article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
we address reversing and anti - reversing properties of interfaces in the following one - dimensional slow diffusion equation with strong absorption @xmath0 where @xmath1 is a positive function , _
e.g. _ , a concentration of some species , and @xmath2 and @xmath3 denote space and time , respectively . restricting the exponents to the ranges @xmath4 and @xmath5 corresponds to the slow diffusion and strong absorption cases respectively
. interfaces sometimes termed ` contact lines ' by fluid dynamicists correspond to the points on the @xmath2-axis , where regions for positive solutions for @xmath1 are connected with the regions where @xmath1 is identically zero . the initial data @xmath6 is assumed to be compactly supported .
the motion of the interfaces is determined from conditions that require the function @xmath1 be continuous and the flux of @xmath1 through the interface to be zero @xcite . in the presence of slow diffusion ( @xmath4 ) , the interfaces of compactly supported solutions have a finite propagation speed @xcite . in the presence of strong absorption ( @xmath5 )
, the solution vanishes for all @xmath2 after some finite time , which is referred to as finite - time extinction @xcite .
therefore , the interfaces for a compactly supported initial data coalesce in a finite time . depending on the shape of @xmath7 and the values of @xmath8 and @xmath9 , the interfaces may change their direction of propagation in a number of different ways .
it was proved by chen _
_ @xcite that bell - shaped initial data remains bell - shaped for all time before the compact support shrinks to a point .
however , the possible types of dynamics of interfaces for this bell - shaped data were not identified in @xcite .
the slow diffusion equation with the strong absorption ( [ heat ] ) describes a variety of different physical processes , including : ( i ) the slow spreading of a slender viscous film over a horizontal plate subject to the action of gravity and a constant evaporation rate @xcite ( when @xmath10 and @xmath11 ) ; ( ii ) the dispersion of a biological population subject to a constant death - rate @xcite ( when @xmath12 and @xmath11 ) ; ( iii ) non - linear heat conduction along a rod with a constant rate of heat loss @xcite ( when @xmath13 and @xmath11 ) , and ; ( iv ) fluid flows in porous media with a drainage rate driven by gravity or background flows @xcite ( when @xmath14 and either @xmath15 or @xmath11 ) .
let us denote the location of the left interface by @xmath16 and the limit @xmath17 , where @xmath1 is nonzero , by @xmath18 .
if @xmath19 , it was proved in @xcite that the position of the interface , @xmath20 , is a lipschitz continuous function of time @xmath3 . in the case
@xmath21 , the function @xmath20 is found from the boundary conditions @xmath22 and @xmath23 where a dot denotes differentiation with respect to time . in the case @xmath24 , the spatial derivatives at @xmath18 are not well defined @xcite . and the zero flux condition ( [ zero - flux ] ) must be rewritten as @xmath25 \displaystyle h^{n } \left ( \frac{\partial h}{\partial x } \right)^{-1 } \big{|}_{x=\ell(t)^+ } , \quad \mbox{\rm if } \;\ ; \dot{\ell } \geq 0 .
\end{array } \right.\end{aligned}\ ] ] one could choose to close the slow diffusion equation ( [ heat ] ) in a variety of ways , _
e.g. _ , by supplying analogous conditions at the right interface , or by supplying a dirichlet or neumann condition elsewhere . for instance
, if @xmath7 is even in @xmath2 , then the solution @xmath1 remains even in @xmath2 for all times , and therefore , the slow diffusion equation ( [ heat ] ) can be closed on the compact interval @xmath26 $ ] by imposing @xmath27 .
however , such details do not concern us here because we are interested in studying the behaviour of solutions to ( [ heat ] ) local to the left interface @xmath28 only .
we reiterate here the main question on the possible types of dynamics in the slow diffusion equation with the strong absorption ( [ heat ] ) . working with bell - shaped ,
compactly supported initial data @xmath7 , one can anticipate _ a priori _ that the compact support of the bell - shaped solution can either : ( i ) decrease monotonically in time , or ; ( ii ) first expand and then subsequently shrink , or ; ( iii ) have more complicated behaviour where multiple instances of expansion and contraction are observed .
this phenomenon brings about both ` reversing ' and ` anti - reversing ' dynamics of an interface . here the term ` reversing ' describes a scenario where the velocity of the left interface @xmath28 satisfies @xmath29 before the reversing time and @xmath30 after the reversing time , whereas the term ` anti - reversing ' refers to the opposite scenario with @xmath30 before and @xmath29 after the reversing time .
the first analytical solution to ( [ heat ] ) exhibiting a reversing interface was obtained by kersner @xcite for the case @xmath21 .
this explicit solution takes the form @xmath31_+,\ ] ] where the plus subscript denotes the positive part of the function , and @xmath32 is an arbitrary parameter .
the interfaces are located symmetrically at @xmath33 with @xmath34 more recently , foster _ et al . _
@xcite considered the case @xmath24 and explored the asymptotic and numerical construction of self - similar solutions for equation ( [ heat ] ) some related , yet different , self - similar solutions to other nonlinear diffusion equations have previously been constructed using a combination of asymptotic analysis and numerical shooting ; see , _
e.g. _ , @xcite .
the self - similar solutions capture the relevant dynamics of reversing interfaces near the corresponding points in the space and time ( which can be placed at the origin of @xmath2 and @xmath3 , without the loss of generality ) . based on a classical point symmetry analysis of the porous medium equation ( [ heat ] ) provided in @xcite the authors of @xcite found that the reversing interfaces can be described via the self - similar reductions @xmath35 where the functions @xmath36 satisfy a pair of second - order differential equations . in this paper
, we explore the case @xmath11 only , when the functions @xmath36 of the self - similar reduction ( [ reduction ] ) satisfy the second - order differential equations @xmath37 we seek positive solutions @xmath36 of the differential equations ( [ ode ] ) on the semi - infinite line @xmath38 that satisfy the following conditions : @xmath39 these first of these conditions , ( [ ( i ) ] ) , are the self - similar counterparts of the condition @xmath40 for the equation ( [ heat ] ) .
in addition , the behaviour of @xmath41 and @xmath42 in the far - field must be matched from the condition @xmath43 the requirement ( [ far - field - matching ] ) is tantamount to enforcing that the solution @xmath1 to the slow diffusion equation ( [ heat ] ) does not ` jump ' as @xmath3 passes through zero this can be verified by taking both the limits @xmath44 and @xmath45 in the self - similar reduction ( [ reduction ] ) .
existence of solutions to the differential equations ( [ ode ] ) on @xmath38 with the required behaviour ( [ far - field - matching ] ) implies , via the self - similar reduction ( [ reduction ] ) , the existence of a reversing ( if @xmath46 ) or anti - reversing ( if @xmath47 ) left interface at @xmath28 , which behaves like @xmath48 after placing the reversing point at the origin of the space time plane . if @xmath49 , the velocity of the interface , @xmath50 , changes sign continuously as @xmath3 passes through zero . by combining formal asymptotic constructions of the solutions @xmath36 near the small and large values with a numerical shooting method , the authors of @xcite claimed that for integer values of @xmath51 , there exists a unique positive value of @xmath52 , which leads to a monotonically growing function @xmath53 on the entire semi - axis @xmath54 . using the far - field matching condition ( [ far - field - matching ] ) , a unique , monotonically growing function @xmath55 is found on @xmath56 for a positive value of @xmath57 .
these solutions correspond to a reversing left interface @xmath28 for @xmath51 .
no solutions exhibiting an anti - reversing interface were found in @xcite . in the present work
, we address the same problem using a dynamical system framework @xcite .
the dynamical system theory allows us to justify the formal asymptotic approximations of @xmath36 for small and large values , as well as to set up an accurate and robust numerical procedure for furnishing appropriate solutions to the differential equations ( [ ode ] ) .
qualitatively , we recover the results of @xcite for @xmath58 , but with a better precision , and we generalize these results for all non - integer values of @xmath59 .
in addition , we demonstrate that the result for @xmath60 , reported in @xcite , is incorrect and no self - similar reversing interface solutions exist for @xmath12 . in addition , we discover new reversing and anti - reversing interface solutions of the same differential equations ( [ ode ] ) for other values of @xmath8 .
our approach explores invariant manifolds for the singular differential equations after appropriate unfolding ( which is sometimes referred to as the blow - up technique @xcite ) .
the main analytical results of this work are given by the following theorems .
+ [ theorem - center - manifold - zero ] for every @xmath49 and @xmath61 , there exists a unique solution of the differential equation ( [ ode ] ) such that @xmath62 as @xmath63 . if @xmath64 , this unique solution has the following asymptotic behaviour @xmath65 whereas if @xmath66 , it has the following asymptotic behaviour @xmath67 [ theorem - center - manifold - infinity ] there exists a one - parameter family of solutions of the differential equation ( [ ode ] ) for the lower sign such that @xmath68 as @xmath69 , and this family has the following asymptotic behaviour @xmath70 where @xmath71 is an arbitrary parameter .
there exists a two - parameter family of solutions of the differential equation ( [ ode ] ) for the upper sign such that @xmath72 as @xmath69 , and this family has the same asymptotic behaviour ( [ center - manifold - infinity - asymptotics ] ) for some @xmath71 .
the main problem is to connect the two asymptotic behaviours of the differential equations ( [ ode ] ) which are defined for small and large values of @xmath36 by theorems [ theorem - center - manifold - zero ] and [ theorem - center - manifold - infinity ] .
we know from @xcite that there exists an exact solution of the connection problem if @xmath73 .
this exact solution is given by @xmath74 however , the exact profile ( [ exact - solution ] ) corresponds to a solution to the slow diffusion equation ( [ heat ] ) with an interface that remains stationary for all time , thus we do not examine it further here .
some connection results for nonzero values of @xmath75 are available for the differential equation ( [ ode ] ) in lemmas [ lemma - connection ] and [ lemma - continuation ] below .
although these results are not sufficient for an analytical solution of the connection problem , we can set up a numerical method , which detects connections of the two solutions described in theorems [ theorem - center - manifold - zero ] and [ theorem - center - manifold - infinity ] .
the remainder of the paper is organized as follows .
the unfolding and invariant manifolds for the differential equations ( [ ode ] ) near small values of @xmath36 are described in 2 .
the corresponding results near large values of @xmath36 are reported in 3 .
the connection problem between the invariant manifolds near small and large values of @xmath36 is considered in 4 .
the relevant numerical technique is implemented in 5 , where the main findings are discussed and compared with the previous results from @xcite .
the paper is concluded in 6 with a discussion of the relevance of the self - similar solutions to the dynamics of ( [ heat ] ) .
we shall rewrite the scalar equations ( [ ode ] ) as vector systems for variables @xmath76 and @xmath77 . in the interests of simplicity of notation
, we drop the plus and minus subscripts in the definitions of the variables @xmath78 and @xmath79 .
the non - autonomous vector system for @xmath78 and @xmath79 is as follows : @xmath80 if @xmath8 is a non - integer , we require the constraint @xmath81 . in either case , only positive solutions for @xmath78 are needed to be considered . the interface , in self - similar variables ,
is assumed to be located at @xmath82 , where @xmath83 a requirement of the condition on the continuity of @xmath84 at @xmath16 . since the value @xmath83 is singular in the non - autonomous system ( [ system ] )
, we shall unfold the singularity by introducing a convenient parametrization of solutions with a new time variable @xmath85 defined by @xmath86 the map @xmath87 is increasing and if @xmath88 as @xmath89 , then @xmath90 for finite values of @xmath85 . with the parametrization @xmath87 , we obtain the autonomous dynamical system in @xmath91 , @xmath92 where the dots stand for the derivatives of @xmath93 in @xmath85 . in what follows ,
we assume that @xmath94 , so that the vector field of the dynamical system ( [ zero - dynamics ] ) is continuously differentiable near zero values of @xmath78 .
again , @xmath81 has to be enforced if @xmath8 is a non - integer .
the family of equilibrium points for the system ( [ zero - dynamics ] ) is given by @xmath95 , where @xmath96 is an arbitrary parameter . if @xmath49 , each equilibrium point is associated with the jacobian matrix @xmath97.\ ] ]
this jacobian matrix has a double zero eigenvalue ( with two linearly independent eigenvectors ) and a simple nonzero eigenvalue @xmath98 .
therefore , the linearization of the dynamical system ( [ zero - dynamics ] ) at the equilibrium point @xmath99 with @xmath100 has a two - dimensional center manifold and a one - dimensional stable manifold , whereas the linearized system with @xmath101 has a two - dimensional center manifold and a one - dimensional unstable manifold .
since the dynamical system ( [ zero - dynamics ] ) is @xmath102 smooth , a straightforward application of the invariant manifold theorems @xcite asserts that the equilibrium point @xmath99 with @xmath103 is located at the intersection of the two invariant manifolds , which are tangential to the invariant manifolds of the linearized system .
we formulate these results in the two following propositions , namely propositions [ proposition - center - manifold - zero ] and [ proposition - stable - manifold - zero ] .
the relevant conclusions on the behaviour of solutions of the differential equations ( [ ode ] ) for small values of @xmath36 , expressed in theorem [ theorem - center - manifold - zero ] , follow from these two propositions .
+ [ proposition - center - manifold - zero ] for every @xmath49 and @xmath103 , there exists a two - dimensional center manifold of the dynamical system ( [ zero - dynamics ] ) near the equilibrium point @xmath99 , which can be parameterized as follows : @xmath104 , \;\ ; u \in ( 0,\delta ) , \;\ ; \xi \in ( a , a+\delta ) , \right\},\ ] ] where @xmath105 is small .
dynamics of the system ( [ zero - dynamics ] ) on the center manifold @xmath106 is topologically equivalent to the dynamics at the truncated normal form @xmath107 in particular , for every @xmath103 , there exists exactly one trajectory on @xmath106 , which approaches the equilibrium point @xmath99 as @xmath89 if @xmath100 and @xmath108 if @xmath101 .
existence of a two - dimensional center manifold @xmath106 , which is tangent to that of the linearized system @xmath109 follows from theorem 4.1 in @xcite .
we develop an approximation of @xmath106 by writing @xmath110 where @xmath111 is @xmath102 at the point @xmath112 with zero partial derivatives .
dynamics along @xmath106 is given by the two - dimensional system @xmath113 the function @xmath114 is to be found by substituting ( [ w - u - f ] ) to the three - dimensional system ( [ zero - dynamics ] ) and using the two - dimensional system ( [ zero - dynamics - center - dynamics ] ) .
then , we obtain a partial differential equation @xmath115 if @xmath49 and @xmath111 is a @xmath102 function at @xmath112 with zero partial derivatives , then equation ( [ zero - dynamics - center - manifold ] ) has a solution such that @xmath116 the representation ( [ w - u - f ] ) and ( [ approximation - center - manifold - zero ] ) is equivalent to ( [ center - manifold - zero ] ) . substituting ( [ approximation - center - manifold - zero ] ) to ( [ zero - dynamics - center - dynamics ] ) and truncating the remainder term , we obtain the truncated normal form ( [ center - manifold - zero - dynamics ] ) . from the second equation of the system ( [ center - manifold - zero - dynamics ] ) , it follows that if @xmath100 , then @xmath117 such that @xmath118 as @xmath89 , whereas if @xmath101 , then @xmath119 such that @xmath118 as @xmath108 . from the first equation of the system ( [ center - manifold - zero - dynamics ] ) , the constant of integration for @xmath120 is arbitrary , so that @xmath121 in the same limit with @xmath122 .
hence , dynamics along the two - dimensional manifold @xmath106 is decomposed between a curve of equilibrium states with @xmath83 and weakly unstable ( if @xmath100 ) or weakly stable ( if @xmath101 ) evolution along a curve parameterized by small positive @xmath78 .
persistence of the dynamics on @xmath106 with respect to the remainder terms in ( [ approximation - center - manifold - zero ] ) follows from analysis of the system ( [ zero - dynamics - center - dynamics ] ) .
[ proposition - stable - manifold - zero ] for every @xmath49 and @xmath101 , there exists a one - dimensional unstable manifold of the dynamical system ( [ zero - dynamics ] ) near the equilibrium point @xmath99 , which can be parameterized as follows : @xmath123 where @xmath105 is small .
dynamics of the system ( [ zero - dynamics ] ) on the unstable manifold @xmath124 is topologically equivalent to dynamics of the linear equation @xmath125 existence of a one - dimensional unstable manifold @xmath126 , which is tangent to that of the linearized system @xmath127 follows from theorem 4.1 in @xcite .
we develop an approximation of @xmath126 by writing @xmath128 where @xmath129 and @xmath130 are @xmath102 with the zero derivative at @xmath83 .
dynamics along @xmath126 is given by the one - dimensional system @xmath131 the functions @xmath132 and @xmath133 are to be found by substituting ( [ xi - w - in - terms - of - u ] ) to the three - dimensional system ( [ zero - dynamics ] ) and using the one - dimensional system ( [ zero - dynamics - stable - dynamics ] ) .
then , we obtain a system of differential equations @xmath134 and @xmath135 if @xmath49 while @xmath129 and @xmath130 are @xmath102 with the zero derivative at @xmath83 , then system ( [ zero - dynamics - stable - manifold-1 ] ) and ( [ zero - dynamics - stable - manifold-2 ] ) has a solution such that @xmath136 the representation ( [ xi - w - in - terms - of - u ] ) and ( [ approximation - stable - manifold - zero ] ) is equivalent to ( [ stable - manifold - zero ] ) . substituting ( [ approximation - stable - manifold - zero ] ) for @xmath137 to ( [ zero - dynamics - stable - dynamics ] ) and truncating the remainder terms at @xmath138 , we obtain the linear equation ( [ stable - manifold - zero - dynamics ] ) .
persistence of the linear dynamics on @xmath126 with respect to the remainder terms in @xmath137 follows from analysis of the differential equation ( [ zero - dynamics - stable - dynamics ] ) . for every @xmath49 and @xmath100
, one can construct a one - dimensional stable manifold @xmath139 of the dynamical system ( [ zero - dynamics ] ) near the equilibrium point @xmath99 , which exists for @xmath90 , @xmath140 , and @xmath141 .
however , this stable manifold does not contain trajectories that approach the equilibrium point @xmath99 as @xmath89 .
_ of theorem [ theorem - center - manifold - zero ] .
_ for every @xmath49 and @xmath142 , proposition [ proposition - center - manifold - zero ] states that the equilibrium state @xmath99 is connected by the trajectories of the dynamical system ( [ zero - dynamics ] ) with @xmath143 as @xmath89 if and only if @xmath100 . in this case
, there exists exactly one trajectory with @xmath140 such that @xmath118 as @xmath89 .
this trajectory belongs to the center manifold @xmath106 , whose dynamics satisfy the system ( [ zero - dynamics - center - dynamics ] ) . from this system ,
we obtain a first - order non - autonomous equation @xmath144 integrating ( [ 1st - non - auto ] ) near @xmath145 , we recover the asymptotic behaviour ( [ center - manifold - zero - asymptotics ] ) . for every @xmath49 and @xmath101 , proposition [ proposition - stable - manifold - zero ]
states that the equilibrium state @xmath99 is connected by exactly one trajectory of the dynamical system ( [ zero - dynamics ] ) with @xmath140 and @xmath118 as @xmath89 .
this trajectory belongs to the unstable manifold @xmath126 with the dynamics satisfying equation ( [ zero - dynamics - stable - dynamics ] ) . from ( [ xi - w - in - terms - of - u ] ) and ( [ approximation - stable - manifold - zero ] ) ,
we obtain @xmath146 \quad \mbox{\rm as } \quad u \to 0.\ ] ] inverting this nonlinear equation near @xmath145 , we recover the asymptotic behaviour ( [ stable - manifold - zero - asymptotics ] ) .
the trajectories departing from the equilibrium point @xmath99 of the dynamical system ( [ zero - dynamics ] ) is expected to arrive at infinite values for @xmath120 and @xmath78 . in order to study the behaviour of trajectories near infinite values for @xmath120 and @xmath78
, we shall define @xmath147 , which maps an infinite value for @xmath78 to a zero value for @xmath148 .
the other variables @xmath120 and @xmath79 must be adjusted accordingly for small values of @xmath148 .
let us consider the following scaling transformation , @xmath149 where @xmath150 is the set of new variables , and the positive parameters @xmath151 and @xmath152 are to be chosen below . substituting the transformation ( [ scaling ] ) to the dynamical system ( [ zero - dynamics ] )
, we obtain the following autonomous system in @xmath91 : @xmath153 where a dot still denotes a derivative with respect to the time variable @xmath85 . the system ( [ scaling - dynamics ] ) is singular at @xmath154 , no matter what positive values of @xmath151 and @xmath152 are chosen . to unfold the singularity
, we can now introduce a convenient parametrization of solutions with the time variable @xmath155 instead of the variable @xmath85 such that @xmath156 the map @xmath157 is increasing and we can consider solutions parameterized by the new time variable @xmath155 such that @xmath158 as @xmath159 ( which could correspond to a finite value for the old time variable @xmath85 ) . with the parametrization @xmath157 , the system ( [ scaling - dynamics ] )
can be rewritten in the equivalent form @xmath160 where a prime now denotes a derivative with respect to the new time variable @xmath155
. a suitable choice of parameters @xmath151 and @xmath152 is given by @xmath161 so that the explicit form of the transformations ( [ scaling ] ) is given by @xmath162 the choice ( [ choice ] ) ensures that the system ( [ scaling - dynamics - equivalent ] ) is rewritten in the simplest non - singular form with a quadratic vector field : @xmath163 the family of equilibrium points for the system ( [ infinity - dynamics ] ) is given by @xmath164 , where @xmath165 is an arbitrary parameter . each equilibrium point is associated with the jacobian matrix @xmath166.\ ] ] this jacobian matrix has a double zero eigenvalue ( with two linearly independent eigenvectors ) and a simple eigenvalue @xmath167 . in the context of reversing and anti - reversing interfaces , we are interested in the behaviour of solutions for which @xmath69 and @xmath168 .
it follows from the transformation ( [ scaling - explicit ] ) , that this requirement restricts our consideration to the family of critical points @xmath169 with @xmath71 . if @xmath71 , the linearization of the dynamical system ( [ infinity - dynamics ] ) at the equilibrium point @xmath169 has a two - dimensional center manifold and a one - dimensional stable ( upper sign ) or unstable ( lower sign ) manifold . since the vector field of the dynamical system ( [ infinity - dynamics ] )
is analytic ( quadratic ) , another straightforward application of the invariant manifold theorems @xcite yields that the equilibrium point @xmath169 with @xmath71 is located at the intersection of the two invariant manifolds , which are tangential to the invariant manifolds of the linearized system .
we formulate these results in the next two propositions , namely propositions [ proposition - center - manifold - infinity ] and [ proposition - stable - manifold - infinity ] .
the relevant conclusions on the behaviour of solutions of the differential equations ( [ ode ] ) for large values of @xmath36 expressed in theorem [ theorem - center - manifold - infinity ] follow as a corollary from these two propositions .
+ [ proposition - center - manifold - infinity ] for every @xmath71 , there exists a two - dimensional center manifold of the dynamical system ( [ infinity - dynamics ] ) near the equilibrium point @xmath169 , which can be parameterized as follows : @xmath170 , \quad \begin{array}{l } x \in ( x_0-\delta , x_0+\delta ) , \\ z \in ( -\delta,\delta ) , \end{array } \right\},\ ] ] where @xmath105 is small and @xmath171 denotes cubic terms in @xmath172 and @xmath173 .
the dynamics of the system ( [ infinity - dynamics ] ) on the center manifold @xmath174 is topologically equivalent to the dynamics at the truncated normal form @xmath175 in particular , there exists exactly one trajectory on @xmath174 , which approaches the equilibrium point @xmath169 as @xmath159 .
existence of a two - dimensional center manifold @xmath174 , which is tangent to that of the linearized system @xmath176 follows from theorem 4.1 in @xcite .
we develop an approximation of @xmath174 by writing @xmath177 and expanding @xmath178 using a taylor series in small values of both @xmath179 and @xmath173 .
dynamics along @xmath174 is given by the two - dimensional system @xmath180 the function @xmath178 is to be found from the partial differential equation @xmath181 + \frac{\partial f}{\partial z } \left [ f ( \pm 1 + f ) \mp \frac{m+1}{2 } xz - \frac{m+3}{2 } z^2 \right ] + z f = 0.\ ] ] equations ( [ infinity - dynamics - center - dynamics ] ) and ( [ infinity - dynamics - center - manifold ] ) suggest the following near - identity transformation of the function @xmath178 given by @xmath182 after the near - identity transformation ( [ near - identity ] ) , the dynamics along @xmath174 are given by the two - dimensional system @xmath183.\end{array } \right.\ ] ] the function @xmath184 is now to be found from the partial differential equation @xmath185 which has a solution such that @xmath186 expansions ( [ near - identity ] ) and ( [ f-1 ] ) yield ( [ center - manifold - infinity ] ) . substituting ( [ f-1 ] ) into ( [ infinity - dynamics - center - dynamics - eq ] ) and truncating at the cubic terms
, we obtain the truncated normal form ( [ center - manifold - infinity - dynamics ] ) . from the second equation of the system ( [ center - manifold - infinity - dynamics ] ) , it follows that there exists a unique solution such that @xmath187 as @xmath159 with @xmath188 .
from the first equation of the system ( [ center - manifold - infinity - dynamics ] ) , the constant of integration for @xmath2 is arbitrary , so that @xmath189 as @xmath159 with @xmath190 .
hence , dynamics along the two - dimensional manifold @xmath174 is decomposed between a curve of equilibrium states with @xmath191 and weakly stable evolution along a curve parameterized by small positive @xmath173 .
persistence of the dynamics on @xmath174 with respect to the remainder terms in ( [ f-1 ] ) follows from analysis of the system ( [ infinity - dynamics - center - dynamics - eq ] ) .
[ proposition - stable - manifold - infinity ] for every @xmath71 , there exists a one - dimensional stable ( upper sign ) or unstable ( lower sign ) manifold of the dynamical system ( [ infinity - dynamics ] ) near the equilibrium point @xmath169 , which can be expressed explicitly : @xmath192 , \quad x \in ( x_0-\delta , x_0+\delta ) , \right\},\ ] ] where @xmath105 is small .
the dynamics of the system ( [ infinity - dynamics ] ) on the manifold @xmath193 is topologically equivalent to the dynamics of the linear equation @xmath194 the linearized system has the stable / unstable manifold : @xmath195 existence of a one - dimensional manifold @xmath193 that is tangent to @xmath196 follows from theorem 4.1 in @xcite .
we notice from ( [ infinity - dynamics ] ) that @xmath154 is an invariant reduction of the three - dimensional system .
therefore , we can set @xmath154 and seek a parametrization of @xmath193 by working with @xmath197 , where @xmath198 and @xmath199 .
dynamics along @xmath193 are given by the one - dimensional system @xmath200 from the three - dimensional system ( [ infinity - dynamics ] ) , we obtain a linear differential equation for nonzero @xmath201 : @xmath202 this equation completed with the initial condition @xmath198 admits a unique solution @xmath203.\ ] ] note that @xmath199 , because @xmath193 is tangent to @xmath196 . substituting @xmath204 into the differential equation ( [ infinity - dynamics - stable - dynamics ] ) and truncating at the quadratic remainder term , we obtain the linear equation ( [ stable - manifold - infinity - dynamics ] ) .
persistence of the linear dynamics on @xmath193 with respect to the remainder term follows from analysis of the system ( [ infinity - dynamics - stable - dynamics ] ) with the expansion ( [ expansion - h ] ) .
_ of theorem [ theorem - center - manifold - infinity ] .
_ for every @xmath71 , proposition [ proposition - center - manifold - infinity ] states that there exists exactly one trajectory with @xmath205 such that @xmath206 as @xmath159 .
this trajectory belongs to the center manifold @xmath174 , whose dynamics satisfy the system ( [ infinity - dynamics - center - dynamics ] ) .
we recover the asymptotic behaviour ( [ center - manifold - infinity - asymptotics ] ) by eliminating @xmath148 from the transformation ( [ scaling - explicit ] ) and setting @xmath207 .
for every @xmath71 , proposition [ proposition - stable - manifold - infinity ] states that the point @xmath169 is an intersection of the two - dimensional center manifold @xmath174 and the one - dimensional stable / unstable manifold @xmath193 for the upper / lower sign .
therefore , the point can be reached in the direction @xmath159 along @xmath208 but can not be reached along @xmath209 .
this guarantees that the trajectory in proposition [ proposition - center - manifold - infinity ] with @xmath205 such that @xmath206 as @xmath159 is unique for the lower sign .
therefore , for every @xmath71 , there exists a one - dimensional set of solutions of the differential equation ( [ ode ] ) for the lower sign such that @xmath68 as @xmath69 . on the other hand ,
the span of the trajectory in proposition [ proposition - center - manifold - infinity ] and the trajectory in proposition [ proposition - stable - manifold - infinity ] is a two - dimensional set that hosts all trajectories with @xmath205 such that @xmath206 as @xmath159 .
therefore , for every @xmath71 , there exists a two - dimensional set of solutions of the differential equation ( [ ode ] ) for the upper sign such that @xmath72 as @xmath69 .
the rate of change along @xmath208 is exponential in @xmath155 and the rate of change along @xmath174 is algebraic in @xmath155 .
therefore , solutions along the two - dimensional set still obey the asymptotic behaviour ( [ center - manifold - infinity - asymptotics ] ) .
let us summarize the results of the previous two sections on the existence of solutions @xmath36 to the differential equations ( [ ode ] ) on the semi - infinite line @xmath210 that satisfy the properties ( [ ( i)])-([(iii ) ] ) .
such solutions are related to the trajectories of the dynamical systems ( [ zero - dynamics ] ) and ( [ infinity - dynamics ] ) , which depart from the equilibrium points where @xmath36 is zero and arrive to the equilibrium point where @xmath36 is infinite . in what follows , we will consider separately the two different systems for @xmath55 and @xmath53 . by propositions
[ proposition - center - manifold - zero ] and [ proposition - stable - manifold - zero ] , for every nonzero @xmath212 , there is a unique trajectory of the dynamical system ( [ zero - dynamics ] ) in variables @xmath93 that departs from the equilibrium point @xmath213 as @xmath89 and belongs to the domain @xmath140 .
this trajectory is contained in the center manifold @xmath214 if @xmath215 and in the unstable manifold @xmath216 if @xmath217 . by propositions [ proposition - center - manifold - infinity ] and [ proposition - stable - manifold - infinity ] , for every @xmath71
, there is a two - dimensional set of trajectories of the dynamical system ( [ infinity - dynamics ] ) in variables @xmath218 that reaches the equilibrium point @xmath169 as @xmath159 and belongs to the domain @xmath205 .
this trajectory is contained in the intersection between the center @xmath174 and stable @xmath208 manifolds .
we shall now establish that the same trajectory departing from the equilibrium point @xmath213 in ( [ zero - dynamics ] ) arrives to the equilibrium point @xmath169 in ( [ infinity - dynamics ] ) .
this trajectory determines a unique solution @xmath55 of the differential equation ( [ ode ] ) with the upper sign satisfying properties ( [ ( i)])([(iii ) ] ) .
+ [ lemma - connection ] fix @xmath219 and consider a one - parameter trajectory of the dynamical system ( [ zero - dynamics ] ) for the upper sign such that @xmath220 as @xmath89 and @xmath140 .
then , there exists a @xmath221 ( or @xmath222 ) such that @xmath223 and @xmath224 as @xmath225 .
we introduce the energy - like quantity for the dynamical system ( [ zero - dynamics ] ) with the upper sign : @xmath226 computing the derivative of @xmath227 in @xmath85 along a solution of system ( [ zero - dynamics ] ) , we obtain @xmath228 if @xmath215 , then @xmath229 , and @xmath227 is a strictly decreasing function of @xmath85 as long as the solution to system ( [ zero - dynamics ] ) exists and @xmath230 . by the representation ( [ center - manifold - zero ] ) of @xmath214 in proposition [ proposition - center - manifold - zero ]
, we have @xmath231 for sufficiently large negative @xmath85 .
now @xmath232 can not vanish for any @xmath85 , because if @xmath233 , then @xmath234 , which contradicts positivity of @xmath79 before vanishing .
therefore , @xmath235 decreases to @xmath236 in finite or infinite time @xmath85 . because @xmath237 , we have @xmath168 if @xmath238 .
if @xmath217 , then @xmath227 is an increasing function of @xmath85 at least for sufficiently large negative @xmath85 . by the representation ( [ stable - manifold - zero ] ) of @xmath216 in proposition
[ proposition - stable - manifold - zero ] , we still have @xmath231 for sufficiently large negative @xmath85 . also , @xmath232 can not vanish for any @xmath85 by the same contradiction , since if @xmath233 , then @xmath234 .
therefore , @xmath239 and @xmath240 are still increasing functions , and there is a finite @xmath221 such that @xmath241 and @xmath242 . for @xmath243 ,
the energy method described above proves again that @xmath224 in finite or infinite time @xmath85 .
we shall now prove that @xmath223 as @xmath225 .
since @xmath244 for all @xmath245 , the map @xmath246 is monotonically increasing , so that we can parameterize both @xmath120 and @xmath79 by @xmath78 and consider the limit @xmath168 . from the last two equations of the system ( [ zero - dynamics ] ) , we obtain @xmath247 since @xmath248 for @xmath85 close to @xmath249 and @xmath231 for all @xmath245 , we estimate @xmath250 integrating this differential inequality , we obtain @xmath251 where @xmath32 is a constant of integration . therefore , as @xmath252 , there exists a constant @xmath253 such that @xmath254 for sufficiently large @xmath78 . from the first two equations of the system ( [ zero - dynamics ] ) , we obtain @xmath255 integrating this differential inequality , we obtain a lower bound for @xmath120 given by @xmath256 where @xmath32 is another constant of integration .
this lower bound yields @xmath69 as @xmath168 .
[ coral-1 ] the trajectory of lemma [ lemma - connection ] such that @xmath223 and @xmath224 corresponds to the trajectory of system ( [ infinity - dynamics ] ) approaching the equilibrium state @xmath169 for some @xmath257 .
consequently , one can define a piecewise @xmath102 map @xmath258 we can use the transformation ( [ scaling - explicit ] ) .
since @xmath252 , then @xmath158 .
from the bound ( [ bound - w ] ) , we obtain @xmath259 therefore , @xmath260 as @xmath252 . on the other hand , from the bound ( [ bound - xi ] ) , we only obtain @xmath261 which is not sufficient to claim that @xmath2 remains bounded as @xmath252 . using the transformation ( [ scaling - explicit ] ) and equations ( [ equation - w ] ) and ( [ equation - xi ] )
, we obtain the following system for @xmath2 and @xmath173 in the variable @xmath78 : @xmath262\ ] ] and @xmath263.\ ] ] it follows from the bound ( [ bound - z ] ) and equation ( [ equation - z ] ) that there is a positive constant @xmath264 such that @xmath265 for sufficiently large @xmath78 .
since @xmath266 is integrable as @xmath168 , then @xmath267 is bounded from below by a negative constant . since @xmath173 is bounded and approaches to zero as @xmath252 , we finally obtain that @xmath2 is bounded from above by a positive constant for all sufficiently large @xmath78 .
finally , @xmath169 is an equilibrium state of system ( [ infinity - dynamics ] ) , therefore , the trajectory approaches @xmath169 for some @xmath257 . integrating equation ( [ equation - x ] )
, we obtain @xmath268 where @xmath269 is an arbitrary constant .
if we can prove that @xmath270 , then @xmath271 this would indicate that the bound ( [ bound - z ] ) is not sharp .
however , we only have numerical data supporting this claim for every @xmath272
. we will show numerically in 5 that both pieces of the map ( [ two - maps ] ) are monotonically increasing for all @xmath272 and intersecting at @xmath273 for @xmath274 .
therefore , the piecewise @xmath102 map @xmath275 is in fact continuous at @xmath274 .
the exact value @xmath276 corresponds to the exact solution ( [ exact - solution ] ) of the scalar differential equations ( [ ode ] ) .
this exact solution is recovered with the following elementary result .
[ lemma - exact ] there exists an exact solution of the dynamical systems ( [ zero - dynamics ] ) and ( [ infinity - dynamics ] ) given by @xmath277 and @xmath278 respectively , where @xmath279 , whereas @xmath280 and @xmath281 are arbitrary positive parameters .
consider the system ( [ infinity - dynamics ] ) and try the reduction @xmath207 , where @xmath282 is constant in @xmath155 .
then , on setting @xmath283 we obtain the following differential equations : @xmath284 this system is compatible with the constraint @xmath283 if @xmath285 and @xmath286 , so that @xmath287 .
the general solution of @xmath285 is @xmath288 for a positive parameter @xmath281 .
the solution is defined for @xmath289 .
consider the system ( [ zero - dynamics ] ) and try the reduction @xmath290 .
then , the system is compatible with the reduction if @xmath291 and @xmath292 .
therefore , @xmath293 which admits a general solution given by @xmath294 where @xmath280 is an arbitrary positive constant .
the solution exists for @xmath295 .
other components @xmath78 and @xmath79 are found from the above relations .
note that in the exact solution ( [ exact - solution - zero ] ) , @xmath296 as @xmath89 .
therefore , the exact solution of proposition [ lemma - exact ] corresponds to the choice @xmath274 in lemma [ lemma - connection ] . by propositions [ proposition - center - manifold - zero ] and [ proposition - stable - manifold - zero ] , for every nonzero @xmath298
, there is a unique trajectory of the dynamical system ( [ zero - dynamics ] ) in variables @xmath93 that departs from the equilibrium point @xmath299 as @xmath89 and belongs to the domain @xmath140 .
this trajectory is contained in the unstable manifold @xmath300 if @xmath301 and in the center manifold @xmath302 if @xmath303 . by propositions [ proposition - center - manifold - infinity ] and [ proposition - stable - manifold - infinity ] , for every @xmath71
, there is a one - dimensional set of trajectories of the dynamical system ( [ infinity - dynamics ] ) in variables @xmath218 that reaches the equilibrium point @xmath169 as @xmath159 and belongs to the domain @xmath205 .
this trajectory is contained in the center manifold @xmath174 .
if we try an argument used in the proof of lemma [ lemma - connection ] , then it becomes clear that most of the trajectories departing from the equilibrium point @xmath299 in system ( [ zero - dynamics ] ) will not arrive to the equilibrium point @xmath169 in system ( [ infinity - dynamics ] ) but instead reach the value @xmath83 in a finite @xmath304 .
this indicates that , first , there are very few values of @xmath52 , for which the trajectories may reach infinite values for @xmath78 , and second , the numerical method should not be based on the trajectories departing from the equilibrium point @xmath299 ( such a shooting method was used previously in @xcite ) . instead , it may be better to look for the one - dimensional trajectory departing the equilibrium point @xmath169 in system ( [ infinity - dynamics ] ) in the negative direction of the time variable @xmath155 . to illustrate the previous point
, we show on figure [ return - to - zero ] the trajectories of the system ( [ zero - dynamics ] ) with @xmath10 starting from the equilibrium point @xmath305 along either center ( for @xmath215 ) or unstable ( for @xmath301 ) manifolds .
the trajectories of the system for @xmath55 extend from small to infinite values of @xmath55 , see panel ( a ) .
contrastingly , the trajectories of the system for @xmath53 turn back and return to small values of @xmath53 , see panel ( b ) .
note that the return time is significantly different between the first two and the last two trajectories .
this indicates the presence of a particular value of @xmath52 , for which there exists a trajectory that extends from small to infinite values of @xmath53 , see ( [ m - is-3-vals ] ) below . )
with @xmath10 for @xmath55 and @xmath53 respectively . in both cases ,
the trajectories start from the equilibrium point @xmath305 and depart along either the center or the unstable manifolds.,title="fig:",scaledwidth=49.0% ] ) with @xmath10 for @xmath55 and @xmath53 respectively .
in both cases , the trajectories start from the equilibrium point @xmath305 and depart along either the center or the unstable manifolds.,title="fig:",scaledwidth=49.0% ] in order to justify our numerical scheme , we shall prove that the trajectory departing the equilibrium point @xmath169 in system ( [ infinity - dynamics ] ) in the negative @xmath155 direction either intersects the plane @xmath83 or the plane @xmath233 of system ( [ zero - dynamics ] ) .
+ [ lemma - continuation ] fix @xmath71 and consider a one - parameter trajectory of the dynamical system ( [ infinity - dynamics ] ) for the lower sign such that @xmath306 as @xmath159 and @xmath205
. then , there exists an @xmath307 ( or @xmath308 ) such that * either @xmath309 and @xmath310 $ ] , * or @xmath311 as @xmath312 , whereas @xmath313 . for convenience ,
let us reverse the time variable , by transforming @xmath314 , and rewrite system ( [ infinity - dynamics ] ) with the lower sign in the negative direction of @xmath155 : @xmath315 by proposition [ proposition - center - manifold - infinity ] , we have @xmath205 and @xmath188 for the trajectory departing from the equilibrium point @xmath169 ( in negative @xmath155 ) along @xmath174 . from the second equation of system ( [ infinity - dynamics - negative - time ] )
, @xmath148 remains an increasing function of negative @xmath155 as long as @xmath173 remains positive .
therefore , we have an alternative : either @xmath173 vanishes before @xmath148 diverges or @xmath148 diverges before @xmath173 vanishes .
the first choice of the alternative gives case ( i ) .
for the second choice , let us consider variables @xmath120 and @xmath79 given by ( [ scaling - explicit ] ) and parameterized by @xmath148 in the limit @xmath316 ( the map @xmath317 is one - to - one and onto ) . by dividing the first and third equations in system ( [ infinity - dynamics - negative - time ] ) by the second equation , we obtain @xmath318 by using variables @xmath120 and @xmath79 given by ( [ scaling - explicit ] ) , we obtain @xmath319 to show that the second choice of the alternative above gives case ( ii ) , we will prove that @xmath79 remains finite as @xmath316 .
this is done by a contradiction .
assume that @xmath320 as @xmath316 .
therefore , there exists @xmath321 such that @xmath322 for all sufficiently large @xmath148 .
then , from the first equation of system ( [ closed - infinity ] ) , we have for sufficiently large @xmath148 , @xmath323 because @xmath324 is integrable at infinity , there is a finite positive @xmath325 such that @xmath326 for all sufficiently large @xmath148 .
then , from the second equation of system ( [ closed - infinity ] ) , we obtain for sufficiently large @xmath148 , @xmath327 since both @xmath324 and @xmath328 are integrable at infinity , there is a finite positive @xmath329 such that @xmath330 for all sufficiently large @xmath148 , contradicting the assumption that @xmath320 as @xmath316 .
therefore , the case ( ii ) is proved .
the trajectory of system ( [ infinity - dynamics ] ) departing from the equilibrium point @xmath169 with @xmath71 in the negative direction of @xmath155 intersects either the half - plane @xmath233 and @xmath81 in system ( [ zero - dynamics ] ) in case ( i ) of lemma [ lemma - continuation ] or the half - plane @xmath83 , @xmath331 in system ( [ zero - dynamics ] ) in case ( ii ) .
moreover , in the corresponding cases , * if @xmath332 , then @xmath333 * if @xmath334 , then @xmath333 .
consequently , one can define two piecewise @xmath102 maps @xmath335 [ corollary - continuation ] in case ( i ) , it is trivial to see that @xmath309 and @xmath336 $ ] corresponds to the half - plane @xmath233 and @xmath81 .
the first equation of system ( [ infinity - dynamics - negative - time ] ) can be written for the variable @xmath120 as follows : @xmath337 where the prime still denotes the derivative with respect to the time variable @xmath155 in the negative direction of @xmath155 .
if @xmath148 remains finite as @xmath312 ( so that @xmath338 ) , then @xmath339 is bounded . in case ( ii ) , it is also trivial to see that @xmath340 and @xmath341 corresponds to the half - plane @xmath83 and @xmath331 .
if @xmath79 remains nonzero in the limit @xmath316 , then , there exists @xmath321 such that @xmath322 for all sufficiently large @xmath148 .
then , the same analysis as in lemma [ lemma - continuation ] applies to the first equation of system ( [ closed - infinity ] ) and shows that @xmath120 remains finite as @xmath316 . _
_ unfortunately , we do not control the value of @xmath120 at the intersection of the two piecewise @xmath102 maps ( [ maps - negative ] ) .
however , we will show numerically in
[ num - scheme ] that the piecewise @xmath102 maps ( [ maps - negative ] ) are typically connected at the points where @xmath342 and @xmath82 . in this case , a true solution @xmath53 of the differential equation ( [ ode ] ) with the lower sign satisfying properties ( [ ( i)])([(iii ) ] ) exist .
let us describe a new numerical approach , based on the results of lemmas [ lemma - connection ] and [ lemma - continuation ] , that will be used to furnish meaningful solutions to the differential equations ( [ ode ] ) , for @xmath53 and @xmath55 . in [ t - neg - shooting ] and [ t - pos - solutions ] we describe the numerical procedures for finding solutions for @xmath53 and @xmath55 respectively . finally , in
[ num - sum ] , we summarize the results of the numerical experiments and compare them with the results found in foster _ et .
_ @xcite . as discussed in [ h - minus - system ]
, we wish to numerically construct a unique trajectory from infinite to finite values of @xmath53 .
to do so , we integrate the system ( [ infinity - dynamics ] ) from near the equilibrium point @xmath169 in the far - field towards the equilibrium point @xmath299 of the system ( [ zero - dynamics ] ) in the near - field .
the numerical procedure is carried out as follows : select a value of @xmath344 .
ideally , one would like to begin by using this choice of @xmath282 to specify unique initial values for @xmath345 , and then numerically integrating the system ( [ infinity - dynamics ] ) backward in the ` time ' variable @xmath155 .
equivalently , one could integrate the system ( [ infinity - dynamics - negative - time ] ) forwards in time .
however , since @xmath169 is an equilibrium point , it is not possible to escape @xmath169 in a finite time .
thus , in order to ensure that any numerical integration scheme can depart from near the equilibrium point along the center manifold , @xmath174 , it is necessary to take a ` small step ' , say @xmath346 , away from @xmath169 using the relevant asymptotic behaviour . using ( [ center - manifold - infinity ] ) and ( [ center - manifold - infinity - dynamics ] ) we find that a trajectory on the center manifold @xmath174 has the local behaviour @xmath347 for small positive values of @xmath348 . having selected values for both @xmath282 and @xmath348 , the behaviours ( [ numeric - begin - t - negative ] )
may be used to specify unique ( pseudo-)initial values for @xmath218 and to begin the numerical integration of the system ( [ infinity - dynamics ] ) in the direction of decreasing @xmath155 . in this study , numerical integration of the system ( [ infinity - dynamics ] )
was carried out using the ode45 routine in matlab with the default settings , except abstol and reltol which were both set to have a value of @xmath349 . selecting an appropriate value of @xmath348
is a somewhat ad - hoc procedure : there is trade - off between taking @xmath348 too small , which renders it difficult for the numerical integration to escape the neighbourhood of the equilibrium point ( leading to poor accuracy of the integration ) , and taking @xmath348 too large which could result in low accuracy of the asymptotic expansion ( [ numeric - begin - t - negative ] ) .
however , we found that choosing @xmath350 gave good results over the ranges of parameters we studied .
robustness of the results with respect to changes in : ( i ) the choice of @xmath348 , and ; ( ii ) the number of terms in the asymptotic expansion ( [ numeric - begin - t - negative ] ) were verified .
the result of corollary [ corollary - continuation ] asserts that all such trajectories will ultimately in either a finite or infinite time intersect with either the plane @xmath351 or @xmath352 , see the maps ( [ maps - negative ] ) . to ensure accurate numerical integration of the system in the near - field , in variables
@xmath93 , it is necessary to ` switch ' from integrating the far - field system ( [ infinity - dynamics ] ) to the near - field system ( [ zero - dynamics ] ) backward in time .
the choice of conditions under which this switch should occur is , again , somewhat arbitrary .
however , as long as the values of @xmath218 and hence the values of @xmath93 are all finite and non - zero , this switching is valid at any point . in this study
, we chose to switch from integrating ( [ infinity - dynamics ] ) to ( [ zero - dynamics ] ) when @xmath353 ( or equivalently when @xmath354 ) .
however , we verified that our results were robust to changes in the choice of switching conditions .
this switching procedure can be readily automated within matlab using the events function to autonomously : ( i ) stop the integration of the system ( [ infinity - dynamics ] ) when specified the conditions are satisfied ; ( ii ) read - off the final values of @xmath218 ; ( iii ) transform these to corresponding values for @xmath93 using the change of variables ( [ scaling - explicit ] ) , and ; ( iv ) begin the integration of ( [ zero - dynamics ] ) backwards in time from the appropriate initial data .
the integration of the system ( [ zero - dynamics ] ) is then continued backward in the time variable @xmath85 until either @xmath351 or @xmath352 .
again , we used the events function to autonomously detect when either of these events occurred and to stop the integration .
it is noteworthy that we found it helpful to use ode15s to integrate the near - field system again , the default settings were used with the exception of abstol and reltol which we both set to @xmath349 .
although ode15s is typically slower than ode45 , it is considerably more appropriate to deal with integrating systems of equations that exhibit apparent stiffness .
this apparent stiffness , manifested as rapid changes in the direction of the trajectory in variables @xmath93 , arises as an artifact of the infinite time required to reach the equilibrium point @xmath299 .
thus , if a trajectory approaches very close to the equilibrium point it appears to be rapidly rejected from that neighbourhood .
when the integration is terminated , we record the following pieces of data : ( i ) the selected value of @xmath282 ; ( ii ) whether the trajectory reached @xmath352 or @xmath351 , and ; ( iii ) the value of either @xmath355 or @xmath356 at the termination point .
these data define a point on one of the two piecewise @xmath102 maps defined in ( [ maps - negative ] ) .
it is by computing a large number of trajectories , each emanating from different values of @xmath282 , that we are able to trace out the forms of these maps .
the non - trivial solutions for @xmath53 that we seek correspond to trajectories emanating from particular equilibrium points in the far - field , say @xmath357 , that reach the near - field equilibrium point , @xmath299 for some finite @xmath358 .
in addition to these non - trivial solutions , we recall that for all values of @xmath59 there exists a trivial solution given by ( [ exact - solution - zero ] ) and ( [ exact - solution - infinity ] ) that emanates from @xmath359 and reaches @xmath360 , as discussed in proposition [ lemma - exact ] .
some representative results for @xmath361 and @xmath362 are shown in figure [ individual - maps ] . in these plots , a suitable non - trivial solution for @xmath53 is found by identifying a value of @xmath363 for which the value of @xmath364 or @xmath365 at the termination point . in figure
[ example - shooting ] , we show some representative computations highlighting the differences between the near - field behaviour of trajectories local to a true solution with @xmath366 and @xmath367 . as is evidenced by figure [ individual - maps ] ,
close to a trajectory with @xmath367 the piecewise continuous @xmath102 maps defined in ( [ maps - negative ] ) are smooth , whereas close to a trajectory with @xmath366 the maps exhibit rapid changes and ( vertical ) cusp - like features . as a result ,
determining negative value(s ) of @xmath52 is more challenging despite resolving @xmath282 to machine precision ( approximately @xmath368 in the standard ieee double precision ) , it is not possible to approximate the value of @xmath369 sufficiently well that an accurate estimate of @xmath370 can be determined .
in these cases , we therefore found it necessary to implement one additional stage in the numerical scheme as follows .
once @xmath369 had been determined up to machine precision , and two ` limiting ' trajectories had been identified ( one emanating from @xmath371 and terminating at @xmath352 , and the other emanating from @xmath372 and terminating at @xmath351 , where @xmath373 ) the expected near - field linear asymptotic behaviour of @xmath374 , according to the expansion ( [ center - manifold - zero - asymptotics ] ) for the solution @xmath53 in theorem [ theorem - center - manifold - zero ] , is clearly visible .
this linear behaviour can then be extrapolated , in the direction of decreasing @xmath120 , until it intersects the @xmath120-axis .
this intersection point is , to a good approximation , the value of @xmath370 corresponding to the trajectory emanating from @xmath369 .
panel ( a ) of figure [ example - shooting ] gives an example of the linear extrapolation procedure described above .
using the procedure described above , for all values of @xmath59 , we recover the trivial solution discussed in proposition [ lemma - exact ] .
in addition , we see that in the case @xmath375 there is only one suitable non - trivial solution with the following data : @xmath376 for @xmath377 and @xmath378 , similar results are observed with one trivial and only one non - trivial solution as follows : @xmath379 and @xmath380 contrastingly , in the case @xmath60 we find that three suitable non - trivial solutions exist with the data : @xmath381 x_0^ * \approx 0.137 , \quad a_- \approx -0.932 , \\*[1 mm ] x_0^ * \approx 0.0592 , \quad a_- \approx -0.546 . \end{array } \right.\ ] ] notably all values of @xmath52 for @xmath60 are negative , whereas for @xmath382 and @xmath362 they are positive .
in addition to the detailed results for @xmath383 and @xmath362 shown in figure [ individual - maps ] , we also show in figure [ all - sols - negative - t ] the values of @xmath369 determined for all values of @xmath8 from @xmath14 to @xmath384 . for the primary red and blue branches , emanating from @xmath375 along the black branch ,
we show the corresponding values of @xmath52 in figure [ negative - t - a - minus ] .
intriguingly , the numerical results indicate that in addition to the exact solution which is valid for all @xmath59 there are a whole host of additional solutions , some with @xmath367 , and others with @xmath366 .
in particular , there is at least one additional trajectory corresponding to a suitable solution for @xmath53 with @xmath367 for all values of @xmath385 .
further , for all @xmath386 there exists at least one additional solution for @xmath53 , although , in this case for a value of @xmath366 .
another noteworthy feature of the plots shown in panels ( a)-(d ) of figure [ all - sols - negative - t ] is that at each value of @xmath387 for @xmath388 additional branches of solutions depart from the branch along which @xmath389 and @xmath390 .
the underlying reason for this structure is as yet not understood .
maps defined in ( [ maps - negative ] ) for @xmath361 and @xmath362 respectively . in all cases the blue , red and black curves show the value of @xmath79 at @xmath352 , the value of @xmath78 at @xmath351 and the value of @xmath120 at the termination point respectively .
the dashed vertical line indicates the value of @xmath391 corresponding to the exact solution ( [ exact - solution ] ) .
the crosses on panel ( a ) mark the data points extracted using the extrapolation procedure discussed in the text.,title="fig:",scaledwidth=49.0% ] maps defined in ( [ maps - negative ] ) for @xmath361 and @xmath362 respectively . in all cases the blue , red and black curves show the value of @xmath79 at @xmath352 , the value of @xmath78 at @xmath351 and the value of @xmath120 at the termination point respectively .
the dashed vertical line indicates the value of @xmath391 corresponding to the exact solution ( [ exact - solution ] ) .
the crosses on panel ( a ) mark the data points extracted using the extrapolation procedure discussed in the text.,title="fig:",scaledwidth=49.0% ] maps defined in ( [ maps - negative ] ) for @xmath361 and @xmath362 respectively . in all cases the blue , red and black curves show the value of @xmath79 at @xmath352 , the value of @xmath78 at @xmath351 and the value of @xmath120 at the termination point respectively .
the dashed vertical line indicates the value of @xmath391 corresponding to the exact solution ( [ exact - solution ] ) .
the crosses on panel ( a ) mark the data points extracted using the extrapolation procedure discussed in the text.,title="fig:",scaledwidth=49.0% ] maps defined in ( [ maps - negative ] ) for @xmath361 and @xmath362 respectively . in all cases the blue , red and black curves show the value of @xmath79 at @xmath352 , the value of @xmath78 at @xmath351 and the value of @xmath120 at the termination point respectively .
the dashed vertical line indicates the value of @xmath391 corresponding to the exact solution ( [ exact - solution ] ) .
the crosses on panel ( a ) mark the data points extracted using the extrapolation procedure discussed in the text.,title="fig:",scaledwidth=49.0% ] for @xmath60 .
more precisely the red and blue trajectories begin at @xmath371 for @xmath392 and @xmath368 .
the black line shows the artificially extrapolated linear behaviour .
panel ( b ) shows some representative trajectories emanating from @xmath393 for @xmath377 . in this case , the trajectories begin at @xmath371 for @xmath394 and @xmath395 .
notably , in the latter case , despite only resolving @xmath369 to 3 significant digits , a good estimate of @xmath52 has already been obtained.,title="fig:",scaledwidth=49.0% ] for @xmath60 .
more precisely the red and blue trajectories begin at @xmath371 for @xmath392 and @xmath368 .
the black line shows the artificially extrapolated linear behaviour .
panel ( b ) shows some representative trajectories emanating from @xmath393 for @xmath377 . in this case , the trajectories begin at @xmath371 for @xmath394 and @xmath395 .
notably , in the latter case , despite only resolving @xmath369 to 3 significant digits , a good estimate of @xmath52 has already been obtained.,title="fig:",scaledwidth=49.0% ] versus @xmath8 .
the red , blue and black curves indicates values of @xmath369 that define trajectories terminating at the near - field equilibrium point with @xmath301 , @xmath303 , and @xmath396 respectively .
panels ( b)-(d ) show zoomed - in regions from panel ( a ) near @xmath10 , @xmath397 , and @xmath398 respectively.,title="fig : " ] + along the red and blue curves emanating from the black curve near @xmath375 .
panel ( b ) shows the same plot zoomed in on positive values of @xmath52.,title="fig:",scaledwidth=49.0% ] along the red and blue curves emanating from the black curve near @xmath375 .
panel ( b ) shows the same plot zoomed in on positive values of @xmath52.,title="fig:",scaledwidth=49.0% ] with @xmath57 for various different values of @xmath399 and @xmath400 .
panel ( b ) : plots of the trajectories emanating from @xmath401 and @xmath402 for @xmath375 .
the constant to which these trajectories tend in the far - field is selected to be the corresponding value of @xmath282 .
, title="fig:",scaledwidth=49.0% ] with @xmath57 for various different values of @xmath399 and @xmath400 .
panel ( b ) : plots of the trajectories emanating from @xmath401 and @xmath402 for @xmath375 .
the constant to which these trajectories tend in the far - field is selected to be the corresponding value of @xmath282 .
, title="fig:",scaledwidth=49.0% ] having successfully found suitable solutions for @xmath53 , we now proceed to compute suitable solutions for @xmath55 . as discussed in
[ h - plus - connect ] , we can numerically construct a unique trajectory from small to infinite values of @xmath55 .
to do so , we integrate the system ( [ zero - dynamics ] ) from the equilibrium point @xmath213 in the near - field towards the equilibrium point @xmath169 of the system ( [ infinity - dynamics ] ) in the far - field .
the numerical procedure is carried out as follows : select a value of @xmath404 . since @xmath213 is an equilibrium point , it is not possible to escape @xmath213 in finite time .
we therefore begin integration of the system ( [ zero - dynamics ] ) by taking a small step , say @xmath405 , away from @xmath213 using the relevant asymptotic behaviour . using ( [ center - manifold - zero ] ) and ( [ center - manifold - zero - dynamics ] ) for @xmath215
, we find that a trajectory exiting the equilibrium point along the center manifold , @xmath214 , has the local asymptotic behaviour @xmath406 for a small positive value of @xmath405 .
by contrast , using ( [ stable - manifold - zero ] ) and ( [ stable - manifold - zero - dynamics ] ) for @xmath407 , we find that a trajectory along the unstable manifold @xmath216 has the local asymptotic behaviour @xmath408 for a small positive value of @xmath405 .
having selected values for both @xmath57 and @xmath405 , either ( [ numeric - begin - t - positive - positive ] ) or ( [ numeric - begin - t - positive - negative ] ) define unique ( pseudo-)initial conditions to begin integrating the system ( [ zero - dynamics ] ) in the direction of increasing time @xmath85 towards the far - field .
we proved in corollary [ coral-1 ] , that the ultimate fate of all such trajectories , in variables @xmath218 , is approaching the equilibrium state @xmath169 for some @xmath257 .
thus , by continuing integration of the system ( [ zero - dynamics ] ) to some large value of @xmath85 , denoted by say @xmath409 , and reading off the value @xmath410 at @xmath411 , we can obtain an arbitrarily accurate approximation of the corresponding value of @xmath282 that is obtained in the far - field a higher degree of accuracy can be achieved by simply increasing the value of @xmath409 . for this purpose we found ode45 with the majority of the default setting to be sufficiently robust . to ensure high numerical accuracy , at the cost of a relatively small increase in computation time ,
both abstol and reltol were decreased to @xmath349 .
in contrast to the case for solutions @xmath53 , we found it unnecessary to ` switch ' from integrating the near - field system ( [ zero - dynamics ] ) to the far - field system ( [ infinity - dynamics ] ) .
typically , we found that taking @xmath412 gave an approximation of @xmath282 correct to 8 significant digits . carrying out this procedure for a variety of choices of @xmath57 we are able to trace out the form of the piecewise @xmath102 map between @xmath57 and @xmath282 defined earlier in ( [ two - maps ] ) . in figure
[ t - positive - map ] , we show this map for @xmath413 and @xmath400 ( see panel ( a ) ) , as well as some representative trajectories of the system ( [ zero - dynamics ] ) for @xmath375 ( see panel ( b ) ) .
in addition to the results shown , other computations for different values of @xmath8 were also carried out and it appears generic that @xmath282 is a monotonically increasing function of @xmath57 .
crucially , it appears that range of the map ( [ two - maps ] ) is the entire semi - axis @xmath414 for @xmath282 .
we have demonstrated that : ( i ) for each value of @xmath59 there exists _ at least _ one value of @xmath363 ( different from the trivial case @xmath391 ) that defines a trajectory emanating from @xmath357 and terminating at a @xmath299 , and thus a suitable solution for @xmath53 , and ; ( ii ) for every value of @xmath415 there exists a unique corresponding value of @xmath282 , thereby defining an infinite family of suitable solutions for @xmath55 .
the one remaining step is therefore to invoke the matching condition ( [ far - field - matching ] ) .
this condition is equivalent to requiring that the far - field behaviour of @xmath55 is characterized by @xmath363 .
thus , given a solution for @xmath53 , the matching condition ( [ far - field - matching ] ) specifies a unique choice of @xmath363 , a unique @xmath57 , and thus a unique solution for @xmath55 , thereby closing the problem .
the solutions found here for @xmath375 and @xmath400 show a qualitative , although not quantitative , agreement with those reported in @xcite . here
, we found that @xmath416 and @xmath417 in foster _ et .
@xcite , they claimed that for @xmath375 : @xmath418 , @xmath419 , and @xmath420 , whereas for @xmath13 : @xmath421 , @xmath422 , and @xmath423 .
additionally , they claim another suitable solution for @xmath60 : @xmath424 , @xmath425 , and @xmath426 . in contrast , here we found that no such solution with @xmath367 exists .
notably this value of @xmath369 reported in @xcite for @xmath12 is very close to the value of @xmath279 . for @xmath60 , using our numerical approach we have been able to identify three other solutions with @xmath366 , namely : @xmath427 a_- \approx -0.932 , \quad a_+ \approx -30.625 , \quad x_0^
* \approx 0.137 , \\*[1 mm ] a_- \approx -0.546 , \quad a_+ \approx -166.623 , \quad x_0^ * \approx 0.0592 . \end{array } \right.\ ] ] we believe that the origin of these discrepancies is due to the low accuracy of the numerical scheme used in @xcite .
indeed , in @xcite , the solutions for @xmath53 were computed by identifying the value of @xmath52 which characterizes solutions in the near - field that extend into far - field with the requisite behaviour , as in panel ( a ) of figure [ return - to - zero ] .
solutions for @xmath55 were computed by finding the value of @xmath57 inferred ( via shooting from the far - field toward the near - field ) by invoking the matching condition in the far - field as in panel ( b ) on figure [ t - positive - map ] .
both numerical methods used in @xcite are ill - posed . here
, we pose the numerical problem as a shooting scheme for uniquely defined piecewise @xmath102 scalar functions , _
i.e. _ the maps defined in ( [ two - maps ] ) and ( [ maps - negative ] ) .
we therefore believe that the results obtained here are more reliable than those in @xcite .
at 10 equally spaced values of @xmath3 between @xmath428 and @xmath429 , and for @xmath430 to demonstrate the continuity of @xmath84 across @xmath431 .
solid and dashed curves show the solution for @xmath343 and @xmath403 respectively .
panel ( a ) shows the anti - reversing dynamics for @xmath60 , @xmath432 , @xmath433 and @xmath434 .
panel ( b ) shows the reversing dynamics for @xmath377 , @xmath421 , @xmath435 and @xmath393.,title="fig:",scaledwidth=49.0% ] at 10 equally spaced values of @xmath3 between @xmath428 and @xmath429 , and for @xmath430 to demonstrate the continuity of @xmath84 across @xmath431 .
solid and dashed curves show the solution for @xmath343 and @xmath403 respectively .
panel ( a ) shows the anti - reversing dynamics for @xmath60 , @xmath432 , @xmath433 and @xmath434 .
panel ( b ) shows the reversing dynamics for @xmath377 , @xmath421 , @xmath435 and @xmath393.,title="fig:",scaledwidth=49.0% ] this work has focused on constructing local ( in both space and time ) self - similar reversing and anti - reversing solutions to the nonlinear diffusion equation ( [ heat ] ) with @xmath59 and @xmath436 .
we have demonstrated how the dynamical theory combined with the numerical scheme can be used to furnish suitable solutions to the differential equations ( [ ode ] ) for @xmath53 and @xmath55 . via the self - similar reductions
( [ reduction ] ) , the solutions to these differential equations can be transformed into physically meaningful solutions for @xmath84 to the nonlinear diffusion equation ( [ heat ] ) with @xmath59 and @xmath436 , which is completed with the no flux boundary condition ( [ zero - flux2 ] ) and the condition the @xmath40 . in this final section
we shall discuss the connection between the self - similar solutions found here and the dynamics of the model ( [ heat ] ) .
it is well - known , and can be readily verified , that the nonlinear diffusion equation with absorption ( [ heat ] ) admits two different travelling wave solutions .
seeking such solutions near a left interface @xmath16 that have the form @xmath437 , for some function @xmath438 and constant @xmath439 , we find that @xmath440 \label{lin - nose } h \sim ( \dot{\ell})^{-1 } ( x-\ell(t ) ) \quad \mbox{as } \quad x \searrow \ell(t ) , \quad \mbox{for } \quad \dot{\ell } > 0.\end{aligned}\ ] ] the former , is an advancing wave local to a left interface whose motion is driven by diffusion , whereas the latter is a receding wave driven by absorption .
this study has therefore elucidated the process by which the wave ( [ non - lin - nose ] ) _ becomes _ ( [ lin - nose ] ) , giving rise to a reversing interface , or vice versa , giving rise to an anti - reversing interface . in
[ t - neg - shooting ] we used the result of lemma [ lemma - continuation ] to numerically construct suitable solutions for @xmath53 . for each value of @xmath59 we identified _ at least _
one suitable solution , defined by a pair of values of @xmath52 and @xmath369 , in addition to the exact solution ( [ exact - solution ] ) in the original time and space variables , this exact solution corresponds to a steady solution for @xmath84 and thus does not constitute a reversing nor an anti - reversing solution . in
[ t - pos - solutions ] , we used lemma [ lemma - connection ] to formulate a numerical scheme for constructing solutions for @xmath55 defined by pairs of values of @xmath57 and @xmath369 .
we showed that the map ( [ t - positive - map ] ) is one - to - one and its range is the entire semi - axis @xmath414 for @xmath282 .
importantly , for each value of @xmath59 , we found that : ( i ) if @xmath407 then @xmath441 , and ( ii ) if @xmath442 then @xmath443 where @xmath444 .
the final stage in constructing solutions for @xmath84 is to invoke the matching condition ( [ far - field - matching ] ) that ensures continuity of @xmath84 across @xmath431 .
owing to the aforementioned properties of ( [ t - positive - map ] ) we are forced to reject any solution for @xmath55 that is defined by a trajectory with @xmath366 and @xmath445 or @xmath367 and @xmath446 on the basis that it necessarily can not match to solution for @xmath53 . in summary
we have found that for @xmath447 up to 5 different solutions are available with @xmath448 .
for @xmath449 there is at least one solution with @xmath450 . for @xmath451 an additional branch of solutions with @xmath448
emerges , whereas for @xmath452 , there is another branch of solutions with @xmath453 .
for @xmath454 it seems quite possible that yet more branches of solutions will emerge .
there is a distinct difference between the interpretation of solutions with @xmath450 in terms of the original model ( [ heat ] ) compared to those with @xmath448 .
the former , correspond to a reversing solution where the left interface advances for @xmath343 , with the behaviour ( [ non - lin - nose ] ) , and then subsequently recedes for @xmath403 with the behaviour ( [ lin - nose ] ) .
contrastingly , the latter corresponds to an anti - reversing solution where the interface recedes for @xmath343 , with the form ( [ lin - nose ] ) , and then advances according to the form ( [ non - lin - nose ] ) for @xmath403 .
one representative local solution for @xmath84 for both types of behaviour one reversing and one anti - reversing are shown in figure [ generic - figure - name ] .
some natural open questions raised by this study are : ( i ) whether any self - similar solutions with non - monotone profiles ( in @xmath120 ) exist _
i.e. _ solutions that do not satisfy ( [ ( ii ) ] ) ; ( ii ) whether the self - similar solutions identified here are stable in the context of the model ( [ heat ] ) , and ; ( iii ) if more than one reversing or anti - reversing solution is stable for a particular value of @xmath8 , what is the mechanism for selecting the appropriate self - similar solution at a particular reversing or anti - reversing event
. * acknowledgements . *
thanks m. chugunova and r. taranets , while j.f .
thanks j. r. king and a. d. fitt for useful discussions regarding this project .
is supported by a postdoctoral fellowship at mcmaster university .
he thanks b. protas for hospitality and many useful discussions .
a part of this work was completed during the visit of d.p . to claremont graduate university .
the work of d.p . is supported by the ministry of education and science of russian federation ( the base part of the state task no .
2014/133 , project no . | we consider the slow nonlinear diffusion equation subject to a constant absorption rate and construct local self - similar solutions for reversing ( and anti - reversing ) interfaces , where an initially advancing ( receding ) interface gives way to a receding ( advancing ) one .
we use an approach based on invariant manifolds , which allows us to determine the required asymptotic behaviour for small and large values of the concentration .
we then ` connect ' the requisite asymptotic behaviours using a robust and accurate numerical scheme . by doing so , we are able to furnish a rich set of self - similar solutions for both reversing and anti - reversing interfaces .
nonlinear diffusion equation , slow diffusion , strong absorption , self - similar solutions , invariant manifolds , reversing interface , anti - reversing interface |
diffuse reflectance spectroscopy in the visible and near - infrared range ( vis - nir drs ) has proved to be useful to assess various soil properties @xcite .
it can be employed to provide more data rapidly and inexpensively compared to classical laboratory analysis .
therefore , drs is increasingly used for vast soil surveys in agriculture and environmental research @xcite . recently
, several studies have shown the applicability of vis - nir drs _ in situ _ as a proximal soil sensing technique @xcite . to predict soil properties from soil spectra ,
a model is calibrated , often using partial least squares ( pls ) regression . however , when calibration is based on air - dried spectra collected under laboratory conditions , predictions of soil properties from field spectra tend to be less accurate @xcite .
usually , this decrease in accuracy is attributed to varying moisture between air - dried calibration samples and field spectra recorded with a variable moisture content .
different remediation techniques have been proposed , ranging from advanced preprocessing of the spectra @xcite to `` spiking '' the calibration set with field spectra @xcite . in our study
, we adopt a slightly different view on the calibration problem .
it does not only apply to the varying moisture conditions between the calibration data set and the field spectra .
indeed , it is also valid if we want to predict soil properties in a range where calibration samples are rare .
mining with rarity or learning from imbalanced data is an ongoing research topic in machine learning @xcite . in imbalanced data sets
frequent samples outnumber the rare once .
therefore , a model will be better at predicting the former and might fail for the latter .
two different approaches exist to take care of the data imbalance : we can either adjust the model or `` balance '' the data .
the latter approach has the advantage that we can use the usual modelling framework .
synthetic minority oversampling technique ( smote ) is one way to balance the data .
it was first proposed for classification @xcite and recently for regression @xcite .
smote oversamples the rare data by generating synthetic points and thus helps to equalize the distribution . in this study
, we propose a strategy to increase the prediction accuracy of soil properties from field spectra when they are rare in calibration .
the goal of this study is to build a calibration model to predict soil organic carbon content ( socc ) from field spectra by air - dried samples spiked with synthetic field spectra .
the studied soil was sampled at the southern slopes of mt .
kilimanjaro , tanzania ( 3@xmath3 4@xmath4 33@xmath5 s , 37@xmath3 21@xmath4 12@xmath5 e ) in coffee plantations . due to favourable soil and climate in this region ,
extensive coffee plantations constitute a frequent form of land use .
we took 31 samples for calibration at 4 different study sites . for validation
, we scanned 12 field spectra at a wall of a soil pit and sampled soil material for chemical analysis at the scanned spots .
we call these validation field spectra f. after collection , the calibration samples were dried in an oven at 45@xmath3c and sieved @xmath6 2 mm .
subsequently , they were scanned with an agrispec portable spectrophotometer equipped with a contact probe ( analytical spectral devices , boulder , colorado ) in the range 3502500 nm with 1 nm intervals .
the same spectrometer was used in the field .
the instrument was calibrated with a spectralon white tile before scanning the soil samples . for the measurement
, a thoroughly mixed aliquot of the sample was placed in a small cup and the surface was smoothed with a spatula .
each sample was scanned 30 times and the signal averaged to reduce the noise . in the following , we call this calibration data set l. socc was measured in a cns - analyser by high temperature combustion with conductivity detectors .
to generate new data to spike the calibration data set l , we used smote @xcite and its extension for regression @xcite .
this algorithm consists of generating new synthetic data using existing data and is summarized below . in our case , we generated new spectra and the related socc using the field spectra f. the new spectra are created by calculating the difference between a field spectrum and one of its nearest neighbours and adding this difference ( weighted by a random number between 0 and 1 ) to the field spectrum .
the socc of the synthetic spectrum is then a weighted average between the socc of the field spectrum and the used nearest neighbour .
smote has two parameters , namely @xmath7 , the number of points to generate for each existing point ( given in percent of the whole data set ) and @xmath8 , the number of nearest neighbours . to study the influence of these parameters we generated six different synthetic data sets s1 through s6 , varying @xmath9 and @xmath10 .
@xmath11 $ ] : original sample @xmath12 @xmath13 $ ] : target value of original sample @xmath14 @xmath15 $ ] : synthetic sample @xmath16 @xmath17 $ ] : target values of synthetic sample @xmath18 @xmath19 @xmath20 : number of synthetic samples to compute for each original sample generate synthetic samples : @xmath21 compute @xmath8 nearest neighbours for @xmath11 $ ] randomly choose @xmath22 @xmath23 $ ] @xmath24 = { \mathit{orig.s}}[i ] + \mathrm{random}(0,1 ) \times { \mathit{diff}}$ ] @xmath25)$ ] @xmath26 @xmath27 we corrected each spectrum ( calibration , validation and synthetic ) for the offset at 1000 and 1830 nm and kept only parts with a high signal - to - noise ratio ( 4502400 nm ) .
then , we transformed the spectra to absorbance @xmath28 and smoothed them using the singular spectrum analysis ( ssa ) .
ssa is a non - parametric technique to decompose a signal into additive components that can be identified as the signal itself or as noise @xcite . finally , we divided each spectrum by its maximum and calculated the first derivative . in order to assess similarities between the calibration , validation and synthetic data sets , we calculated the principal component analysis ( pca ) of the ( uncorrected original ) spectra l and f and projected the synthetic data into the space spanned by the principal components .
we calibrated seven different pls models .
for model i we used the data set l , the spectra scanned under laboratory conditions .
model ii through vii were calibrated on l spiked with synthetic spectra s1 through s6 . to find the best model i through vii
, we varied the number of pls components between 1 and 15 .
based on the predictions in the leave - one - out cross - validation ( loocv ) we calculated the corrected akaike information criterion @xcite aic@xmath29 , where @xmath30 is the number of calibration samples , @xmath31 the number of pls components and @xmath32 the root mean - squared error .
the latter is defined as @xmath33 , where @xmath34 are the predicted and @xmath35 the measured soccs .
we selected the model with the smallest aic@xmath36 as the most plausible .
to assess the model quality , we used the @xmath32 , the mean error @xmath37 and the coefficient of determination @xmath38 , where @xmath39 is the mean socc .
smote has two random components because it selects spectra randomly ( with replacement ) among the nearest neighbours and weights the difference between spectra by a random number ( between 0 and 1 ) . to study the influence of these random steps
, we generated 100 different datasets s1 through s6 .
each data set was then used to spike the calibration data set l , to build a new pls model and to predict the data set f.
the first principal components ( pcs ) explain 85.4% and 11.2% of variance , respectively .
we can clearly identify two distinct groups of samples : the calibration data set l and the field spectra f ( fig .
[ fig : pca ] ) .
in other words , the data sets l and f differ .
the synthetic points that were projected into the space spanned by the pcs resemble the field spectra as expected .
the distinct characteristics of the data sets l and f accord well with the difficulties to predict the data set f by using the laboratory spectra l only ( table [ tab : results : calib ] and table [ tab : results : valid ] ) . although the loocv of model i yields a moderate @xmath32 and a large @xmath1 , the validation on the data set f fails . spiking the calibration data
set l with synthetic spectra increases the prediction accuracy of the socc in the data set f. actually , the @xmath32 decreases and @xmath1 increases with increasing number of synthetic points both for the loocv and the validation ( table [ tab : results : calib ] and table [ tab : results : valid ] ) .
however , the number of model parameters also increases from 2 to 7 .
the monte carlo results show only a small variability in the interquartile range . however , some synthetic data sets in model v produced @xmath1 values smaller than @xmath20.53 , the value we obtain in model i on air - dried samples only .
this might be due to the combination of neighbours during smoting .
in general , models with 5 neighbours were more accurate than those with 3 neighbours .
however , the number of neighbours had a smaller influence on the prediction accuracy than the number of synthetic points .
it is difficult to decide _ a priory _ how many synthetic points should be included in the calibration .
indeed , in a classification problem the goal is to approximate an equal distribution of different classes such that the rare class becomes an ordinary one . in regression , however , we do not know which features of the data make them rare . for our data , the range of socc in the data set l is larger than in the data set f. therefore , we conclude that concentration is not responsible for the difference between these data sets . based on the monte carlo results we chose one synthetic data set from model vi , namely the one with the median number of model parameters and the best @xmath1 in the validation .
thus , the calibration data set includes 31 air - dried and 24 synthetic spectra . compared to model i
, spiking the air - dried data set l with these synthetic spectra clearly improves the prediction of the data set f ( fig .
[ fig : models ] ) . [ cols="<,<,>,>,>,>,>,>",options="header " , ]
we propose a framework to predict soil properties from _ in situ _ acquired field spectra by spiking air - dried laboratory calibration data by synthetic ones generated from these field spectra . in general , the prediction accuracy increases when a sufficient number of synthetic points is included in the calibration . however , because it is difficult to determine this number _ a priori _ , we recommend to generate several synthetic data sets to find an appropriate model .
this study is part of the project dfg for 1246 `` kilimanjaro ecosystems under global change : linking biodiversity , biotic interactions and biogeochemical ecosystem processes '' and was supported by the deutsche forschungsgemeinschaft .
b. stenberg and r. a. viscarra rossel , `` diffuse reflectance spectroscopy for high - resolution soil sensing , '' in _ proximal soil sensing _ , r. a. viscarra rossel , a. b. mcbratney , and b. minasny , eds . , pp .
springer , 2010 .
k. d. shepherd and m. g. walsh , `` infrared spectroscopy - enabling an evidence - based diagnostic surveillance approach to agricultural and environmental management in developing countries , '' , vol .
1 , pp . 119 , 2007 .
vgen , k. d. shepherd , and m. g. walsh , `` sensing landscape level change in soil fertility following deforestation and conversion in the highlands of madagascar using vis - nir spectroscopy , '' , vol .
3 , pp . 281294 , 2006 .
r. a. viscarra rossel , s. r. cattle , a. ortega , and y. fouad , `` in situ measurements of soil colour , mineral composition and clay content by vis
nir spectroscopy , '' , vol .
3 , pp . 253266 , 2009 .
t. h. waiser , c. l. s. morgan , d. j. brown , and c. t. hallmark , `` in situ characterization of soil clay content with visible near - infrared diffuse reflectance spectroscopy , '' , vol .
2 , pp . 389396 , 2007 .
b. minasny , a. b. mcbratney , v. bellon - maurel , j .-
roger , a. gobrecht , l. ferrand , and s. joalland , `` removing the effect of soil moisture from nir diffuse reflectance spectra for the prediction of soil organic carbon , '' , vol .
118124 , 2011 . | diffuse reflectance spectroscopy is a powerful technique to predict soil properties .
it can be used _ in situ _ to provide data inexpensively and rapidly compared to the standard laboratory measurements .
because most spectral data bases contain air - dried samples scanned in the laboratory , field spectra acquired _ in situ _ are either absent or rare in calibration data sets .
however , when models are calibrated on air - dried spectra , prediction using field spectra are often inaccurate .
we propose a framework to calibrate partial least squares models when field spectra are rare using synthetic minority oversampling technique ( smote ) .
we calibrated a model to predict soil organic carbon content using air - dried spectra spiked with synthetic field spectra .
the root mean - squared error of prediction decreased from 6.18 to 2.12 mg g@xmath0 and @xmath1 increased from @xmath20.53 to 0.82 compared to the model calibrated on air - dried spectra only .
= 1 diffuse reflectance spectroscopy , soil , partial least squares , calibration , smote |
the class of compact steep - spectrum sources ( csss ) has been discussed by many authors ( e.g. pearson et al .
1985 ; fanti it et al.1990 ) .
it is now generally believed that csss are physically small objects with sub - galactic dimensions , whose structure and sizes are possibly affected by the ambient gas in the central regions of the parent optical objects .
double - lobed sources represent @xmath0 of the objects in a sample of csss ( fanti et al .
1995 ) selected from the 3cr catalogue ( laing et al . 1983 ) and from the peacock & wall ( 1982 ) catalogue for a linear size @xmath1kpc , spectral index @xmath2 ( s@xmath3 ) and radio power @xmath4whz@xmath5 at 178mhz for the 3cr or @xmath6whz@xmath5 at 2.7ghz ( @xmath7kmsec@xmath5mpc@xmath5 and @xmath8 ) .
fanti at al .
( 1995 ) suggest that double - lobed csss represent the _ young _ ( @xmath9 yrs ) precursors of the larger double - lobed radio sources .
the remaining 30% might represent a different population , made of sources where the jets are brightened either by doppler boosting or by interaction with the ambient medium .
there are observational results that suggest a connection between radio source evolution and the css and gps ( gigahertzed peaked sources ) phenomena ( see , for example , odea et al .
1991 ; stanghellini et al .
these results also support the hypothesis that the css phenomenon results from expansion within the ambient environment of radio sources at intermediate and high redshifts .
this may be due to the existence of a more dense ism in the inner portions of host galaxies with csss .
this density enhancement would influence the behaviour of the radio - emitting regions . generally speaking ,
the cores are often weak in csss , suggesting that if boosting is present , then jets and cores are misaligned with respect to the kpc - scale jets ( spencer 1994 ) .
moreover , distorted structures on the milli - arcsecond scale are common in those csss which are quasars with strong jets . in this paper
we focus our attention on a sample of csss selected because of the large bends in the jets in their inner regions
. we will consider whether the large misalignments are due to relativistic effects or produced by jet - cloud interactions in narrow line regions . in the next section we will describe the sample , our observations of four of the selected sources and the source structures .
we will then summarize the source parameters derived from the observations , and estimate the probability of the collision of a jet with dense gas cloud .
finally , we will discuss the results of our investigations .
we have used the images of sources classified as csss from the compilation by dallacasa & stanghellini ( 1990 ) . from this , we have chosen a sub - sample using quite simple criteria . first , all the sources dominated by jet emission were selected . from this , a list of sources was chosen to include all the sources showing large bent jets on the sub - arcsecond scale . a bend in a jet
is considered _ large _ when the jet major axis changes direction by an angle @xmath10 ( @xmath11 ) .
such a value was adopted to avoid ambiguities in selecting candidates from the published images .
these are due to the uncertainty in determining the position angle ( pa ) of the jet major axis .
we are also aware that the images were made from observations done using a variety of arrays and frequencies , some of which were inappropriate for our purpose .
the selected sources are listed in table 3 below .
observations of four of them by us are presented here .
the four sources ( 0548@xmath12165 , 1741@xmath12279 , 2033@xmath12187 and 2147@xmath12145 ) have been observed with vlbi at 1.6ghz and 5ghz using different arrays and recording systems .
calibration observations on largely - unresolved radio sources were made during each observing session along with the observations of the target source .
table 1 summarizes the observations . after the correlation process
the bulk of the data reduction was done using the _ aips _ package .
the sources have been imaged using both _ aips _ and _ difmap _ ( shepherd et al .
1995 ) .
the source parameters in table 2 are as follows : column 1 : source name ; column 2 : observing frequency ; columns 3,4,5 : beam major axis , minor axis in mas and pa in degrees ; column 6 : r.m.s .
noise in the map far from the source ; column 7 : component label ; column 8,9,10 : major axis , minor axis in mas and pa in deg of each component ; column 11 : component peak brightness in mjy / beam ; column 12 : component total flux density in mjy ; column 12 : component spectral index .
0548@xmath12165 & b , s , j2,l , i , o , w & 5 ghz & mk3-b & 12may1995 & 12 hrs & mpifr & b , s , j1,l , o , w , k , h , g , & 1.6 ghz & mk2 & 21sep1992 & 04 hrs & cit & vlba & & & & & 1741@xmath12279 & b , j2,l , i , o , w & 5 ghz & mk3-b & 14sep1993 & 10 hrs & mpifr & b , j1,l , o , w , k , h , g , & 1.6 ghz & mk2 & 21sep1992 & 09 hrs & cit & vlba & & & & & 2033@xmath12187 & b , j2,l , i , o , w & 5 ghz & mk3-b & 01mar1994 & 10 hrs & mpifr & s , o , w , c , d , k , h , g , & 1.6 ghz & mk2 & 14mar1986 & 08 hrs & cit & hr , fd , ov & &
& & & 2147@xmath12145 & b , l , o , w & 5 ghz & mk3-b & 26feb1994 & 11 hrs & mpifr _ note : _ * b * effelsberg , * c * cambridge , * d * defford , * fd * fort davis , * g * green bank , * h * hartebeestehoek , * k * haystack , * hr * hat creek , * j1 * jodrell - lovell , * j2 * jodrell - mk2 , * l * medicina , * i * noto , * o * onsala85 , * ov * owens valley , * s * simeiz , * w * westerbork , * vlba : * pt , kp , la , br , fd , nl , ov & dens .
& @xmath13 & @xmath14 & maj & min & pa & noise & & maj & min & pa & peak & total & & mhz & mas & mas & @xmath15 & mjy / b & & mas&mas&@xmath15&mjy / b & mjy & 0548@xmath12165 & 1662 & 14 & 10 & 8 & 10 & a & - & - & - & 116 & 131 & 0.2 & & & & & & j & - & - & - & 62 & 162 & & & & & & & b & 20 & 9&15 & 693 & 1571 & 1.3 & & & & & & c & - & - & - & 16 & 18 & & 4975 & 6 & 6 & & 0.2 & a & - & - & - & 92.6 & 109.1 & & & & & & & j & - & - & - & 8.3 & 49.1 & & & & & & & b & 14 & 9 & 9 & 110.9 & 376.5 & & & & & & & c & 21 & 9&34 & 15.8 & 92.0 & 1741@xmath12279 & 1662 & 12 & 9 & 2&2.5 & a & - & - & - & 53 & 57 & @xmath160.3 & & & & & & b & 29&11&104 & 10 & 44 & 1.3 & & & & & & c & 36 & 9&110 & 6 & 25 & 1.8 & & & & & & d & - & - & - & 18 & 23 & 1.3 & & & & & & e & 13 & 8 & 7 & 77 & 178 & 1.4 & 4960&8.5 & 5.7 & 79&0.12 & a & 4 & 2&69 & 75.9 & 82.9 & & & & & & & b & 13&12&108 & 3.3 & 10.7 & & & & & & & c & 6 & 2&172 & 2.3 & 3.5 & & & & & & & d & 16&11&72 & 18.5 & 42.7 & & & & & & & e & 12 & 6&29 & 18.5 & 42.7 & 2033@xmath12187 & 1662 & 19 & 3 & 9 & 0.7 & a & 8 & 4 & 164 & 303.2 & 499.2 & 1.2 & & & & & & a@xmath17&33&3 & 164 & 12.6 & 23.9 & & & & & & & a@xmath18&12 & 4&33 & 64.0&154.8 & & & & & & & b & 10 & 3&175 & 63.5 & 133.9 & 1.1 & & & & & & c & && & 13.6 & 14.4 & & & & & & & d & 10 & 5 & 168 & 80.3 & 193.2 & 1.2 & 4960&9.7&4.8 & 65 & 0.07 & a & 7 & 5&71 & 94.8 & 136.0 & & & & & & & b & 7 & 1&79 & 33.1 & 38.6 & & & & & & & d & 7.3&4.4&68 & 10.6 & 12.6 & & & & & & & d@xmath17&&& & 29.9 & 41.5 & 2147@xmath12145 & 4960&12.6&8.2&70 & 0.2 & a & && & 394.5 & 486.5 & & & & & & & b & 19 & 6&55 & 40.5 & 56.5 & & & & & & & c & 14 & 7&107 & 9.1 & 9.6 & & & & & & & d & 17&1 & 153 & 23.8 & 61.5 & & & & & & & e & && & 5.0 & 3.6 & & & & & & & f & && & 6.7 & 6.1 & the four sources above plus the other sources listed in the following tab.3 are described briefly here : * 3c43 * ( 0127@xmath12233 ) the vla image from pearson et al .
( 1985 ) shows a highly misaligned triple structure .
the merlin map by sanghera et al .
( 1995 ) shows that the northernmost component a in the evn @xmath1918 cm image by spencer et al .
( 1991 ) is likely to be the core .
this is consistent with the @xmath1950 cm vlbi observations of rendong et al .
spencer ( 1994 ) imaged 3c43 at @xmath1918 cm with merlin .
that image shows a bridge of emission between the main components and the northern component .
the jet is straight and collimated for the first 150 mas , then it changes pa by @xmath20 and , after 75 mas the pa changes again by @xmath21 pointing towards east .
according to junor et al .
( in preparation ) the central component is 3.6% polarized at 8.4ghz with the vla .
the polarized intensity is sufficiently strong to provide an estimate of rotation measure , rm , of @xmath22rad m@xmath23 in the source s rest frame .
the intrinsic magnetic field direction in the central component follows the curvature of the source faithfully . * 3c99 * ( 0358@xmath12004 ) the radio source 3c99 has a triple structure on arcsecond scales .
the outer components are located rather asymmetrically relative to the nucleus and have very different surface brightnesses ( mantovani et al .
1990 ) . along the major axis
, 3c99 has an angular size of @xmath246 arcsec and a linear size of @xmath2421 kpc .
it is associated with an n galaxy which has been detected close to the central component ( spinrad et al .
the central component is unpolarized in 8.4ghz vla observations ( mantovani et al .
1997 ) and it is likely to be the nucleus of 3c99 .
the vlbi image of the source at @xmath1918 cm shows that the central component consists of several blobs of emission with the two prominent ones being significantly misaligned with the overall axis of the source . * 3c119 * ( 0429@xmath12415 ) an image of the source structure with an angular resolution of 5 mas is presented by nan ren - dong et al .
component a has an inverted spectral index and it is identified with the core of 3c119 .
the morphology of 3c119 is rather peculiar .
the jet emerging from the core component is not well collimated .
it contains several blobs of emission . the major axis pa changes direction by @xmath25 to reach the component c , at about 40 mas from the core .
from there , it changes direction several times to form an almost circular structure .
the core has a radio luminosity which is @xmath26 of the source total luminosity at 5 ghz .
taylor et al .
( 1992 ) listed 3c119 among sources with very large rotation measures ( rm@xmath273400 rad m@xmath23 ) .
a number of possible explanations have been discussed in the paper by nan ren - dong et al.(1991 ) for the source s brightness distribution ( e.g. rotational shear of the radio jet by an ambient rotating gaseous disk , precession in the nucleus , and the source expanding in a cavity in the interstellar medium ) , but those authors admit that they can not reach a satisfactory conclusion . * 3c147 * ( 0538@xmath12498 ) 3c147 has been observed by several investigators ; see , for example , the collection of images by alef et al .
those authors have also observed 3c147 with 5ghz vlbi over three epochs .
the source shows an unusually - complex , nonlinear structure which varies with time .
superluminal separation of two components in the core region was observed also .
new 8.4ghz data ( alef , private communication ) confirm a mildly - superluminal separation velocity of @xmath28 .
the jet is embedded in a diffuse emission region and shows a change in the projected orientation of its major axis of @xmath29 at @xmath30 mas from the core . a vla image at 1 ghz ( van bruegel et al .
1984 ) shows a weak component north to the main one in a position which is opposite to the jet respect to the core .
3c147 shows large rm s of @xmath31 and @xmath32 rad m@xmath23 in the rest frame of the source for the main component and for the extension to the nne respectively ( junor et al .
submitted ) .
* 0548@xmath12165 * on arcsecond scales , the source shows an asymmetric structure with a strong , unresolved component coincident with a quasar at z=0.474 and a much weaker secondary component about 3 arcsec to the north .
the faint , extended component is weakly polarized .
most of the polarization comes from the main component .
this source has a rm of 1934 rad m@xmath23 .
the polarized emission is also strongly depolarized between 15 ghz and 5 ghz ( mantovani et al .
1994 ) .
the mas scale structure of 0548@xmath12165 from a full - track , global @xmath19 18 cm program is shown in fig.1 .
the source has a straight jet pointing west which changes direction dramatically @xmath29 at @xmath3380 mas from the core .
the de - rotated magnetic field is aligned parallel to the east - west direction ( mantovani et al .
1994 ) like the straight part at the beginning of the jet .
evn observations at @xmath19 6 cm ( fig.2 ) confirm the main structure seen at @xmath1918 cm .
fig.2 also shows that the jet increases its width but it remains collimated and it seems to show a wiggling structure .
the core is thought to be the easternmost component , since it has a flat spectrum .
the detection of the secondary component about 3 arcsec north , suggests that either there is a counter jet or the vlbi jet bends back to the north . * 3c287 * ( 1328 + 254 ) the image of 3c287 at @xmath196 cm with 7 mas resolution shows a regularly curving jet - like structure ; this bears some similarity to that in 3c119 ( fanti et al .
the source brightness decreases smoothly along the curved jet .
it is not clear where the core is located so we can not add its parameters to table 3 .
fanti et al.1989 suggest that the main component ( a in their fig.3 ) is the possible site for the core .
this is the most compact feature visible at @xmath196 cm but it does have a spectral index of @xmath330.5 between @xmath1918 cm and @xmath196 cm ( nan ren - dong et al .
1988 ) .
* 1442@xmath12101 * ( oq172 ) this object is unresolved by merlin .
the vlbi image shows a very compact source @xmath34 mas in extent at @xmath1918 cm ( dallacasa et al .
it has a core - jet structure with the core located in the northern part of the radio emission .
the source shows a bend in the jet major axis pa of @xmath29 at a separation of @xmath35mas .
1442@xmath12101 has a redshift of 3.544 and a very high integrated rm of 22400 rad m@xmath23 in the source s rest frame ( taylor et al .
recent vlba observations at 5ghz by udompresert et al .
( 1997 ) indicate that the rm is 40,000 rad m@xmath23 in the rest frame of the quasar . at 10 mas from the nucleus the rm falls to less than 100 rad m@xmath23 .
the very high rm is found near to the core ; it is likely that this is not associated with material which could influence the bending of the jet . * 1629@xmath12680 * this source was observed at @xmath1913 cm and @xmath193.6 cm ( dallacasa et al .
the x - band image shows a straight jet @xmath36mas long .
the s - band image shows a mild bend in the jet which finally changes the direction of the axis by @xmath29 at @xmath37mas from the core . *
1741@xmath12279 * the vla map at 8.4 ghz of 1741@xmath12279 shows two bright components roughly aligned e - w , a wiggling jet 5 arcsec long aligned n - s to the north of the two bright components and a region of weak emission 4 arcsec to the south - south - east .
the two compact components have about 5% of the total polarized emission at 8.4 ghz ( mantovani et al .
the magnetic field is parallel to the curved line joining the two components , changing direction smoothly by an angle of @xmath38 ; this suggests that there is a bend in the emitting region .
the vlbi @xmath39 cm and @xmath40 cm maps , figures 3 & 4 respectively , show several knots along the axis between these two components .
the eastern component is the nucleus ( it has an inverted spectrum ) and the western one shows an elongation north - south . when compared with the @xmath41 cm vla image , it is possible to imagine that this elongation is the location of a sharp bend or cusp in the jet s apparent path . * 2033@xmath12187 * this source is unresolved by the vla a - array at 15ghz ( angular size @xmath42arcsec ) and is unpolarized .
we present two vlbi images here .
the first was obtained with a global array at @xmath39 cm ( fig.5 ) ; the second with the evn at @xmath40 cm ( fig.6 ) . in these images
, we see a straight jet and a dramatic bend at a small distance , @xmath43mas , from component ` a ' .
note that component ` a ' is the brightest feature in 2033@xmath12187 at both frequencies and shows a steep spectral index ( @xmath44 ) .
the position of the core is unknown .
the resolution in the north - south direction is insufficient to resolve the westermost component at @xmath39 cm which has a low brightness region of emission extending south .
the @xmath40 cm image shows a linear morphology from component ` a ' to ` d ' , where the jet changes direction sharply .
* 2147@xmath12145 * vlbi observations of 2147 + 145 were made for the first time by cotton et al .
( 1984 ) at @xmath39 cm . that image shows a core - jet structure with the major axis pa of 45@xmath45 .
recent vla images of 2147 + 145 made at 8.4 and 15ghz , when compared with the previous observations by cotton ( 1983 ) , show that the total flux density at 15 ghz has increased by @xmath3313% .
in addition , a new component is found north of the core in pa @xmath46 , separated by 0.35 arcsec ( mantovani et al .
this component lies in a direction which differs by more than @xmath47 from that found for the jet in the @xmath39 cm vlbi map of cotton _ et al . _ ( 1984 ) .
the vlbi source can be modelled by four gaussian components that lie along a path which bends smoothly towards north .
it is reasonable to expect that the jet continues with an increasingly pronounced bend to allow the flow to reach the component to the north .
the evn observations of 2147 + 145 at @xmath196 cm ( fig.7 ) have produced an image that shows a core - jet structure which , at first sight , agrees with the @xmath1918 cm image of cotton ( 1984 ) .
a sharp bend in the jet occurs at @xmath3340 mas from the core .
the flow changes direction by @xmath48 and a couple of weaker components are detected further north .
the component ` a ' , possibly the core , has a steep spectrum .
the spectral index ranges from 0.38 ( peak emission ) to 0.7 ( total emission ) between @xmath1918 cm and @xmath196 cm .
some of the parameters for the overall mas - scale structure that can be derived from the available images of these sources are summarized in table 3 .
the table contents are as follows : column 1 : name ; column 2 : references for the images ; column 3 : separation ( in milliarcsecond ) between the core and the first large bend in the jet ; column 4 : linear separation , assuming @xmath49 km sec@xmath5mpc@xmath5 and @xmath8 ; column 5 : difference in the jet pa before and after the bend ( in degrees ) ; column 6 : redshift ; column 7 : optical identification ; column 8 : rotation measure in radm@xmath23 corrected by the redshift .
the symbol ` ' means measurements are unavailable .
the symbol ` n ' means polarization was not detected ; ; column 9 : references for rms .
rm@xmath50 & ref & & mas & pc & deg & & & rad m@xmath23 & 0127@xmath12233 3c43 & a & 225 & 780 & 93 & 1.459 & q & @xmath161088 & 1 0358@xmath12004 3c99 & b & 15 & 46 & 60 & 0.425 & g & 0 & 2 0429@xmath12415 3c119 & c & 39 & 142 & 55 & 1.023 & q & 3400 & 3 0538@xmath12498 3c147 & d & 200 & 664 & 90 & 0.545 & q & @xmath163144/630 & 4 0548@xmath12165 & e & 80 & 253 & 90 & 0.474 & q & 1934 & 5 1328@xmath12254 3c287 & f & & & 58 & 1.055 & q & & 1442@xmath12101 oq172 & g & 15 & 39 & 90 & 3.531 & q & 22400 & 3 1629@xmath12680 4c68.18 & h & 76 & 226 & 66 & 2.475 & q & & 1741@xmath12279 b2 & e & 300 & 862 & 84 & 0.372 & q & @xmath16148/@xmath16279 & 2 2033@xmath12187 & e & 40 & 129 & 80 & ( 0.5 ) & & n & 2 2147@xmath12145 & e & 23 & 75 & 90 & ( 0.5 ) & & n & 2 _ note
_ references - column 2 - : a. spencer at al .
( 1991 ) ; b. mantovani et al .
( 1997 ) ; c.nan ren - dong et al .
( 1991 ) ; d. alef et al .
( 1990 ) ; e. this paper ; f. fanti et al .
( 1989 ) ; g. dallacasa et al .
( 1995 ) ; h. dallacasa et al.(1997 ) . -
column 9 - : 1 .
junor et al.(in preparation ) ; 2 .
mantovani et al .
( 1997 ) ; 3 .
taylor et al .
( 1992 ) ; 4 .
junor et al .
( 1997 ) ; 5 .
mantovani et al .
( 1995 ) . because of the selection criteria adopted , the sources listed in table 3 are characterized by the presence of bright jets .
the change in the jet major axis position angle has a mean value of @xmath51 .
the emission from the core compared to the total source emission is weak ( @xmath52 ) in 3c43 , 3c99 , 3c119 and 3c147 .
however , it is strong ( @xmath53 ) in 0548@xmath12165 , 1629@xmath12680 , 1741@xmath12279 and 2147@xmath12145 . in two cases , namely 3c287 and 2033@xmath12187 , it is still unclear where the core is located .
it is worthwhile mentioning that all of the sources in table 3 have been identified with quasars with the exception of 3c99 , which has been classified as an n - galaxy .
note , too , that the sources 2033@xmath12187 and 2147@xmath12145 are associated with unusually - faint optical objects ( cotton et al .
1989 ) .
the sites where the bends occur are very close to the cores of the sources . the angular separation from the core to that location ranges between 15 and 300mas , which corresponds to a linear separation of 40900pc , which is well inside the respective narrow line regions ( nlrs ) . in general , most of the sources have a * very * high rotation measure .
the sources are too compact to allow arcsecond - scale rm images .
those sources which are not polarized are also the most compact among those listed ; this might be interpreted as due to large changes in the magnetic field direction within the angular resolution of the synthesized beam . in such cases ,
vlbi polarimetry is required ; see , for example , the observations of 1442@xmath12101 ( udomprasert et al . 1997 )
the list of csss compiled by dallacasa & stanghellini ( 1990 ) contains @xmath33130 objects . about 30% of these ( i.e. @xmath54 objects )
are dominated by the emission from the jet structure .
table 3 contains 11 objects with strongly bent jets , which represents about 30% of this sub - sample of csss ( or @xmath338% of the total number of csss ) which is a reasonably - significant number . if we consider instead the complete sample of csss defined by fanti et al .
( 1995 ) which contains 43 objects then five of them , i.e. @xmath3311% , are listed in table 3 .
this confirms that a strongly - bent jet is a noticeable phenomenon among this class of object . as a comparison sample
, we have taken the compilation of 293 flat - spectrum radio sources constructed by taylor et al .
( 1996 ) which gives a complete flux density limited vlbi sample .
following wilkinson ( 1995 ) , 82% of objects in this sample are classified as asymmetric core - jet sources .
we have found only 4 sources in this sample which exhibit a bent jet fitting the selection criteria , and on the same scales , that we have used for the csss listed in table 3 . this number represents less than 2% of the objects in the complete sample and
is much less than that found for the css complete sample .
this last number should be treated with some care , however .
the observations of the flat spectrum sample were made with a high resolution vlbi array , limited in sensitivity by the snap - shot observing strategy adopted .
deeper observations might change the number of objects with large bent jets .
however , we consider the large fraction of objects with a strongly - bent jet among csss as an indication that the propagation of the jet is strongly affected by the ambient medium .
let us calculate the probability of a collision between the jet and the dense gas clouds in the nlr surrounding the radio source .
the outer gas is characterized by temperature of @xmath55 , a particle density of @xmath56@xmath57 and by a filling factor @xmath58 ( mccarthy 1993 ) .
the number of clouds , b , that the jet will encounter in a distance _
l _ is given by : @xmath59 where _ n _ the total number of clouds , _
l _ is the size of a narrow line region , _
d _ the diameter of a cloud , and _
n _ is the cloud spatial density @xmath60 . the diameter _ d _ and the density _
n _ of these dense gas clouds in active galaxies with @xmath61 is still uncertain .
the term b can be expressed more appropriately in terms of the filling factor . the radio jet propagates in a single preferred direction
; thus we are interested in determining the linear filling factor @xmath62 , which is defined as : @xmath63 the term on the right is , by definition , the _ volume filling factor _ @xmath64 . such an expression is valid if the jet diameter is @xmath65 , the cloud s diameter . from this , one can derive @xmath66 and @xmath67 can be calculated as : @xmath68 to estimate b , we have adopted the following values : ( a ) the linear size _
l _ of the nlrs can be taken to be in the range 110 kpc .
this range is suggested by the bends which occur inside the first kpc from the core and from the fact that we are dealing with csss whose linear size can not exceed 1520 kpc by definition ; ( b ) the filling factor @xmath69 is @xmath70 .
however , from the luminosity in strong lines like h@xmath71 in the nlr ( @xmath72 ergss@xmath5 in seyfert 2s ) and electron density of 10@xmath73 @xmath57 , the filling factor can reach values as large as 10@xmath23 ( see , for example , peterson 1997 ) ; ( c ) in nearby agn , the nlrs are often partially resolved by the observations and the inferred sizes are generally @xmath74100 pc .
this gives an upper limit to the diameter @xmath75 for a cloud . in our own galaxy ,
dense clouds are characterized by sizes of 110 pc ( cowie & sangaila 1986 ; cox & reynolds 1987 ) .
thus the cloud diameters can be taken to be in the range 10100 pc . moreover , from the vlbi observations we see that the jets are not resolved in a direction transverse to the major axis , with an angular resolution of @xmath76mas ( about 10 pc for @xmath77 ) . from these considerations , and assuming that the jet diameter is @xmath65 , then b will fall in the range 0.10.01 .
this last value is not just the number of collisions between the radio jet and the dense gas cloud in a css .
it also represents the fraction of sources which suffer for such an interaction . here
we wish to stress that the chances of such a collision are not very small . the dallacasa & stanghellini ( 1990 ) list of csss contains more than one hundred objects .
the percentage of csss which show large bends in the propagation direction of the radio jet is a number which is roughly in agreement with the value derived here .
the above considerations also indicate that the the probability of multiple bends is quite small .
sub - arcsecond resolution vlbi maps of csss often show that these sources have strongly - distorted structures with recognizable jet - like features ( fanti et al . 1986 ; spencer et al .
1991 ) , consistent with strong dynamical interactions between the jets and the ambient media .
although the largest distortions are often seen in sources dominated by jets , and there are suggestions that this might at least partly be due to projection effects ( e.g. spencer 1994 ) , there are clear indications that intrinsic distortions due to interactions with a possibly - dense inhomogeneous gaseous environment play an important role . for example , some of the most distorted and complex structures are found in objects with very weak cores , as in the quasars 3c43 and 3c119 .
such objects should be inclined at large enough angles to the line of sight that projection effects are not very significant .
generally speaking , csss usually show very weak cores compared to the jet emission while in superluminal sources the doppler boosted cores are the components with the more dominant emission . moreover , the only csss known so far which exhibit superluminal separation between two components , thus implying a small angle of inclination to the line of sight , are 3c147 ( @xmath78 ; alef et al . 1990 ) and 3c138 ( @xmath79 ; cotton et al .
in addition to the distorted structures , an increased asymmetry in the location of the outer components also suggests that the components are evolving through a dense asymmetric environment in the central regions of these galaxies ( sanghera et al .
1995 ; saikia 1995 ) .
such an asymmetry between the mas structure and the arcsec structure is present where an arcsec structure is detected for sources in table 3 .
most csss show low ( @xmath33 1@xmath80 ) percentage polarizations at or below 5 ghz ( saikia 1991 ; saikia , singal & cornwell 1987 ) .
the median polarization , however , increases with frequency , ( van breugel et al .
1992 ; saikia 1995 ) suggesting that the polarimetric behaviour is due to faraday depolarization .
this pattern is consistent with the observation of garrington and akujor ( 1996 ) that smaller sources show stronger depolarization .
a number of sources with very high rotation measures in the range of several thousands of radm@xmath23 are also csss .
most of the sources in table 3 follow this trend having large depolarization and do have rather large rms .
sub - arcsec polarimetry has often provided evidence in favour of the interaction of these components with dense clouds of gas .
for example , the southern component of 3c147 has a much higher rm than the northern component , is brighter , closer to the nucleus , and has the expected signatures of a jet colliding with a cloud of gas on the southern side of the galaxy ( junor et al .
1997 ) . to explain both the low polarizations at centimeter wavelengths and the small linear sizes
, it has been proposed that csss are cocooned in dense gaseous envelopes ( e.g. mantovani et al .
1994 ) .
existing ground - based optical observations show that narrow - line luminosities are comparable for csss and large - sized radio sources . however ,
broader narrow - line profiles found for csss suggest that the interaction of the jet with the ambient medium produces additional accelerations ( gelderman 1992 ; morganti 1993 ; gelderman & whittle 1994 ) . from the information available in the literature , it is not clear whether there is enough gas to support the so - called frustration scenario for csss , in which they are confined to small dimensions by dense interstellar media , thus hindering their evolution into classical frii radio sources .
the structures of the sources shown here , with such large bent jets in the vicinity of the core , support the view of a strong interaction between the jet flow and dense gas clouds .
this is in agreement with our estimate of the probability of having an interaction between the jet and a dense cloud of gas inside the first kpc from the nucleus .
we predict that the widths of the narrow emission lines in these sources will be broader that for nlrgs ( narrow line radio galaxies ) .
the interaction between the radio source and the ism has been investigated , with time - dependent numerical simulations of jets associated with csss , by de young ( 1993 ) in the following scenarios : _ a ) _ the ism is an uniform homogeneous medium , with a smooth density gradient ; _ b ) _ the ism is in two phases a hot tenuous gas together with a population of cold dense clouds ; _ c ) _ the interaction is with a single large feature such a very dense cloud . de young shows that csss of low and intermediate luminosity can be confined by an ism of average density 10 times that of the galaxy .
confinement of the highest luminosity sources requires an ism of unrealistic density and total mass .
the population as a whole can be confined if the jets are intrinsically more efficient , by factors of 10 or more , than is commonly assumed for ` normal ' radio sources .
the physics of jet - cloud collisions have been investigated by de young ( 1991 ) and norman & balsara ( 1993 ) in 3-d hydrodynamical simulations .
norman & balsara do identify a fluid dynamical mechanism which maintains the collimation of the jet for all deflection angles @xmath81 .
the reflected jet inherits the stability properties of the original jet .
de young shows that deflection of a jet in the case of jet - loud encounters over the short times ( @xmath82 years ) and distances ( 2 to 4 kpc ) needed for csss may easily occur .
fanti et al . ( 1995 ) have distinguished between double - lobed csss , which represent @xmath83 of the population and complex morphologied or asymmetric jet - dominated csss .
they suggest that double - lobed csss represent the precursors of the large , double - lobed radio sources .
the remaining 30% are sources with the jets brightened by doppler boosting or by interaction with the ambient medium .
the sources in table 3 are classified as core - jet objects .
the evidence that doppler boosting can play a role in magnifying the large distortions seen in the jets is very weak .
radio structures , large rms , and broader narrow line profiles support the view that large bent jets can be caused by interactions between the jets and dense gas clouds in the nlrs .
even a rough statistical computation indicates that the probability of a bump in the nlrs is statistically significant .
3-d hydrodynamical simulations of extragalactic jets reveal interesting features that further support this interpretation . in this picture , sources like 3c119 and 3c287 require more than one collision to createstructures that are spiral - like .
such cases can not be excluded statistically , though they are very unlikely .
alef w. & preuss e. 1990 proceedings _ css & gps radio sources _ , dwingeloo , eds . c. fanti , r.fanti , c.p . odea and r.t .
schilizzi , p. 149
barthel p. 1989
_ apj _ * 336 * , 606 van bruegel , w.j.m . ,
miley , g & heckman t. 1984 _ a.j . _ * 89 * , 5 van bruegel , w.j.m . , fanti c. fanti r. _ et al .
_ 1992 _ a&a _ * 256*,56 cawthorne t. , wardle j.f.c . , roberts d.h .
_ et al . _
1993 _ apj _ * 416 * , 519 conway j.e . &
. 1993 _ ap.j .
_ , * 411 * , 89 cotton w.d . 1983 _ ap.j . _ ,
* 271 * , 51 cotton w.d . , owen f.n .
& geldzahler b.j .
1984 _ ap.j . _ * 277*,l41 cotton w.d . ,
owen f.n . ,
mahoney m.j .
1989 _ ap.j .
_ , * 338 * , 37 cotton w.d . , dallacasa d. , fanti c. et al .
1997 _ a&a _ in press cowie l. & songaila a. 1896 _ ara&a _ , * 24 * , 499 cox d.p . & reynolds r.j .
1987 _ ara&a _ , * 25 * , 303 dallacasa d. & stanghellini c. 1990 _ css & gps radio sources _ , dwingeloo , eds . c. fanti , r.fanti , c.p .
odea & r.t .
schilizzi , p. 224
dallacasa d. , fanti c. , fanti r. _ et al .
_ 1995 _ a&a _ * 295 * , 27 dallacasa d. , bondi m. , alef w. & mantovani f. 1997 _ a&as _ in press de young , d.s .
1991 _ apj _ * 371 * , 69 de young , d.s .
1993 _ apj _ * 402 * , 95 fanti , c. fanti r. schilizzi , r. t. _ et al .
_ 1986 _ a&a _ * 170 * , 10 fanti r. , fanti c. schilizzi , r. t. _ et al .
_ 1990 , _ a&a _ * 231 * , 333 fanti , c. , fanti r. , dallacasa d. _ et al .
_ 1995 _ a&a _ * 302 * , 317 fejes , i. , porcas r.w .
& akujor c.e .
1992 _ a&a _ * 257 * , 459 garrington s. & conway r. g. ( 1991 ) _ mnras _ * 250 * , 198 garrington s. & akujor c. e. 1996 _ extragalactic radio sources _ , bologna , eds .
r. ekers , c. fanti & l. padrielli , p. 77
gelderman r. 1992 _ astrophysical jets
d. bulgarella , m. livio , & c.p .
odea ( cambridge university press ) , p. 16
gelderman r. & whittle m. 1994 _ apjs _ * 91 * , 491 gopal - krishna & wiita , p.j .
1991 _ aj _ * 402 * , 95 kato t. , tabara h. , inoue m. & aizu k. 1987 _ nature _ * 329 * , 223 kus , a. 1994 _ compact extragalactic radio sources _ , socorro , new mexico , eds .
j. a. zensus & k. i. kellermann , p. 91
junor , w. , salter , c.j .
, saikia , d.j . , _ et al .
_ 1997 _ nature _ , submitted laing r.a .
, riley j.m . , longair m.s .
1983 , _ mnras _ * 204*,151 mantovani f. , saikia d.j . , browne i.w .
_ 1990 _ mnras _ * 245*,427 mantovani f. , junor w. , fanti r. _ et al .
_ 1992 _ mnras _ * 257 * , 353 .
mantovani , f. , junor , w. , fanti , r. _ et al .
_ 1994 _ a&a _ * 292 * , 59 mantovani , f. , junor , w. , fanti , r. _ et al .
_ 1997 _ a&as _ , * 125 * , 573 marscher , a.p . & shaffer , d.b .
1980 _ a.j .
_ , * 85 * , 668 mccarthy , p.j
. 1993 it a.r.a&a , * 639 * morganti r. 1993 _ sub - arcsecond radio astronomy _ eds . davis r. j. & booth r. s. , ( cambridge university press ) nan ren - dong , schilizzi r.t . , fanti , c. _ et al .
_ 1991 , _ a&a _ * 245 * , 449 norman m.l . & balsara d.s .
1993 _ jet in extragalactic radio sources _ eds . k. meisenheimer & h.j .
roeser , springer verlag , berlin , p. 229
odea c.p .
, baum s.a . &
morris g.b . 1990 _ a&as _ ,
* 82 * , 261 pearson t. , perley r.a . & readhead a.c.s .
1985 , _ aj _ * 90 * , 738 peacock j.a . &
1982 _ mnras _ * 198 * , 843 peterson b.m . 1997
_ an introduction to active galactic nuclei _ , cambridge university press , p. 101
saikia , d. j. 1990 _ css & gps radio sources _ , dwingeloo , eds . c. fanti , r.fanti , c.p .
odea & r.t .
schilizzi , p. saikia , d.j . ,
. national academy of sciences , * vol .
92 * , p. 11417
saikia , d.j . ,
singal , a.k . ,
cornwell , t.j .
1987 _ mnras _ * 224 * , 379 sanghera h.s .
, saikia d.j .
, ldke e. _ et al .
_ 1995 _ a&a _ * 295 * , 629 shepherd m.c .
, pearson t.j . &
taylor g.b . , 1995 _ baas _ * 26 * , 987 spencer r.s . ,
schilizzi r.t .
, fanti c. _ et al .
_ 1991 _ mnras _ * 250 * , 225 spencer r.s .
1994 _ compact extragalactic radio sources _ , socorro , new mexico , eds .
j. a. zensus & k. i. kellermann , p. 35
spinrad h. , djorgovski s. , marr j. & anguilar l. 1985 _ publs astr.soc.pacif._ , * 97*,932 stanghellini c. , odea c.p .
, baum s.a .
1993 _ apjs _ , * 88 * , 1 tabara h. & inoue m. 1980 _ a&a _ * 39 * , 379 taylor g.b . ,
inoue m. & tabara h. 1992 _ a&a _ * 264 * , 421 taylor g.b . , vermeulen r.c . , readhead a.c.s .
_ 1996 _ ajsuppl .
_ , * 107 * , 37 udomprasert p.s . ,
taylor g.b .
, pearson t.j . & roberts d.h . 1997 , _ apjletters _ , in press venturi t. , pearson t.j . ,
barthel p.d . & herbing t. 1993 _ a&a _ * 271 * , 65 wilkinson , p.n . , akujor , c.e . ,
cornwell , t.j . ,
saikia , d.j . , 1991 _ mnras _ * 248 * , 86 wilkinson , p.n . 1995 , _ quasars ans agn : high resolution radio imaging _ , national academy of science colloquium , irvine , eds . m.h.cohen & k.i .
kellermann , p. 11342 | the class of compact steep - spectrum sources is dominated by double - lobed objects ( 70% ) .
the remaining 30% are jet - dominated objects , with the jet brightened either by doppler boosting or by interaction with the ambient media .
we show that there is both observational and statistical evidence in favour of an interaction between jets and dense gas clouds .
such an interaction should happen in the narrow line regions .
the images of four csss observed by us with vlbi are also presented
. these sources do show large bent jets in the first kpc from the nucleus . |
in this article , we describe the recent development of the 2mrad horizontal crossing angle scheme for the ilc . to date , two parallel designs have emerged : the first coming from the uk and france and the second from slac and bnl .
we shall describe both schemes here , although they share many common features and are now being developed in parallel under the unified slac - bnl - uk - france task force collaboration .
the work presented focuses on the performance at 1 tev .
the benefits of the scheme are well documented @xcite : for small crossing angles , the loss of luminosity is small ( crab correction may not be necessary and it may be possible to partially correct this loss by exploiting the finite @xmath0 at the ip for local chromaticity correction lattices ) , no electrostatic separators or kicker magnets are needed and the conditions are improved for physics ( e.g. better forward coverage ) . a known weakness of this scheme is however its more difficult conditions for extracting cleanly the spent disrupted and energy - degraded beam , in comparison with larger crossing - angle schemes where separate magnetic channels can be used for in - and outgoing beams .
the work presented here covers the designs developed at slac , bnl , the uk and france . in section [ euroscheme ]
we shall describe the scheme developed in europe and discuss its performance at 1 tev . in section [ slacscheme ] , we shall discuss the performance of the slac / bnl scheme , when extended to 1 tev from the initial design at 500 gev , and we shall draw our conclusions in section [ conc ] .
in this section we shall describe the 2mrad interaction region layout and extraction line for the 2mrad horizontal crossing angle scheme .
the final doublet magnets have been optimised for the extraction of a 500 gev beam . a similar optimisation exists for the baseline beam energy of 250 gev .
it has been shown that the doublet parameters calculated for a high energy beam also provide acceptable extraction at lower energy .
the superconducting magnet closest to the ip , denoted qd , is of critical importance to the ir layout properties and is chosen to be a lhc low-@xmath1-insertion quadrupole .
this provides the required aperture and field strength to accommodate both the incoming and outgoing ( disrupted ) beams .
note that the outgoing beam possesses a long low energy tail and , by virtue of the crossing angle , is off - axis in qd .
the other final doublet magnet , qf , is constructed from a normal conducting magnet and is separated from qd by 3 m . for the tev machine , qd and
qf are 2.3 m and 1.9 m long , respectively . after passing through qd
, the outgoing disrupted beam enters the extraction line , which provides beam transport to the dump and downstream diagnostics ( the geometry is fixed by the linear matrix element @xmath2 from the ip to the exit of qd ) .
the lhc low-@xmath1-region quadrupoles are constructed from nbti and can achieve a gradient of 215 tm@xmath3 with an available aperture for the beam of 62 mm .
note that higher gradients are currently under development , which will aid the present application .
lhc studies of the tolerable power depostion indicate local and integral values of 0.4 mwg@xmath3 and 5 wm@xmath3 respectively ; this approximately translates into a maximum power deposition from charged particles into qd of around 10w .
note that in all these studies , unless otherwise noted , we assume the parameters of the tev ilc parameters working group and , where possible , assume the worst possible parameter set for extraction . in this work ,
we follow @xcite and assume a photon cone half - opening angle of 0.5mrad in all cases .
this ensures that the cone contains all the photon power , apart from 100w .
this remaining power needs to be lost in suitable collimators . ensuring extraction of the photons past qf immediately requires a crossing angle of greater than 1.6mrad .
figure [ figqdel ] shows the power deposition into qd from charged particles produced during the beam - beam interaction , as a function of crossing angle .
these power deposition calculations are described in detail in @xcite .
the charged particle loss comes from two sources : the low energy tail of the disrupted beam and from radiative bhabha events produced in the interaction of colliding particles ( also refered to as the `` compton tail '' ) .
the latter contribution is suppressed in regions of phase space of low transverse momentum exchange , where the virtual photons participating in the scattering process can have a transverse position indeterminacy exceeding the transverse beam size . the suppression from this so - called beam - size effect
is illustrated in the curves of fig .
1 . conservatively , the results without it are however used for the worst - case scenarios considered here . if we assume a maximun power loss of 10w , we find that for the case of a tev machine with a bunch population of 2@xmath410@xmath5 , we can tolerate a crossing angle no larger than 1.6mrad .
this result is dominated by the compton tail .
the other case we have considered in figure [ figqdel ] shows a larger permitted crossing angle , and hence easier extraction .
this case , with a vertical offset of 150 nm at the ip , is studied because it maximises the low energy tail of the disrupted beam . for further details of these calculations
see @xcite .
all of these curves were produced using the us cold machine parameters , for which the key parameters ( for this study ) are similar to the wg 1 new ilc nominal parameters @xcite .
10@xmath5 bunch population and the right - hand plot is a tev machine with the vertical offset of 150 nm at the ip . in both cases
the blue line ( the lower curve at small angle ) denotes the power loss with the beam size effect turned on and the red line ( the upper curve at small angle ) denotes the power loss with the beam size effect turned off .
note that for the right - hand plot , both curves have the vertical offset.,width=226 ] 10@xmath5 bunch population and the right - hand plot is a tev machine with the vertical offset of 150 nm at the ip . in both cases
the blue line ( the lower curve at small angle ) denotes the power loss with the beam size effect turned on and the red line ( the upper curve at small angle ) denotes the power loss with the beam size effect turned off .
note that for the right - hand plot , both curves have the vertical offset.,width=226 ] the purpose of the extraction line is to transport the disrupted beam to the dump with controlled and acceptable losses , handle the undisrupted beam in the case of single beam operation ( e.g. during commissioning ) and to provide diagnostics on the disrupted beam .
the possible diagnostics include compton polarimetry and energy spectrometry .
the linear dispersion for the extraction line is shown in the left - hand plot of figure [ figplots ] .
the linear dispersion allows identification of the major optical structures , denoted by regions of non - zero dispersion .
the first region is the dispersion generated by the beam being off - axis in qd ( the first magnet in the line ) .
this dispersion is cancelled by a mirroring dipole .
the second dispersive region is a `` bend - back '' linear - achromatic structure , which bends the beam parallel to the beam at the ip .
this is necessary to perform compton polarimeter and `` undo '' the spin precession occuring in upstream dipoles .
the third region is the diagnostic chicane ; the current requirements for this chicane is a secondary beam focus and some dispersion , such that the dispersive beam size dominates the betatron beam size
. requirements of beamsize / laser wavelength correlation and of minimum background requirements are under study . ) to achieve 0.5ppm beam loss at each element , for the case of the tev machine nominal parameters ( with ( dashed line ) and without ( solid line ) a vertical offset).,width=226 ] the first magnet in the extraction line is of critical importance , due to the need to place the first disrupted beam magnet as close as possible to the ip , balanced against the constraints of transverse space . for this reason ,
a half - quadrupole current sheet magnet is used @xcite .
these magnets , developed for the tesla extraction line @xcite provide the required aperture and maximum pole tip field whilst allowing the magnet to be placed 8 m after the exit of qd ( with 50 mm transverse separation ) . note that other magnet designs are under study .
the right - hand plot of figure [ figplots ] shows the optimisation of the apertures for a fixed power loss .
we have used the tev machine nominal parameters and demanded a power loss of 0.5ppm at each element .
we also considered a vertical offset of 120 nm . the apertures were calculated using explicit all - orders ray - tracing of the disrupted beam .
the calculated apertures are acceptable , although the very downstream apertures for the tev machine need some optimisation .
the study of this scheme is on - going .
it is necessary to include the effects on the outgoing beam of all the fields from nearby magnets used to focus the incoming beam , in particular qf
. this can be done by way of suitable multipole expansions , once the detailed field map is computed .
it has been shown to be particularly critical to predict reliably the transport of the low - energy tail of the extracted beam @xcite .
furthermore the magnet technology is of prime importance , and substitutes for the half - quadrupole described here are under consideration , along with the use of sextupoles in the extraction line to aid the control of the low energy beam tail .
the layout of the charged particle and photon beam dumps is also a topic for further work , along with detailed calculations of backgrounds in the detector and diagnostic sections .
in this section , we describe the performance of the slac / bnl - designed 2 mrad scheme , when extended to the upgrade centre - of - mass energy of 1 tev .
this scheme , described in @xcite , was studied at 1 tev using the nominal parameters and a vertical offset of 100 nm .
this gives a low energy beamstrahlung tail reaching down to 18% of maximum energy , when computed with high statistics . extending the design to 1 tev
required a number of modifications to the initial design for 500 gev : proper scaling of the lengths of sextupoles to keep the pole tips within maximum achievable fields , increasing l * from 3.51 m to 4.1 m and reducing the length of qd0 to reduce over focusing of low energy tail particles . in order to reduce the losses of the low energy tail particles on the magnets ,
a collimator was introduced in the vertical plane within the final doublet .
the horizontal and vertical beam envelopes for this scheme are shown in figure [ figslac ] , for particles in different energy ranges .
note that this is a first look at the tev upgrade of this scheme and further optimisation and power losses on the magnets needs to be evaluated .
we have described the 2mrad horizontal crossing angle schemes under development for the ilc .
the first part described the scheme developed in the uk and france , and focused on its propeties at 1 tev .
the second describes the study of the slac and bnl scheme from these proceedings , when extending the design to 1 tev .
both schemes at 1 tev can work with nominal parameters but the high luminosity case seems difficult .
further development work is ongoing .
finally we note that both schemes share many common features . | the current status of the 2mrad crossing angle layout for the ilc is reviewed .
the scheme developed in the uk and france is described and the performance discussed for a tev machine . secondly , the scheme developed at slac and bnl is then studied and modified for a tev machine .
we find that both schemes can handle the higher energy beam with modifications , and share many common features . |
in view of the actual experimental possibilities , we recall some general characteristics of three - body charged systems with arbitrary masses and two different charges , in order to derive the charge - mass dependence of the leading order strength of the attractive dipole interaction produced by a bound two - body subsystem ( with individual charges @xmath5 and @xmath6 ) in the third particle with charge @xmath0 .
we start our investigation by considering the old and well - known case of the interaction of a positronium ( @xmath7 , a bound - state of an electron @xmath8 and a positron @xmath9 , where @xmath10 is the absolute value of the electron charge ) with a spectator electron .
this is the ionized negative positronium ( @xmath11 ) .
early calculations , by wheeler in 1946 @xcite , have already predicted a bound state for such system , confirmed by mills in 1981 @xcite .
further investigations on the properties of @xmath12 system @xcite , as well as on other coulombic three - body systems , since 1960 s up to recent years @xcite , have also been motivated by the increasing interest in matter - antimatter interactions @xcite .
up to 1995 , the theoretical and experimental advances in understanding coulombic three - body systems and matter - antimatter interaction can be found , respectively , in two reviews : @xcite and @xcite .
as emphasized in @xcite , small number of leptons , electrons , muons and their antiparticles , are important to test fundamental theories of quantum electrodynamics ; and systems with small number of protons and antiprotons can also provide relevant tests of the strong interaction .
the actual interest on the properties of few - body charged systems is evidenced by the recent report on the production of a molecular di - positronium @xmath13 @xcite .
for a recent review , particularly concerned on the stability of quantum charged few - body systems , see @xcite . in ref .
@xcite , rost and wintgen have explored and classified the @xmath11 dynamics considering the observation that it has a molecular structure similar to the ionized hydrogen molecule @xmath14 [ @xmath15 .
they have also reported the existence of a @xmath16 resonance pattern unknown in three - body coulomb systems .
such results , combined with results obtained for the hydrogen ion @xmath17 [ @xmath18 , lead them to suggest the existence of a similar resonance spectrum for all @xmath19 coulomb systems with charge and mass ratios such that @xmath20 and @xmath21 , respectively .
more recently , by considering a molecular adiabatic ( ma ) treatment for the @xmath11 , it was also reported an accumulation of three - body resonances _ above _ the two - body threshold @xcite .
surprisingly , besides the number of recent studies concerned with few - charged quantum systems @xcite , and the recognized relevance of a mass - charge universal relation in view of actual experimental facilities @xcite , a particular straightforward condition relating charges and masses for a general three - body system ( with arbitrary masses ) is still missing in the literature . in the case of exotic molecular three - body systems , where the subsystem is hydrogen - like ( with @xmath22 replaced by @xmath23 or @xmath24 ) , it was shown in @xcite that a relation for the structure of the spectrum can be obtained in the frame of born - oppenheimer approximation , following spectral properties of long - ranged @xmath25 interactions @xcite , which are known to be conformally invariant @xcite .
the study of resonance patterns , which can occur in few - body interactions , became very relevant in trapped ultracold atom experiments , as the presence of several resonances at experimentally accessible magnetic fields can allow precise tunning of atom - ion interaction @xcite .
theoretical predictions such as the increasing number of three - body bound states when the two - body scattering length goes to infinity , known as efimov effect @xcite , can actually be checked experimentally in ultracold atomic laboratories @xcite .
the spectrum of efimov states , exhibiting a geometrical scaling , is generated by an attractive effective potential proportional to @xmath25 , where @xmath26 is the distance of one of the particles to the center - of - mass of the remaining pair , considering short - range two - body interactions @xcite ( on the scaling mechanism and conformal invariance behind this effect , see @xcite ) .
as it will be shown , the long - ranged coulombic interactions , for certain configurations of three charged particles , can exhibit the same kind of effective interaction . for recent relevant applications in ultracold laboratories , of a study with three charged systems
, we can mention the possibility of exciton @xcite or positronium @xcite condensed gas interacting with a charged particle . in case of excitons ( electron - hole bound pair in a semiconductor ) , the electron and hole in the interacting pair can acquire effective individual masses distinct from the free electron mass @xcite . when interacting with a slow charged particle , a charge - mass dependent resonance pattern should emerge .
so , well - based simple charge - mass conditions can be very helpful to analyze the relation between effective masses and the observed spectrum .
motivated by the above discussion we show examples in atomic physics of charged three - body systems in which appears an effective long - range @xmath25 potential , where the strength is modulated by the arbitrary individual masses and charges , restricted to the condition that the bound subsystem is neutral .
a robust mass - charge critical condition is derived for the weakly - bound three - body spectrum _
below _ the state @xmath27 of the subsystem , in case of arbitrary masses . here
, we should observe that the degeneracy between the @xmath28 and @xmath29 levels of the chargeless two - body subsystem is broken by the dipole potential .
when it is attractive it rises to the spectrum _ below_@xcite that we consider ; the repulsive part is responsible for a set of resonances _ above _ the @xmath30 state @xcite .
the approach is strictly valid for the cases where the dimer is made up with _ point - like _ charged particles .
however , it can also be taken as a first approach when the charged particles are more complex objects .
the basic equations of our formalism , in the adiabatic approximation , are given in the next section , which lead to the effective @xmath25 long - range interaction and a spectrum geometrically scaled . in section iii , after analyzing the range of values for the strength of the effective interaction in terms of the particle masses , we illustrate the main results with examples . a general discussion on the applicability of the present approach and a summary of our conclusions are provided in section iv .
in this section , we present the basic formalism for a general charged three - body system , where the masses ( @xmath31 ) can be arbitrary , and the charges are such that we have a bound chargeless two - body subsystem ( @xmath6 ) interacting with a slow charged particle @xmath0 ( the charges are in units of the absolute value of the electron charge @xmath32 ) .
the system hamiltonian is given by @xmath33 where @xmath34 is the distance vector between the charges @xmath5 and @xmath35 and @xmath36 the distance vector from the center - of - mass of the subsystem to the third ( spectator ) particle .
here , we define @xmath37 as the reduced mass of the subsystem @xmath38 , @xmath39 ; and @xmath40 the corresponding reduced mass of the subsystem @xmath41 interacting with the third particle : @xmath42^{-1}$ ] .
the adiabatic approximation implies in solving a coupled equation for the total wave - function @xmath43 , such that we first solve the schrdinger equation in the variable @xmath44 for the wave - function @xmath45 , using @xmath46 as a parameter , with energy solution @xmath47 : @xmath48 next , the total energy @xmath49 is given by @xmath50 the separation of the schrdinger equation by means of the adiabatic approximation is justified in case the energy of the relative motion of the center - of - mass of the two - body subsystem , with respect to the third particle , is small in comparison with the subsystem binding energy . here , the main interest is the behavior of @xmath51 as @xmath26 goes to infinity .
the case where the third particle interacts with the @xmath52subsystem .
using the expansion for the coulomb potential , with @xmath53 and @xmath54 the usual legendre polynomials of order @xmath55 , we have @xmath56\psi(\vec{r},\vec{r } ) .\label{eq05}\end{aligned}\ ] ] assuming a neutral subsystem ( @xmath57 ) , the leading order term ( conformally invariant ) in the asymptotic expansion @xmath58 is given by @xmath59\psi(\vec{r},\vec{r } ) , \label{eq06}\ ] ] where @xmath60 is the angle between the vectors @xmath34 and @xmath36 .
the next - to - leading order contribution to the adiabatic potential is @xmath61 if @xmath62 , otherwise a term like @xmath63 appears , which however does not contribute due to parity conservation .
note that the higher order terms in eq.([eq05 ] ) break the conformal invariance .
the interaction between the third particle and the subsystem @xmath38 at large distances is dominated by a dipole potential , which breaks the degenerated character of the opposite parity states . according to ref .
@xcite , there is an accumulation of resonances below the threshold of the @xmath27 excited state of the positronium negative ion , due to an attractive dipole interaction , which appears as the leading asymptotic term of the potential at large hyperspherical radius .
therefore , it is reasonable to assume that for large values of @xmath26 the first low lying levels of the @xmath52subsystem ,
i.e. , the first excited @xmath64 and @xmath65 states are weakly perturbed by the dipole interaction and the degeneracy is broken . for @xmath58 ,
the distorted wave function of the @xmath52subsystem is a linear combination of atomic orbitals ( lcao ) , composed by @xmath66 , where @xmath67 ( with @xmath68 ) are the usual spherical harmonic function and @xmath69 ( with @xmath70 ) are the two - body hydrogenic like wave functions for the charges @xmath5 and @xmath71 . inserting @xmath45 in eq .
( [ eq06 ] ) , we obtain @xmath72 where @xmath73 is the energy of the first excited state of the subsystem @xmath38 .
the two possible signs in ( [ eq07 ] ) come from the diagonalization of the eigenvalue equation ( [ eq06 ] ) in the 2s-2p subspace . from eqs .
( [ eq07 ] ) , ( [ eq08 ] ) and ( [ eq04 ] ) , we obtain the corresponding equation for the third particle motion related to the center - of - mass of the subsystem : @xmath74 here we note how the degeneracy between the @xmath28 and @xmath29 levels of the chargeless two - body subsystem is broken by the dipole potential .
the motion of the wave function @xmath75 is given by an attractive type potential .
the effect discussed in ref .
@xcite of level condensation near the @xmath76 state of positronium is analogous to this situation , as the long - range attractive potential @xmath25 is responsible for forming such resonances .
the other solution to @xmath77 gives a long - range repulsion which means that , asymptotically , no bound states can be formed .
this repulsive @xmath25 interaction is responsible for the set of resonances above the @xmath30 states @xcite .
focusing our approach in the weakly - bound states of eq .
( [ eq09 ] ) , we end up with an attractive dipole interaction , @xmath78 , generated by the particles with masses @xmath79 and @xmath80 , where the dimensionless @xmath81 is given by @xmath82 now , to solve eq .
( [ eq09 ] ) , we employ a well - known result derived for attractive dipole potential ( see @xcite ) , which implies in a spectrum of infinite weakly - bound levels @xmath83 below @xmath84 , if @xmath85 ; otherwise , no other bound state .
so , the following condition emerges for the existence of such spectrum : @xmath86 in order to examine the condition suggested in @xcite for the occurrence of level condensation , we restrict the above to an @xmath19 system with @xmath87 : if @xmath88 and @xmath89 , we can easily verify that in all the cases we will have @xmath85 , implying in a level spectrum below the state @xmath27 of the subsystem @xmath90 .
the mass ratio @xmath91 in this case can only control the level spacing of the spectrum .
however , if the bound subsystem is of a particle and antiparticle ( @xmath92 ) , with the captured particle having mass @xmath93 , the condition @xmath94 will give us @xmath95 .
this can be exemplified with the possible configurations of three - body systems with protons , electrons and their antiparticles , such as @xmath96 , @xmath97 and @xmath98 , where level spectrum is expected to occur only for @xmath96 and @xmath97 .
equation ( [ eq11 ] ) gives us a quite general relation which allows infinite number of energy levels that scale geometrically in most of three - body systems involving the usual stable charged particles and their anti - particles .
the number of states are very dense when @xmath99 , as the ratio between the levels is shown to be given by @xcite @xmath100 implying in the geometrically scaled energy levels @xmath101 where @xmath102 is a reference energy , solution of eq .
( [ eq09 ] ) , determined by non - adiabatic effects .
observe that eq .
( [ eq09 ] ) is not well defined for small values of @xmath26 ( of about few bohr radii ) , leading to the collapse of the reference ground - state energy @xcite .
therefore , the dipole potential must be modified at some short distance , which in our case corresponds to a region where non - adiabatic effects start to be relevant .
two other important remarks related to eqs .
( [ eq10])-([eq11 ] ) : ( i ) the strength of the dipole interaction , as well as the corresponding level ratio @xmath103 ( when the occurrence of infinite number of levels is possible ) , depend only on the ratio of reduced masses @xmath104 and ratio of the charges @xmath105 , implying that systems such as @xmath106 , @xmath107 , or @xmath108 have the same @xmath81 and energy ratios @xmath103 .
( ii ) another relevant characteristic of ( [ eq10 ] ) is its dependence on the specific three - body configuration and identification of the spectator particle : let us consider two possible configurations for the same three - body system , where @xmath109 and @xmath110 and @xmath111 . with @xmath112 for the configuration @xmath113
, we obtain @xmath114 an obvious example is that of @xmath115 or @xmath116 , where @xmath117 should lead to a dense level spectrum below the @xmath27 state of the subsystem . for the counterpart configuration , @xmath118 , @xmath119
, a similar spectrum is not expected to occur .
figure [ lambda ] illustrates our conclusions for the values of @xmath81 as a function of the mass ratios @xmath120 and @xmath121 , taking @xmath122 and @xmath123 .
the critical value of @xmath124 is shown by the dashed straight line . as observed , only in the cases that @xmath125 it is possible to reach the value @xmath124 for some ratios of @xmath120 .
there is no infinite number of levels in case that @xmath126 , independently of @xmath120 . when @xmath127 , the only solution is @xmath128 ; and , when @xmath129 , @xmath130 for all choices of @xmath80 and @xmath79 .
a few examples of three - body charged systems , with the corresponding values of @xmath81 and level ratios @xmath131 are given in table i. as verified , level condensation below the @xmath27 state of the subsystem ( shown in the first two lines of the table ) is not possible when the mass of the captured particle is much smaller than the lighter particle of the subsystem . in this table
we also present a few results motivated by the actual interest in matter - antimatter interaction , and also motivated by experiments with exotic mesonic atoms [ where protons(@xmath132 ) , deuterons(@xmath133 ) or tritons(@xmath134 ) are combined with muons(@xmath135 ) , kaons(@xmath136 ) or pions(@xmath137 ) ] .
our @xmath138 and @xmath139 results correspond to the ones given in @xcite .
obviously , the same results apply , respectively , for @xmath140 and @xmath141 .
table i displays the striking difference of such results when compared with the ones for @xmath142 and @xmath143 . related to @xmath11
we should note that our approach gives @xmath144 , to be compared with @xmath145 obtained by botero @xcite in a more involved calculation .
@lcc|lcc@ + system&@xmath81&@xmath146 & system&@xmath81 & @xmath103 + @xmath118 & 0.0065 & & @xmath147 & 0.058 & + @xmath148 & 0.012 & & @xmath149 & 0.025 & + @xmath150 & 8 & 9.55 & @xmath151 & 6 & 13.74 + @xmath96 & 24 & 3.63 & @xmath152 & 23.75 & 3.66 + @xmath153 & 6.06 & 13.55 & @xmath154 & 6.81 & 11.63 + @xmath155 & 5508 & 1.09 & @xmath156 & 1428 & 1.18 + @xmath157 & 1.66 & 198.62 & @xmath158 & 1.28 & 488.34 + @xmath159 & 5 & 17.87 & @xmath160 & 12 & 6.25 + @xmath161 & 0.66 & 1.8@xmath162 & @xmath163 & 0.44 & 1.8@xmath164 + @xmath165 & 24.77 & 3.56 & @xmath138 & 31.22 & 3.09 + @xmath166 & 44.89 & 2.56 & @xmath139 & 57.82 & 2.29 + @xmath167 & 65.04 & 2.18 & @xmath168 & 84.45 & 1.98 + @xmath169 & 32.13 & 3.04 & @xmath170 & 41.84 & 2.65 + @xmath171 & 38.49 & 2.76 & @xmath172 & 48.11 & 2.48 +
before our conclusions , it is relevant a discussion on the applicability of the born - oppenheimer approximation to the particular configuration of excited states in three - charged systems .
the separation of variables approximation , expressed in the assumption of a product form for the wave function of the three - body system , considered in eq .
( [ eq04 ] ) , breaks down when the third particle gets close to the neutral subsystem , with @xmath26 of the order of the corresponding bohr radius .
for some three - body systems , as for example systems with proton and antiproton , nuclear force effects can also be relevant to be considered in addition to coulomb effects .
while in the asymptotic region the excited - state three - charged wave function can be described well by the bohr - oppenheimer approximation , as the size of the system gets smaller the magnitude of non - adiabatic effects is enhanced . in this case , in order to go beyond the adiabatic approximation , the appropriate description of the wave function will consist of an expansion with several higher - order terms , such that one can not disregard the coupling between them . however , once one large three - charged state is formed with a size much larger than the dipole radius , i.e. , with the wave function having major contribution from the asymptotic region , the non - adiabatic effects can be translated to a boundary condition on the wave function , at a radius characteristic of the region where non - adiabatic effects start to be important , which can be roughly estimated to be of the order of few bohr radii . within our present approximation
we can not obtain the first large bound excited state that spreads out in the asymptotic region of the dipole potential , as its energy is determined by non - adiabatic contributions .
but , once this energy is known , the energies of the other large excited states follow the expression derived from the dipole interaction , emerging the geometrically scaled pattern given in eq .
( [ eq12 ] ) . in summary , considering a charged particle captured by a dimer ( bound two - body subsystem ) , within an adiabatic approximation to the coulombic three - body system with arbitrary masses , we derived the strength of the attractive dipole interaction with the critical condition for level condensation below the first excited state of the dimer , as given in eqs .
( [ eq10 ] ) and ( [ eq11 ] ) .
the adiabatic approach is justified when the energy of the relative motion of the dimer center - of - mass , with respect to the third particle , is small compared to the subsystem binding energy ; and physically reasonable for a weak dipole potential interacting with the spectator particle . the particular straightforward charge - mass expression obtained for the leading term of the dipole interaction , as well as the condition to occur the geometrically scaled energy spectrum , can be very useful as an insight into the analysis of three - body charged systems with different mass relations . of particular interest is the use of such approach to study interaction properties of matter and antimatter .
another relevant application can be found in the interaction of excitons with electrons in semiconductors @xcite . as the electron and hole in the interacting pair of excitons can acquire effective individual masses distinct from the free electron mass @xcite
, a general charge - mass conditions can be very helpful to analyze the relation between effective masses and the observed spectrum .
we finally note that the present approach can be very useful to study capture reactions of a charged particle by a neutral two - body system , in view of possible three - body configurations . from a configuration without spectrum below the state @xmath27
, one can generate a system with very dense spectrum by exchanging the spectator particle with the same charge particle of the neutral system .
an exchange mechanism can change drastically the observed spectrum : for @xmath96 we have @xmath1733.63 ( levels below the first excited state of @xmath7 ) , whereas for @xmath174 we have @xmath17313.74 ( levels below the first excited state of the hydrogen atom ) .
lt thanks profs .
s.b . duarte and r. shellard for the hospitality in cbpf .
we acknowledge financial support from fundao de amparo pesquisa do estado de so paulo and conselho nacional de desenvolvimento cientfico e tecnolgico . 99 j. a. wheeler , ann .
sci.*48 * , 219 ( 1946 ) .
a. p. mills , jr .
lett.*46 * , 717 ( 1981 ) .
j. botero and c.h .
greene , phys .
lett . * 56 * , 1366 ( 1986 ) ; j. botero , phys .
rev . a*35 * , 36 ( 1987 ) .
j. m. rost and d. wintgen , phys .
lett.*69 * , 2499 ( 1992 ) .
z. papp , j. darai , j. zs .
mezei , z. t. hlousek , c-.y .
hu , phys .
* 94 * , 143201 ( 2005 ) .
r. krivec , m. i. haftel , and v. b. mandelzweig , phys .
a * 47 * , 911 ( 1993 ) ; s.k .
adhikari , phys .
lett . a * 294 * , 308 ( 2002 ) ; g.w.f .
drake and m. grigorescu , j. phys .
b : at.mol .
* 38 * , 3377 ( 2005 ) ; f. fleischer , _ et al .
* 96 * , 063401 ( 2006 ) ; m. puchalski and a. czarnecki , phys .
* 101 * , 183001 ( 2008 ) .
w. kolos , c.c.j .
roothaan , and r.a .
sack , rev .
* 32 * , 178 ( 1960 ) .
pekeiris , phys .
* 126 * , 1470 ( 1962 ) ; phys .
a * 12 * , 493 ( 1975 ) .
frost , m. inokuti , and j.p .
lowe , j. chem . phys .
* 41 * , 482 ( 1964 ) ; r.k .
bhaduri and y. nogami , phys .
a * 13 * , 1986 ( 1976 ) ; s.g .
lie , y. nogami , and m.a .
preston , phys .
a * 18 * , 787 ( 1978 ) ; a. martin , j.m .
richard and t.t .
wu , phys .
a * 46 * , 3697 ( 1992 ) ; r.a .
sultanov , phys . rev .
a * 50 * , 2376 ( 1994 ) ; t.n .
et al . _ ,
science * 286 * , 2474 ( 1999 ) ; v.i .
korobov , phys .
a * 61 * , 064503 ( 2000 ) ; k. varga , j. usukura , and y. suzuki , phys .
lett . * 80 * , 1876 ( 1998 ) ; t.k .
rebane and a.v .
filinsky , phys . of atomic nuclei * 65 * , 41 ( 2002 ) ; s. takahashi and k. takatsuka , j. of chem . phys . * 124 * , 144101 ( 2006 ) ; s. paul , eur .
j. d * 46 * , 483 ( 2008 ) .
lin , phys .
rep . * 257 * , 1 - 83 ( 1995 ) .
armour , j .-
richard , and k. varga , phys . rep . * 413 * , 1 ( 2005 ) .
schultz , r.e .
olson , and c.o .
reinhold , j. phys .
b : at . mol .
* 24 * , 521 ( 1991 ) .
cassidy and a.p .
mills , jr .
, nature * 449 * , 195 ( 2007 ) .
s. kilic , j .-
karr , and l. hilico , phys .
a * 70 * , 042506 ( 2004 ) .
l. landau and e. lifchitz , _ quantum mechanics _ ( mir , moscow , 1966 ) ; p.m. morse and h. feshbach , _ methods of theoretical physics _ ( mcgraw - hill , new york , 1953 ) ; [ see , also , yu .
n. ovchinnilav and i. m. sigal , ann . of phys .
* 123 * , 274 ( 1979 ) ] .
camblong , l.n .
epele , h. fanchiotti , c.a .
garcia canal , phys .
* 87 * , 220402 ( 2001 ) .
e. braaten and h .- w .
hammer , phys .
rep . * 428 * , 259 ( 2006 ) .
z. idziaszek , t. calarco , p.s .
julienne , and a. simoni , phys .
rev . a * 79 * , 010702(r ) ( 2009 ) .
v. efimov , phys .
b * 33 * , 563 ( 1970 ) . t. kraemer et al , nature * 440 * , 315 ( 2006 ) . a. c. fonseca , e. r. redish , and p. e. shanley , nucl .
a*320 * , 273 ( 1979 ) .
a. delfino and t. frederico , phys .
c*53 * , 62 ( 1996 ) ; a. delfino , t. frederico , and l. tomio , few - body systems * 28 * , 259 ( 2000 ) . l.v .
butov , _ et al .
_ , nature * 417 * , 47 ( 2002 ) ; l.v .
butov , a.c .
gossard , and d.s .
chemla , nature * 418 * , 751 ( 2002 ) .
tan , n.d .
drummond , and r.j .
needs , phys .
b * 71 * , 033303 ( 2005 ) .
p.m. platzman and a.p .
mills , jr . , phys .
b * 49 * , 454 ( 1994 ) . d.b .
cassidy , _ et al .
lett . * 95 * , 195006 ( 2005 ) .
( see also http://focus.aps.org/story/v16/st16 ) . | within an adiabatic approximation to the three - body coulomb system , we study the strength of the leading order conformaly invariant attractive dipole interaction produced when a slow charged particle @xmath0 ( with mass @xmath1 ) is captured by the first excited state of a dimer [ with individual masses and charges @xmath2 ) and ( @xmath3 ) ] .
the approach leads to a universal mass - charge critical condition for the existence of three - body level condensation , @xmath4}>\left|{q_1}/(24\ ; q_3)\right|$ ] , as well as the ratio between the geometrically scaled energy levels .
the resulting expressions can be relevant in the analysis of recent experimental setups with charged three - body systems , such as the interactions of excitons , or other matter - antimatter dimers , with a slow charged particle . |
in quantum mechanics , objects are described by wavefunctions .
these take the form of complex superpositions of various evolutionary alternatives , or states .
although successful in describing many aspects of the quantum world , this picture often leads to troubling interpretations when extrapolated to the macroscopic level .
one issue that has suffered long debate is the fact that one never observes a superposition of states .
rather , one only observes a system s _ basic _ or _ stationary _ states .
we are therefore forced to provide a mechanism by which quantum wavefunctions reduce to their stationary states .
this process is called _ wavefunction collapse _ or _
state reduction_. motivated by the basic conflicts which exist between general relativity and quantum mechanics , a number of authors have proposed the idea that wavefunction collapse is an objective phenomenon which arises due to gravitational effects @xcite .
for example penrose @xcite has suggested a scheme in which a superposition of two stationary quantum states should be fundamentally unstable if there exists a significant mass displacement between them . in this case
there should be some characteristic timescale @xmath3 for decay into the basic states .
although a detailed estimate of @xmath3 would require a full theory of quantum gravity , under this hypothesis it is reasonable to expect that for non - relativistic systems @xmath4 where @xmath5 is the gravitational self - energy of the difference between the mass distributions of the two states .
the explicit nature of the basic states in this consideration is somewhat unclear .
we can not simply regard the position of a lump of mass as a basic state , because then we would be forced to regard any general state of a particle as a superposition . as a possible solution to this problem
, penrose proposes that these ( non - relativistic ) basic states are solutions of the schroedinger equation @xmath6 where the additional term represents a coupling to a certain gravitational potential @xmath7 . this potential is determined ( via the poisson equation ) by the expectation value of the mass distribution in the state determined by the wavefunction .
for single particle systems , the matter density is determined by the probability density from the wavefunction , and so @xmath8 where @xmath9 is newton s gravitational constant , and @xmath10 is the mass of the single particle .
equations ( [ e2],[e3 ] ) are dubbed the schroedinger - newton ( sn ) equations @xcite .
a preliminary investigation of the properties of the solutions to the sn equations was recently carried out by moroz et .
al . @xcite . under the assumptions of spherical symmetry in @xmath11 dimensions , and by demanding only that @xmath7 and @xmath12 be everywhere smooth
, they discovered a discrete family of bound state solutions , labelled by an integer @xmath13 .
each solution is a normalizable wavefunction , and the @xmath14th solution has @xmath14 zeros .
the energy eigenvalues associated with each of these solutions are negative , and monotonically converge to zero for large @xmath14 .
these results can be justified analytically @xcite .
the energy eigenvalues are the differences between a given bound state and a continuum ` superposition ' state , and so provide via ( [ e1 ] ) an estimate of the timescale of self - collapse of a single particle of mass @xmath10 .
the energy eigenvalues scale like @xmath15 , and so particles of small mass have extremely long self - collapse times for a nucleon mass the estimate is @xmath16s @xcite . a recent related study by soni is commensurate with these results @xcite . relaxing
the assumption of spherical symmetry is in general a difficult task due to the non - linearity of the sn equations .
however there are two situations in which this is fairly straightforward : cylindrical symmetry with no angular momentum and planar symmetry .
rewriting the sn equations for these cases effectively reduces them to @xmath17 and @xmath18 dimensional situations respectively .
these cases , along with the spherically symmetric case , can be simultaneously recovered by rewriting the spherically symmetric sn equations in @xmath0 dimensions .
motivated by the above , we consider in this paper an analysis of the @xmath0-dimensional spherically symmetric sn equations , for @xmath19 .
although the higher - dimensional cases are of less direct physical interest that the @xmath20 cases , such a study affords us some insight into the dimensional behaviour of the sn system .
this behaviour may be of more than pure pedagogical interest since many candidate approaches to quantum gravity are typically cast in higher dimensions ( superstring theory being the obvious example ) .
any solution to the sn equations ( [ e2],[e3 ] ) must be normalizable ( i.e. square - integrable ) . integrating the probability density over all space yields @xmath21 where @xmath22 is a dimensionless number , and so the wavefunction must be rescaled to ensure there is unit probability of finding the particle somewhere in space .
writing @xmath23 , the sn equations then become @xmath24 and we see that the normalization factor enters the system due to its non - linearity . redefining variables in ( [ sna],[snb ] ) via @xcite @xmath25 where @xmath26 yields @xmath27 where we can assume that @xmath28 is real without loss of generality .
the parameters @xmath29 and @xmath30 have units of @xmath31 and @xmath32 energy respectively .
the system ( [ e6],[e7 ] ) is invariant under the rescaling transformation @xmath33 independent of the dimension @xmath0 , where @xmath34 has units of inverse length . using this transformation we can rewrite the system in terms of fully dimensionless functions @xmath35 of dimensionless variables . for the spherically symmetric case the @xmath0-dimensional laplacian operator is @xmath36 and
so ( [ e6],[e7],[newnorm ] ) become @xmath37 where the prime denotes differentiation with respect to @xmath38 .
we must require that the integral in ( [ normint ] ) be finite in order for the sn system to be physically meaningful .
we therefore seek solutions to the sn system that are finite for all values of @xmath39 ; continuity requirements imply that these solutions are everywhere smooth .
note that @xmath40 is a dimensionless quantity .
however , our equation ( [ normint ] ) clearly has a problem for @xmath41 , where we lose our ability to rescale using @xmath34 .
because we still require the wavefunction to be normalized to unity , we are forced to introduce a constant @xmath42 in the @xmath41 case so that @xmath43
. then our equations ( [ e8 ] ) , ( [ e9 ] ) , and ( [ normint ] ) become @xmath44 we will not always include this constant @xmath45 explicitly in the following discussion . it will simply be understood to occur in the rescaling when @xmath41 .
smoothness implies that @xmath46 and @xmath47 are finite . with this information we can rewrite ( [ e8],[e9 ] ) as @xmath48
it is straightforward to show that picard s theorem @xcite is satisfied by this system of equations , and so given @xmath49 and @xmath50 a unique solution to ( [ e8],[e9 ] ) exists within a range @xmath51 .
the @xmath52 versions of the integral equations ( [ e8a],[e9a ] ) require some care . integrating ( [ e8],[e9 ] ) explicitly for these cases yields @xmath53 for @xmath54 and @xmath55 for @xmath56 .
both of these sets of equations may be obtained formally from ( [ e8a],[e9a ] ) by inserting these values of @xmath0 , the latter case being understood as the limit @xmath57 .
note that for all of these cases the function @xmath58 is montonically decreasing since @xmath59 a result valid for all @xmath60 .
hence if @xmath61 , then @xmath62 and so @xmath63 will also diverge for large @xmath38 .
consequently only @xmath64 is of physical interest . for @xmath65
the function @xmath66 initially . by rewriting @xmath67 equation ( [ e8 ] ) can be rewritten in the form @xmath68
s=0 \label{e11a}\ ] ] which for @xmath69 is bessel s equation .
hence near the origin we expect @xmath70 to have oscillatory behaviour .
however when @xmath58 becomes negative eq .
( [ e11a ] ) is similar to the modified bessel equation and the behaviour of @xmath70 will be a linear combination of exponentially amplified and damped functions .
normalization requirements imply that only the exponentially damped solutions are allowed . in this case
@xmath63 decays like an exponential times an inverse power of @xmath38 and so the integral in ( [ e9a ] ) will be finite , yielding a finite @xmath58 at large @xmath38 .
this behaviour was already noted for the @xmath71 case @xcite ; we see here that it is valid for arbitrary @xmath72
the form ( [ e8],[e9 ] ) of the sn equations is not well - suited for numerical study .
a more convenient form is @xmath73 we are interested in finding solutions to the system ( [ e11],[e12 ] ) for that are smooth and finite for all @xmath38 .
these are the bound state solutions referred to earlier .
we will require our initial conditions , @xmath74 and @xmath75 to be greater than zero , a constraint which avoids both the trivial solutions @xmath76 and the non - normalizable solutions @xmath77 , and is consistent with the rescaling freedom ( [ rescale ] ) .
it is straightforward to obtain the power - series solutions to ( [ e11],[e12 ] ) near the origin : @xmath78 where the functions @xmath79 and @xmath80 and their first derivatives are required to be finite at the origin . using the rescaling freedom to set @xmath81 ( which is equivalent to setting @xmath82 )
, we see that for any dimension @xmath0 the solutions near the origin depend only on a single free parameter .
integrating the system ( [ e11],[e12 ] ) using a fehlberg fourth - fifth order runge - kutta method in maple we find that in each dimension @xmath0 an infinite set of discrete bound states appears as expected .
these can be identified by the number of local extrema .
an infinite amount of precision is required in the choice of @xmath49 such that the solutions do not diverge as @xmath38 increases . this value of @xmath49 marks the transition between solutions in which @xmath63 diverges to @xmath83 and solutions where @xmath63 diverges to @xmath84 .
as we increase the accuracy of @xmath49 for a specific bound state , the value of @xmath38 at which @xmath63 blows up increases .
we will see that the values for @xmath49 and the distance between bound states vary significantly with dimension .
the general method for finding solutions to the @xmath0-dimensional system is the same as that in the @xmath71 case @xcite .
choose a value of @xmath85 for which @xmath63 has @xmath14 zeros and diverges to @xmath86 at finite @xmath38 . then select a slightly smaller value of @xmath87 which has @xmath88 zeros and diverges to @xmath89 for some finite @xmath38 . between these two values of @xmath49
is a value @xmath90 for which @xmath63 has @xmath14 zeros and is smooth and finite for all values of @xmath38 and is square - integrable
. the potential @xmath58 will also be smooth and finite , with finite energy eigenvalue @xmath91 . by successively narrowing this interval the bound - state value of @xmath49
can be achieved to any desired accuracy .
the bound state wavefunction with @xmath88 zeros may be obtained by choosing another value @xmath92 and repeating the procedure with @xmath93 .
we find that the values @xmath94 are rapidly decreasing functions of the dimension @xmath0 , and that the bound state solutions appear to accumulate at @xmath2 .
we are limited by numerical accuracy in carrying out our investigations for dimensions @xmath95 due to this effect
. a summary of this data is presented in table 1 .
[ cols="^,^,^,^,^ " , ] table 1 .
initial conditions @xmath96 for the first @xmath97 to fourth @xmath98 bound state wavefunctions , for one to six spatial dimensions .
the accumulation of the higher bound states for large dimensions results in the decreasing accuracy of the lower right hand entries of the table .
figures 1,2 and 3 illustrate the solutions for @xmath79 for @xmath54 , for the first , second and third bound state transitions .
figures 4 to 8 illustrate the first bound state transitions for dimensions @xmath56 to @xmath2 .
figure 9 illustrates the accumulation effect at @xmath2 .
we turn now to the problem of determining the energy eigenvalues for the bound states . from ( [ e9 ] ) and ( [ e9a ] ) we find that the large-@xmath99 expansion of @xmath100 is @xmath101 where @xmath102 these expansions are formally valid for all @xmath103 ; the explicit expansions for @xmath52 are @xmath104 and @xmath105 where @xmath106 is an arbitrary length scale .
the energy eigenvalue is given by @xmath107 the quantity @xmath108 is the point at which the potential @xmath109 vanishes . for @xmath110 @xmath111 .
however for @xmath112 the potential diverges there and the situation is more delicate . for @xmath54
a newtonian gravitational potential diverges linearly with @xmath113 . here
we extract the @xmath38-independent term from ( [ e15 ] ) to obtain @xmath114 ; the result is still given by the right - hand side of ( [ e18 ] ) with @xmath54 .
effectively we have set @xmath115 in ( [ e15 ] ) even though this equation is a large @xmath99 expansion .
for @xmath56 the newtonian potential diverges at both large and small @xmath116 .
the normalization point @xmath117 is therefore arbitrary , and we set @xmath118 so that ( [ e18 ] ) remains valid in this case as well .
this is tantamount to setting @xmath119 .
a natural normalization point is to choose @xmath106 to be the point at which the potential @xmath120 .
note that under the rescaling transformation ( [ rescale ] ) @xmath121 , where @xmath122 and so the combination @xmath123 is invariant .
the parameter @xmath34 ( or alternatively @xmath50 ) is completely arbitrary , serving the sole function of setting the length scales of the problem in units of @xmath124 . solving ( [ normint ] ) for @xmath34 yields
@xmath125 with this normalization , we find that @xmath126 and so the @xmath0-dimensional energy is given in units of @xmath127 .
the quantities @xmath128 and @xmath129 are straightforwardly solved numerically from ( [ e14a],[e14b ] ) .
the preceding expression can be rewritten as @xmath130 where @xmath131 is the planck mass in @xmath132 dimensions ( in one spatial dimension the quantity @xmath133 is unitless , and can be absorbed into the normalization constant @xmath22 ) .
for @xmath134 the energy eigenvalue is an increasing function of the particle mass , whereas for @xmath135 it is a decreasing function .
the most rapid increase is @xmath136 for @xmath71 , with only linear and quadratic dependence in @xmath54 and @xmath17 respectively .
the time - scale for collapse of the state vector is therefore most rapid in a world with three spatial dimensions for any bodies whose mass is appreciably larger than the planck mass . in a world of more than four dimensions the collapse is fastest for the lightest - mass particles , leading to behaviour which is at complete odds with that expected in the macroscopic world .
if we interpret the lower - dimensional cases to be situations in three spatial dimensions with planar or cylindrical symmetry , a somewhat different picture emerges . inserting the constants into ( [ eplanck ] ) , the dependence of the energy eigenvalues for the planar , cylindrical and spherically symmetric cases are respectively @xmath137 where @xmath138 is the planck length and @xmath139 is the dimension of the box ( or cylinder ) in which the particle is confined . from the perspective of state - vector reduction
, a particle prepared in a plane - wave or cylindrical state will have a radically different time - scale of collapse than the same particle prepared in a spherically symmetric state .
a neutron in a spherically symmetric state will have a collapse time of @xmath140 whereas a neutron confined to a square box @xmath141 m in cross - sectional size will have a collapse time of only @xmath142 .
both of these times are much longer than the age of the universe .
however by confining a neutron to a rectangular pipe whose width is on the order of 10 picometers , the collapse time becomes on the order of @xmath143 , or about 12 days .
such an experiment would push the limits of current technologly . note that for @xmath41 the probability is scale - invariant , and it is not possible to solve ( [ normint ] ) for @xmath34 .
the energy scale in this dimension is independent of the wavefunction normalization .
if we work through the problem using the @xmath45 rescaled equations ( [ e9d4 ] ) and ( [ normintd4 ] ) , we can recover an expression for the energy eigenvalues in this dimension using ( [ e - eigen ] ) . when rescaling equations ( [ e13 ] ) we can choose to absorb @xmath45 in the rescaling of the expansion coefficients : @xmath144 where @xmath145 and @xmath146 are equivalent to ( [ e14c ] ) and ( [ e14d ] ) . then solving ( [ normintd4 ] ) for @xmath45
we get @xmath147 we use this normalization to solve for the energy eigenvalue when @xmath41 @xmath148 clearly we are left with an unsolved parameter , @xmath149 , which is at this point completely arbitrary .
this is a consequence of the introduction of the rescaling constant @xmath45 in the @xmath41 case .
a graph of the invariant @xmath150 is instructive because it reflects the behaviour of the energy eigenvalue solutions with bound state that is independent of the value of the @xmath0-dimensional newton constant .
figure 10 shows that the energy eigenvalues are positive and increase with bound state for @xmath52 , consistent with what we expect for the form of the classical newtonian gravitational potential in these dimensionalities .
figure 11 shows that the energy eigenvalues are negative for dimensions greater than 3 . in this graph , the values of @xmath149 and @xmath151
are set to unity for @xmath41 .
we have presented the preliminary results of our numerical and analytical investigation of the schroedinger - newton equations in @xmath0 dimensions under the assumption of spherical symmetry .
consistent with previous analyses in 3 dimensions , we have demonstrated numerically the existence of a discrete set of `` bound - state '' solutions for the equations , which are associated with different energy eigenvalues .
the bound states appear to accumulate for dimensions greater than or equal to six .
the bound state energies for @xmath52 are positive and increase with increasing bound state number .
for @xmath152 the energies are negative and converge to zero with increasing eigenvalue number . this is consistent with moroz @xcite , who found that for @xmath71 the energy eigenvalues converge to zero as the order of the bound state increases . for higher dimensions ,
the variation of the eigenvalues with bound state becomes less pronounced , as illustrated in figure 11 . for @xmath153
we find that the eigenvalues accumulate , and we are unable to determine any energy differences between bound states to 10 significant digits .
moroz ventured further in the three dimensional study to assign numerical estimations for timescales of collapse , using the lowest bound state solutions to determine energy eigenvalues .
however the analogous calculation using general @xmath0 dimensions is not strictly correct due to possible dimensional variation in newton s constant @xmath154 which appears in ( [ e5 ] ) . we will not attempt to predict the variation of @xmath154 with dimension here , although we note that kaluza - klein theory suggests that @xmath154 is proportional to the three dimensional @xmath155 divided by the square root of the volume element of the extra space dimensions .
we leave any such numerical estimates for future work
. however if we interpret the lower - dimensional cases to be systems in three spatial dimensions with either planar or cylindrical symmetry , a different picture emerges .
the collapse time scale depends not only on the mass of the particle but also on the size of the box in which it is confined . if the gravitational influence of the box can be neglected and if the gravitational field of the neutron in a plane wave state has planar symmetry , then we find that the collapse timescales differ enormously between the two situations .
a free neutron under its own gravitational self - influence will experience a collapse of its wavefunction on a timescale many orders of magnitude larger than the age of the universe , whereas one confined to a rectangular pipe whose cross - sectional width is about 10 picometers in size will decay in about two weeks .
an obvious extension of our work involves introducing external field into our version of the sn equations , in the form of a point source perturbation .
this can be added to the schroedinger equation to give the perturbed system @xmath156 here @xmath157 is the mass which generates the perturbing field .
this system is quite interesting in the respect that the form of the bound state solutions is drastically altered from that of the `` free field '' equations considered in this paper .
a preliminary study of this system indicates that each bound state solution seems to have three regions . for small @xmath157 ,
the solutions asymptotically approach those of the free field or non - perturbed case , as expected . for moderate values of @xmath157
, the equations are goverened by both the free field and point source field , causing the values of @xmath49 , for which the solutions do not diverge , to alter considerably .
finally , for very large @xmath157 , the solutions are goverened almost entirely by the point source perturbation , and become similar to the hydrogen atom solutions of the schroedinger equation .
the effect of incresing dimension causes the `` transition '' values of @xmath157 ( which separate the different regions of the solutions ) to decrease .
a full study of the point source perturbation is currently in progress .
* acknowledgments * this work was supported in part by the natural sciences and engineering research council of canada .
9 f. karolyhazy , nuovo cim . * a42 * ( 1966 ) 390 ; t.w.b .
kibble _ quantum gravity 2 : a second oxford symposium _ eds .
isham , r. penrose and d.w .
sciama ( oxford : oxford university press , 1981 ) pp 6380 ; f. karolyhazy , a. frenkel and b. lukacs _ quantum concepts in space and time _ eds .
r penrose and c j isham ( oxford : oxford university press 1986 ) pp 109 ; l. diosi phys .
lett . * * a120 ( * * 1987 * ) * 377 ; l. diosi , phys . rev . * a40 * ( 1989 ) 1165 ; i.c .
percival proc .
* a451 * ( 1995 ) 503 . | we investigate a @xmath0 dimensional generalization of the schroedinger - newton equations , which purport to describe quantum state reduction as resulting from gravitational effects . for a single particle ,
the system is a combination of the schroedinger and poisson equations modified so that the probability density of the wavefunction is the source of the potential in the schroedinger equation . for
spherically symmetric wavefunctions , a discrete set of energy eigenvalue solutions emerges for dimensions @xmath1 , accumulating at @xmath2 . invoking heisenberg s uncertainty principle to assign timescales of collapse correspoding to each energy eigenvalue , we find that these timescales may vary by many orders of magnitude depending on dimension .
for example , the time taken for the wavefunction of a free neutron in a spherically symmetric state to collapse is many orders of magnitude longer than the age of the universe , whereas for one confined to a box of picometer - sized cross - sectional dimensions the collapse time is about two weeks .
* studies of the schroedinger - newton equations in @xmath0 dimensions * + r. melko and r.b .
mann + dept . of physics ,
university of waterloo waterloo , ont n2l 3g1 , canada + pacs numbers : 04.62 , 02.90 , 03.65 , 03.65 g , 04.20 + + |
monolayer mos@xmath0 has attracted much attention due to its promising applications in electronics,@xcite optoelectronics,@xcite valleytronics@xcite and also spintronics.@xcite for the application of spintronic devices , the suitable spin lifetime and spin diffusion length are required.@xcite this indicates the importance of the investigations on the spin relaxation and spin diffusion in this material .
very recently , spin relaxation has been studied in monolayer mos@xmath0.@xcite wang and wu@xcite calculated the in - plane spin relaxation time of electrons due to the dyakonov - perel@xcite ( dp ) and elliot - yafet@xcite ( ey ) mechanisms with the intra- and inter - valley processes included .
they pointed out that the dp mechanism , which results from the inhomogeneous broadening@xcite together with any scattering process , dominates the spin relaxation .
the inhomogeneous broadening is from the spin - orbit coupling ( soc ) of the conduction band@xcite @xmath1\hat{\bf z } , \label{soctime}\ ] ] where the @xmath2-axis is perpendicular to the monolayer mos@xmath0 plane ; @xmath3 , @xmath4 and @xmath5 are the strengths of the soc ; @xmath6 represents the k(k@xmath7 ) valley .
the first term of the soc , which is momentum independent , only induces the inter - valley dp spin relaxation whereas the last two terms are momentum dependent , which lead to the intra- and inter - valley spin relaxation processes .
in addition , as only the last term ( i.e. , negligible anisotropic cubic one ) causes the dp spin relaxation with the electron - impurity scattering , the electron - impurity scattering is shown to play a marginal role in the spin relaxation.@xcite in contrast to the spin relaxation in time domain , the inhomogeneous broadening in spin diffusion for in - plane spins is determined by the spin precession frequency@xcite @xmath8\hat{\bf z } \label{socspatial}\ ] ] when the spin diffusion is along the @xmath9-axis . here ,
@xmath10 stands for the effective mass . due to the existence of @xmath11 ,
all three terms become momentum dependent , which can induce the intra- and inter - valley relaxations for in - plane spins along the diffusion .
this is different from the case of the spin relaxation in time domain as previously mentioned .
in addition , @xmath11 also makes the first two terms ( i.e. , the leading ones ) anisotropic .
this suggests that the electron - impurity scattering may play an important role in the in - plane spin diffusion , which is of great difference from the case of the spin relaxation in time domain in monolayer mos@xmath0,@xcite but similar to the case of the spin diffusion in semiconductors@xcite and single - layer graphene.@xcite as for the out - of - plane spins , the spin diffusion length is infinite since the spin precession frequency @xmath12 [ see eq .
( [ socspatial ] ) ] is along the out - of - plane direction .
however , this is not the case in the presence of an out - of - plane electric field .
very recently , bishnoi and ghosh@xcite investigated the out - of - plane spin diffusion with this electric field applied .
they showed that the out - of - plane spins relax during the spin diffusion since the out - of - plane electric field induces a rashba soc , which provides an inhomogeneous broadening in the spatial domain for out - of - plane spins.@xcite however , the rashba soc they used is incomplete according to the recent work by kormnyos _
et al._.@xcite in addition , the electron - electron coulomb and electron - impurity scatterings , which have been shown to play an important role in spin diffusion in semiconductors@xcite and single - layer graphene,@xcite are absent in their work.@xcite moreover , they also overlooked the inter - valley electron - phonon scattering , which is of crucial importance in spin relaxation in time domain.@xcite it is noted that in addition to the out - of - plane electric field , an in - plane magnetic field can also lead to the out - of - plane spin relaxation along the spin diffusion.@xcite this is because the in - plane magnetic field ( * b * ) gives rise to a spin precession frequency in the spatial domain as @xmath13 , which provides an inhomogeneous broadening for out - of - plane spins during the spin diffusion . in this work
, we investigate the electron spin diffusion in monolayer mos@xmath0 in the absence of the external electric and magnetic fields . as the contribution of the spin - flip scattering due to the ey mechanism
is negligible,@xcite we only take into account the spin conserving scattering . with the electron - impurity ( inter - valley electron - phonon )
scattering included , the intra - valley ( inter - valley ) diffusion process for in - plane spins is studied by analytically solving the kinetic spin bloch equations ( ksbes).@xcite we find that the intra - valley process dominates the in - plane spin diffusion , which is very different from the case of the spin relaxation in time domain in monolayer mos@xmath0 where the inter - valley process can be comparable to or even more important than the intra - valley one.@xcite moreover , it is shown that the in - plane spin diffusion length decreases with the increase of the impurity density but increases with increasing electron density in both the degenerate and nondegenerate limits .
very interestingly , with the electron - electron coulomb scattering further taken into account , the in - plane spin diffusion length shows an opposite electron density dependence in the nondegenerate limit compared to the one with only the electron - impurity scattering .
this paper is organized as follows . in sec .
ii , we introduce our model and the ksbes . in sec .
iii , we investigate the in - plane spin diffusion by analytically solving the ksbes .
we summarize in sec . iv .
the effective hamiltonian of the conduction band near the k(k@xmath7 ) point in monolayer mos@xmath0 reads @xmath14 according to the latest work by wang and wu.@xcite here , @xmath15 with @xmath10 representing the effective mass ; @xmath16 is given in eq .
( [ soctime ] ) ; @xmath17 are the pauli matrices .
we then construct the microscopic ksbes@xcite to investigate the electron spin diffusion in monolayer mos@xmath0 .
the ksbes can be written as@xcite @xmath18 with @xmath19 being the density matrices of electrons at position @xmath9 and time @xmath20 .
the diffusion terms are described as @xmath21 by assuming that the spin diffusion is along the @xmath9-axis .
the coherent terms @xmath22 can be found in ref . .
as for the scattering terms @xmath23 , we neglect the spin - flip ones due to the ey mechanism since the contribution of the ey mechanism is negligible.@xcite here , we only include the spin conserving terms , i.e , the electron - electron coulomb , electron - impurity , intra - valley electron - acoustic phonon , electron - optical phonon , and also the inter - valley electron - phonon@xcite ( electron - kta , -kla , -kto , and -klo ) scatterings . here ,
kta , kla , kto and klo correspond to the transverse acoustic , longitudinal acoustic , transverse optical and longitudinal optical phonon modes at the k point , respectively .
the detailed expressions of the above scattering terms are given in ref . and
the scattering matrix elements are given in ref . .
we first focus on the intra - valley diffusion process for in - plane spins by simplifying the ksbes with only the electron - impurity scattering included in the scattering term . by taking the steady - state condition ,
the fourier components of the density matrix with respect to @xmath24 are given by@xcite @xmath25\nonumber\\ & & \mbox{}=-\frac{i}{2\hbar}[(2\lambda_c+a_1k^2)\sigma_z,\rho^{l}_{k}(x ) ] -\rho^{l}_{k}(x)/\tau_i^l,\end{aligned}\ ] ] where the valley index @xmath26 is neglected for the intra - valley process .
here , @xmath27\equiv ab - ba$ ] is the commutator ; @xmath28 ; @xmath29 stands for the @xmath30th - order momentum scattering rate with @xmath31 . @xmath32 and @xmath33 represent the impurity density and electron - impurity scattering matrix element , respectively .
it is noted that in recent experiments , the mobilities at room temperature are reported to be of the order of @xmath34-@xmath35@xmath36v@xmath37s@xmath37 in monolayer mos@xmath0 on sio@xmath0 substrate.@xcite by fitting to the mobilities in these experiments , one obtains the impurity densities ( e.g. , @xmath38@xmath39 corresponding to the mobility @xmath35@xmath36v@xmath37s@xmath37).@xcite with these impurity densities , the electron - impurity scattering is in the strong scattering limit ( in time domain ) , i.e. , @xmath40 ( e.g. , @xmath41 for @xmath38@xmath39 ) @xmath42 , with @xmath43 denoting the ensemble average . then , one can only keep the lowest two orders of @xmath44 ( i.e. , 0 , 1 ) and obtain@xcite @xmath45c_1 ^ 2/2+ic_2[\sigma_z,\rho^0_k(x)],\end{aligned}\ ] ] with @xmath46 and @xmath47 . by solving this equation under the boundary conditions ,
the steady - state spin vector @xmath48 $ ] can be obtained . since the system is isotropic for in - plane spins , we choose the injected spin polarization along the @xmath9-axis without losing generality . with the boundary condition @xmath49 and @xmath50 , we have @xmath51 with @xmath52 we then turn to investigate the inter - valley spin diffusion process by analytically solving the ksbes with the inter - valley electron - phonon scattering included in the scattering term .
similar to the intra - valley process , one obtains @xmath53&=&-\frac{i(2\lambda_c+a_1k^2)\mu}{2\hbar}\nonumber\\ & & \hspace{-4cm}\mbox{}\times[\sigma_z,\rho^{l}_{\mu k}(x)]-[{\rho^{l}_{\mu k}(x)}/{\tau_v^0}-{\rho^{l}_{-\mu k}(x)}/{\tau_v^l}]\end{aligned}\ ] ] based on the elastic scattering approximation.@xcite here , @xmath54 is the @xmath30th - order inter - valley momentum scattering rate with @xmath55 and @xmath56 being the phonon number and the inter - valley electron - phonon scattering matrix element , respectively .
it is noted that differing from the electron - impurity scattering , the inter - valley electron - phonon scattering is in the weak scattering limit ( in time domain ) according to the recent work by wang and wu,@xcite i.e. , @xmath57 may be relevant in the inter - valley process . for simplicity ,
we first retain the lowest two orders of @xmath44 ( i.e. , 0 , 1 ) and obtain @xmath58c_1 ^ 2/2\nonumber\\ & & \hspace{-1cm}\mbox{}+ic_3\mu[\sigma_z,2\rho^0_{\mu k}(x)-\rho^0_{-\mu k}(x)c_4],\end{aligned}\ ] ] with @xmath59 and @xmath60 . from this equation together with certain boundary conditions , we can obtain the steady - state spin vector @xmath61 $ ] . similar to the intra - valley process , we also set the initial spin polarization along the @xmath9-axis . under the boundary condition that @xmath62 and @xmath63
, we have @xmath64 where @xmath65 with higher orders ( @xmath66 ) of @xmath67 included , we find that the inter - valley spin diffusion length @xmath68 varies slightly ( not shown).@xcite in the calculation , the effective mass @xmath69 with @xmath70 being the free electron mass;@xcite the coefficients of the soc @xmath71mev [ ref . ] and @xmath72mev @xmath73.@xcite the parameters related to the scattering matrix elements are given in ref . . with these parameters ,
we calculate the intra- and inter - valley spin diffusion lengths in both the degenerate and nondegenerate limits . in the degenerate limit , the intra- and inter - valley spin diffusion lengths are given by @xmath74 and @xmath75 according to eqs .
( [ lxintra ] ) and ( [ lxinter ] ) , respectively . here , @xmath76 is the fermi wavevector . to compare the relative importance between the intra- and inter - valley processes in the spin diffusion ,
we calculate @xmath77 as the electron - impurity scattering is in the strong scattering limit in time domain ( i.e. , @xmath78 ) and the inter - valley electron - phonon scattering is in the weak scattering limit in time domain ( i.e. , @xmath79 ) , we have @xmath80 .
this indicates that the intra - valley process dominates the in - plane spin diffusion , which is very different from the case of the spin relaxation in time domain in monolayer mos@xmath0 where the inter - valley process is shown to be comparable or even more important than the intra - valley one.@xcite in the following , we only calculate the spin diffusion length due to the intra - valley process @xmath74 according to eq .
( [ lxintra ] ) . in fig .
[ fig1](a ) , we plot the electron density dependence of the in - plane spin diffusion length .
the impurity density is taken to be @xmath81@xmath39 so that the electron - impurity scattering is in the strong scattering limit in time domain .
the corresponding mobility is of the order of @xmath82@xmath36v@xmath37s@xmath37 , which is about @xmath83-@xmath84 orders of magnitude larger than those reported in the experiments.@xcite we find that the spin diffusion length due to the electron - impurity scattering ( curve with @xmath85 ) increases with the increase of the electron density .
this can be understood as follows . the intra - valley spin diffusion length is approximately given by @xmath86 according to eq .
( [ lxintra ] ) under the condition that @xmath78 as previously mentioned .
when the electron density increases , the fermi wavevector @xmath76 increases whereas the electron - impurity scattering rate @xmath87 decreases in the degenerate limit,@xcite leading to the decrease of @xmath88 and therefore the increase of the spin diffusion length .
in addition to the electron - impurity scattering , the electron - electron coulomb scattering is also taken into account whereas the intra - valley electron - phonon scattering is neglected due to the negligible contribution.@xcite as done in the spin relaxation in time domain,@xcite we calculate the in - plane spin diffusion length according to eq .
( [ lxintra ] ) with the effective momentum scattering rate @xmath89 replacing the one due to the electron - impurity scattering @xmath90 . here
, @xmath91 represents the momentum scattering rate due to the electron - electron coulomb scattering with @xmath92 and @xmath93 being the fermi energy and boltzmann constant , respectively.@xcite the results at @xmath94 , @xmath95 and @xmath96k are shown in fig .
[ fig1](a ) . by comparing these results with the one calculated with only the electron - impurity scattering ( curve with @xmath85 )
, we find that that the electron - impurity scattering plays a leading role in the in - plane spin diffusion .
this is very different from the case of the spin relaxation in time domain in monolayer mos@xmath0 where the contribution of the electron - impurity scattering is marginal.@xcite moreover , we find that the intra - valley spin diffusion length decreases with the increase of the temperature when the electron density remains unchanged .
this is because the electron - impurity scattering is insensitive to the temperature in the degenerate limit@xcite whereas the electron - electron coulomb scattering increases with increasing temperature in the degenerate limit,@xcite leading to the decrease of the spin diffusion length as the temperature increases [ see eq .
( [ lxapp ] ) ] . as function of the electron density @xmath97 in the degenerate limit .
the curve with @xmath85 stands for the calculation from eq .
( [ lxintra ] ) with only the electron - impurity scattering included whereas the curves with @xmath98 , @xmath99 and @xmath100 represent the calculation based on eq .
( [ lxintra ] ) with the effective momentum scattering rate @xmath101 in place of @xmath102 at temperature @xmath94 , @xmath95 and @xmath96k , respectively . the impurity density @xmath103@xmath39 .
in addition , @xmath104 due to the electron - impurity scattering as function of the impurity density @xmath32 with @xmath105@xmath39 is shown in the inset ; ( b ) @xmath104 as function of @xmath97 in the nondegenerate limit .
the curves with @xmath85 , @xmath99 and @xmath98 stand for the calculation from eq .
( [ lxintra ] ) with the momentum scattering rate being @xmath102 , @xmath106 and @xmath101 , respectively . @xmath103@xmath39 and @xmath107k.,title="fig:",width=321 ] as function of the electron density @xmath97 in the degenerate limit .
the curve with @xmath85 stands for the calculation from eq .
( [ lxintra ] ) with only the electron - impurity scattering included whereas the curves with @xmath98 , @xmath99 and @xmath100 represent the calculation based on eq .
( [ lxintra ] ) with the effective momentum scattering rate @xmath101 in place of @xmath102 at temperature @xmath94 , @xmath95 and @xmath96k , respectively . the impurity density @xmath103@xmath39 .
in addition , @xmath104 due to the electron - impurity scattering as function of the impurity density @xmath32 with @xmath105@xmath39 is shown in the inset ; ( b ) @xmath104 as function of @xmath97 in the nondegenerate limit . the curves with @xmath85 , @xmath99 and
@xmath98 stand for the calculation from eq .
( [ lxintra ] ) with the momentum scattering rate being @xmath102 , @xmath106 and @xmath101 , respectively .
@xmath103@xmath39 and @xmath107k.,title="fig:",width=321 ] in addition , we also investigate the impurity density dependence of the in - plane spin diffusion length with the electron density @xmath105@xmath39 as shown in the inset of fig .
[ fig1](a ) , with only the electron - impurity scattering included .
it is seen that the spin diffusion length decreases with the increase of the impurity density , which can be easily understood from eq .
( [ lxapp ] ) .
then we turn to the nondegenerate limit .
the intra- and inter - valley spin diffusion lengths are calculated by @xmath108 and @xmath109 according to eqs .
( [ lxintra ] ) and ( [ lxinter ] ) , respectively . here , @xmath110 stands for the `` thermal '' wavevector.@xcite similar to the degenerate - limit case , the in - plane spin diffusion in the nondegenerate limit is also dominated by the intra - valley process . with the impurity density @xmath103@xmath39
, we calculate the electron density dependence of @xmath111 due to the electron - impurity scattering and the result is shown in fig .
[ fig1](b ) .
we find that the intra - valley spin diffusion length increases with the increase of the electron density , which results from the decrease of @xmath102 due to the increase of the screening in the nondegenerate limit.@xcite with the effective momentum scattering rate @xmath89 instead of @xmath90 , we also take into account the electron - electron coulomb scattering .
it is noted that differing from the case of the degenerate limit , @xmath112 in the nondegenerate limit with @xmath113 being the relative static dielectric constant.@xcite it is seen that the spins relax faster along the spin diffusion with increasing electron density , which is opposite to the situation with only the electron - impurity scattering included .
this can be understood as follows .
as the electron density increases , the enhancement of @xmath114 leads to the decrease of the intra - valley spin diffusion length due to the electron - electron coulomb scattering ( curve with @xmath99 ) , which suppresses the increase of @xmath108 contributed by the electron - impurity scattering .
it is noted that the feature of the spin diffusion is similar to the one of the spin relaxation in the weak scattering limit in time domain.@xcite specifically , the intra - valley spin diffusion length decreases with increasing the scattering .
this is because the spin precession frequency in the spatial domain is proportional to @xmath11,@xcite which gives strong inhomogeneous broadening along the spin diffusion .
similar behavior has also been shown in the case of spin diffusion in symmetric silicon quantum wells under an in - plane magnetic field@xcite since the spin precession frequency in the spatial domain has similar momentum dependence.@xcite
in conclusion , we have investigated the electron spin diffusion in monolayer mos@xmath0 in the absence of the external electric and magnetic fields .
the anisotropic spin precession frequency in the spatial domain leads to a marked contribution of the electron - impurity scattering in the in - plane spin diffusion .
this is of great difference from the spin relaxation in time domain in monolayer mos@xmath0 where the contribution of the electron - impurity scattering is negligible . with the electron - impurity and inter - valley electron - phonon scatterings separately included in the ksbes , we analytically study the intra- and inter - valley diffusions for in - plane spins .
the intra - valley process is shown to play a dominant role in the in - plane spin diffusion , which is different from the case of the spin relaxation in time domain in this material where the inter - valley process can be comparable to or even more important than the intra - valley one . in the dominant intra - valley process , it is shown that the in - plane spin diffusion is suppressed by increasing impurity density but enhanced with the increase of the electron density in both the degenerate and nondegenerate limits .
interestingly , with the electron - electron coulomb scattering further included , a decrease of the spin diffusion length is observed as the electron density increases in the nondegenerate limit .
this work was supported by the national natural science foundation of china under grant no .
11334014 , the national basic research program of china under grant no .
2012cb922002 and the strategic priority research program of the chinese academy of sciences under grant no .
xdb01000000 .
it is noted that the cubic term in eq .
( [ soctime ] ) is neglected here due to its marginal contribution to the spin diffusion contributed by the electron - impurity scattering as previously mentioned .
this is very different from the case of the spin relaxation in the time domain in monolayer mos@xmath0 . the relative error between the inter - valley spin diffusion length with the orders @xmath115 and @xmath116 is about @xmath117 % and the one between the inter - valley spin diffusion length with the orders @xmath116 and @xmath118 is about @xmath119 % . with the last term ( i.e. , the negligible one ) in @xmath12 [ see eq .
( [ socspatial ] ) ] neglected , the spin precession frequency in the spatial domain in monolayer mos@xmath0 is given by @xmath120 , which is similar to the one provided by the external magnetic field@xcite [ i.e. , @xmath13 ] . | electron spin diffusion is investigated in monolayer mos@xmath0 in the absence of external electric and magnetic fields .
the electron - impurity scattering , which is shown to play a negligible role in spin relaxation in time domain in this material , has a marked effect on the in - plane spin diffusion due to the anisotropic spin precession frequency in the spatial domain . with the electron - impurity and inter - valley electron - phonon scatterings separately included in the scattering term
, we study the intra- and inter - valley diffusion processes of the in - plane spins by analytically solving the kinetic spin bloch equations . the intra - valley process is found to be dominant in the in - plane spin diffusion , in contrast to the case of spin relaxation in time domain , where the inter - valley process can be comparable to or even more important than the intra - valley one . for the intra - valley process , we find that the in - plane spin diffusion is suppressed with the increase of impurity density but effectively enhanced by increasing electron density in both the degenerate and nondegenerate limits .
we also take into account the electron - electron coulomb scattering in the intra - valley process .
interestingly , we find that in the nondegenerate limit , the intra - valley spin diffusion length presents an opposite trend in the electron density dependence compared to the one with only electron - impurity scattering . |
the nature of fluctuations in both the metallic state @xcite,@xcite , and in disordered insulators @xcite , @xcite , has been a matter of interest for both theoretical and experimental studies .
whereas in the metallic regime the basic aspects of fluctuations have been elucidated , in the regime of hopping transport the nature of fluctuations is still an open field .
the deep insulating regime , dir , where transport occurs via variable range hopping ( vrh ) , is defined as the regime where the localization length is the smallest scale compared to the elastic mean free path and hopping lengths , i.e. , @xmath2 respectively@xcite .
coherence effects are possible in this regime because phase breaking events occur at the hopping length@xcite , which is larger than @xmath3 .
important signatures of quantum interference in disordered insulators are the classic magneto - fingerprints , or reproducible fluctuations in the conductance with magnetic field , and a low field positive magneto - conductance .
an important property of mesoscopic conductance fluctuations in the metallic phase is their ergodicity . at the mesoscopic level ,
the sample size is less than thermal diffusion length or the dephasing length , whatever is shorter , such that sample to sample fluctuations are visible and the system does not self - average .
although it was not rigorously proven , the ergodic hypothesis was meant as the ability of the magnetic field ( or energy ) to induce conductance fluctuations equivalent to sample to sample fluctuations ( lee - stone criterion ) @xcite . in contrast , experimental results show that log - conductance mesoscopic fluctuations in the dir without spin - orbit scattering are not ergodic in the lee - stone sense@xcite , i.e. , the variance over samples is larger than the variance over field .
such samples involve hopping lengths that are , at most , 6 to 10 times the localization length .
precise measurements of ladieu et al@xcite and orlov et al@xcite have shown that a ) field fluctuations do not decorrelate disorder fluctuations , b ) field fluctuations do not change the identity of the hop , c ) the field average of the variance over the samples is larger than the sample average of the variance over the field and d ) there exist a decorrelation field @xmath4 defined by the field correlation function , which defines a equivalent new sample .
the question of the ergodic nature of fluctuations with and without so has not been addressed , to our knowledge , from the theoretical side .
our plan of this work is first to address the problem of fluctuations and the question of ergodicity in dir within the nss model .
we undertake this task through the verification of concepts concerning ergodicity , which we first define with mathematical rigor in the next mostly technical section .
then in section [ sec : nssmodel ] we explain the nss model and define the random processes we want to analyze . in section[sec : fluctuations ] we carry out the program described in section [ sec : ergodicconcepts ] and verify the non - ergodic behavior of fluctuations . in section [ sec : corrcoopdiff ]
we define the main theoretical objects of this work , the cooperon and diffuson analogs of weak localization theory , with the help of which , we can explain to some degree the decaying behavior of the correlation function .
finally we conclude by discussing and comparing our results with experiments .
in the preceding section one we introduced the question of ergodicity of fluctuations . here we introduce the mathematical concepts that will permit us to establish the ergodic nature of the fluctuations . given a physical quantity @xmath5 depending on the disordered hamiltonian @xmath6 and magnetic field @xmath7 , we denote by @xmath8 the sample to sample average , or disorder average , and by @xmath9 the field average for a given sample or disorder realization . in order to be able to estimate the sample average from the field average of a given sample
the following conditions must be satisfied : a ) @xmath10 @xmath11^{2}}\rightarrow 0 $ ] and b ) @xmath12 .
the verification of both conditions is known as _ ergodicity in the mean square sense _ ( mss ) , or the random function @xmath13
is said to be ergodic in the mean - square limit@xcite .
the condition @xmath14 is a measure of global stationarity . which means that these averages are independent of @xmath7 .
one can cast conditions a ) and b ) into a single statement on the disorder fluctuations of the field average , i.e .
, @xmath15 , where @xmath16 means variance over disorder@xcite .
this property implies that for one realization of disorder there are enough equivalent samples within the magnetic scale , such that the average in the field , with regard to disorder , does not depend in statistical sense on the particular realization .
this means that @xmath17@xmath18 , where @xmath19 is the number of realizations , and a sharp distribution of @xmath20 over disorder holds .
one could also ask for the possibility of making estimates of @xmath21 from @xmath22 , ( here @xmath23 means the variance over the field b of @xmath24 ) , or more generally , try to make an estimate of another function of the basic process @xmath25 .
the necessary and sufficient conditions such that one can estimate @xmath26 from one realization of disorder with the field average @xmath27 ( the so called law of strong numbers ) , is given by slutski s theorem .
one can write : @xmath28@xmath29@xmath30 @xmath31 , with : @xmath32}\times \overline{[g(f({\cal h},b))-\overline{% % \ g(f({\cal h},b))}\ ] } \label{defcorrelationfunction}\ ] ] where @xmath33 is the correlation function for @xmath34 and @xmath35}$ ] .
one can easily realize that a strong decaying behavior of the correlation function with correlation lenght @xmath36 will be a sufficient condition for ergodicity in the mss , i.e. @xmath37 .
in fact , it is also a necessary condition
. usually , for the application of the theorem stationarity in the wide sense of @xmath38 can be assumed , i.e. , @xmath39 does not depend on @xmath7 and @xmath40should depends only on @xmath41@xcite . in this work
we are interested in testing the ergodicity in two cases : @xmath42 for the average and @xmath43for the variance .
the first case corresponds to the usual meaning of ergodicity in statistical mechanics .
the second case , usually named the lee - stone criterion , refers to the equivalence of sample to sample fluctuations and magnetic field fluctuations .
we examine now the fluctuations in the dir in two and three dimensions and define the random process @xmath44 .
this is obtained from the nguyen , spivak and shklovskii@xcite model ( nss ) .
the nss model s crucial insight is that coherence is maintained within a mott hopping length , where the conductance is a sum of coherent _ forward directed feynman paths _ which interfere which each other .
the nss model describes the quantum behavior of the critical ( bottleneck ) hop in the miller - abrahams network@xcite .
the existence of many randomly oriented critical hops tend to average the macroscopic conductance , eliminating fluctuations@xcite . here
, we focus on the low temperature regime where critical hops do not trivially self average@xcite@xcite , i.e. , the percolation correlation length @xmath45 is such that @xmath46 , where @xmath47 is the percolation correlation length exponent , @xmath48 is the spatial dimension and @xmath49 a disorder parameter .
this is the mesoscopic regime @xcite .
we will first find the fields above which the process @xmath50 can be considered stationary and test the ergodicity criterion in the mean square limit .
we then calculate the correlations functions and their decaying behavior .
invoking slutski s theorem , we confirm non - ergodic behavior as established by the finiteness of the mean square criterion , ( @xmath51 @xmath52 ) .
we then proceed to test the ergodicity of fluctuations .
non - ergodicity of fluctuations is implied by the weakly decaying behavior of the correlation function .
we find this consistently . here
, we find good quantitative agreement with the measurements of orlov et al @xcite .
we derive the cooperon and diffuson analogs of the weak localization theory for dir with the help of which the non ergodic behavior can be explained .
furthermore , the predictions for the case with so are made . in the two dimensional nss model ,
impurities are placed on the sites of a lattice of main diagonal length @xmath53 ( the hopping length , @xmath54 mott s law ) .
we apply a magnetic field @xmath7 , perpendicular to the plane , changing only the phases of the electron paths .
the overall tunneling amplitude is computed by summing all forward directed paths between two diagonally opposed points , each contributing an appropriate quantum mechanical complex @xmath55 matrix weight given by the hamiltonian : @xmath56 where @xmath57 is the site energy , and @xmath58 represents the nearest neighbor couplings or transfer terms which includes a randomly chosen su(2 ) matrix describing a spin rotation due to strong so scattering . within the nss model , we choose site energies to be @xmath59 with equal probability @xcite @xcite . without so the coupling terms are diagonal in spin space @xmath60 and the green s function between the initial and final site is given by @xmath61 , \label{enss}\ ] ] where @xmath62 is the phase gained through path @xmath63 due to the magnetic vector potential , @xmath64 represents all directed paths that go from @xmath65 to @xmath66 through the lattice and @xmath67@xcite . in the presence of spinorbit scattering the green s function is the @xmath68 matrix : @xmath69[\prod_{i_{\gamma ^{\prime } } } u_{i_{\gamma ^{\prime } } } ] e^{\sum{i\phi_{i_{\gamma^{\prime}}}}}\ , \label{enssso}\ ] ] the green s function consist of sum of terms , one for each path , each being a product of random numbers and random su(2 ) matrices , and a deterministic disorder independent phase factor from the magnetic vector potential@xcite .
the complex function @xmath70 ( a complex matrix function in the presence of so ) contains the interference information including correlations due to crossing of paths , and the factor @xmath71 is the leading contribution to the exponential decay of the localized wavefunction .
we use the transfer matrix approach in order to compute @xmath72 , exactly , for each realization of disorder@xcite .
our random processes in question represent the log - conductance .
they are @xmath73 and with so @xmath74where @xmath75@xcite . in his work
we measure the magnetic field @xmath7 or changes in magnetic field @xmath41 in flux units @xmath76 , where @xmath77 is the flux quantum .
figure [ fig1 ] shows typical fluctuations of the log - conductance as a function of the sample and the magnetic field . without so
, the figure clearly shows that the average of @xmath78 dominates the fluctuations , i.e. , the average behavior is visible for a single sample .
as the average @xmath79 first increases proportional to @xmath7 , crossing over to a slower growth as @xmath80 dictated by the magnetic length @xmath81 @xcite , the process is not stationary .
however , in the latter regime of slow growth , one finds a field above which the process can be considered as essentially stationary in the same fashion as in the metallic regime . in this regime , while the log - conductance tends to saturate , the fluctuations persist as in mesoscopic fluctuation theory in metals@xcite .
furthermore , we note that the average behavior is periodic in half the flux quantum @xmath82 per @xmath83 ( only one half of a period is shown ) .
this periodicity reveals an average field coupling to @xmath84@xcite which has been demonstrated theoretically@xcite . in three dimensions ,
the fluctuations are appreciably larger than the average behavior .
once more , persistent fluctuations beyond the average conductance saturation field are observed .
the existence of such persistent fluctuations were first surmised by sivan et al@xcite and zhao et al@xcite .
with so , there is no tendency to build an average , and there are in general soft changes in the fluctuations in marked contrast with the sharp changes without so .
this peculiarity will be explained later .
one marked feature of figure 1(with and without so ) , are the fact that disorder fluctuations do not decorrelate the field fluctuations , which is the reason of the remarkable similarity with fig.2 of the work of @xcite .
the key point in both figures is that the the fluctuations do no decorrelate at the scale of fields shown in the figures , suggesting non - ergodic behavior .
following the concepts of section two , we analyze the ergodicity of the log - conductance fluctuations more carefully . to achieve this
we have to check two points : first if stationarity is reasonably fulfilled and second , we have to verify the ergodicity condition and the decaying behavior of the correlation function .
fig.[fig2]a shows the quantities @xmath85(stars ) and @xmath86^{2}}$ ] ( diamonds ) in the absence of so ( @xmath87 ) .
the field averaging interval is @xmath88 $ ] . in this figure as the field @xmath89 is increased from zero to 0.09 in @xmath90 $ ] units , the diamonds move downward overlapping the stars sooner for smaller t. therefore , the field above which the process can be considered quasi stationary increases with the hopping length t such that both quantities tend to coincide for larger @xmath91 .
this feature is of importance for calculating other quantities , as any question on ergodicity presumes at least quasi - stationarity . in the case of so ( @xmath92 ) , fig.[fig2]b shows effective stationarity of the process @xmath93 essentially independent of @xmath94 and @xmath53 . @xmath95 does not tend to zero with increasing @xmath96 ; on the contrary , it increases with a power of t@xcite , which is an indicator of non - ergodic behavior as there is no self - averaging as one increases the hopping length t ( see kramer and mackinnon @xcite ) .
the last result establishes non - ergodic behavior in the mean square sense with and without so . to further substantiate this result
we calculate the correlation functions .
[ fig3]a shows the correlation function for three values of the hopping length @xmath53 and some values for @xmath94 . one sees a very weakly decaying behavior on the physical field scale ( @xmath76 ) and a tendency to decay faster for bigger @xmath91 , indicating that when the process becomes quasi stationary there is a tendency to a faster decaying correlation function .
[ fig3]b with so shows that the correlations function depends essentially only on @xmath97 and the decaying is even weaker with a functional form depending on t. now , the basic argument against ergodicity is that the decaying behavior of the correlation functions is such that it is not possible to construct enough ensembles from the field fluctuations data and therefore non - ergodic behavior is established . to illustrate this point
, we write @xmath98 $ ] . in order to have enough emsembles in the field scale within the validity of the model ( @xmath99 @xmath100 in our units )
, there should be a decorrelation field @xmath4 @xmath101 such that large number of samples could be defined , so that one would get @xmath102-could be satisfied .
this , on the other hand , implies that the condition for ergodicity of the variance , which in general requires a stronger decaying behavior of the correlation function for process @xmath103 than the condition required for ergodicity in the mean square sense , is not at all fulfilled@xcite .
therefore , one should expect that the lee and stone criterion of ergodicity on the relative magnitude of the field and sample fluctuations is not realized . to check this point
one has to compare the magnitude of the variance in field and sample to sample fluctuations .
the idea of further averaging over disorder and over the field , respectively , is the same as in statistical mechanics using different initial conditions to improve the statistics .
figure [ fig4 ] shows the averages @xmath104 @xmath105 without so we find the characteristic @xmath91dependence shown in fig.[fig2 ] shows up as a crossing of both types of averages .
there is again a clear tendency to saturate as @xmath91increases whereas the difference of the saturated averages widens with increasing t. this tendency is clearly seen in the case with so where there is a much weaker dependence on @xmath91 as expected from fig.[fig2 ] and the crossing observed in the previous case is absent .
in this section we develop the concepts of the cooperon and diffuson in the context of strong localization .
these objects are used to explain qualitatively and semi - quantitatively non - ergodic behavior in the mean square sense and the relative magnitude of the field and sample fluctuations found in experiments . for this purpose ,
the following relation is straightforwardly derived : @xmath106={\rm var}_{d}[f(b+\delta b , t)]+{\rm var}% _ { d}[f(b , t)]+2c(f({\cal h},b,\delta b , t ) ) \label{nonlinear}\ ] ] this relation is valid for both processes @xmath107 and @xmath108@xcite . for the sake of clarity , the composite process inside the brackets on the left side of eqn.[nonlinear ]
is denoted by the process @xmath109\ $ ] and with so by @xmath110 $ ] , such that the left hand side of eqn .
[ nonlinear ] is in each case @xmath111 $ ] and @xmath112 $ ] , respectively .
these two functions are shown in fig .
[ fig5 ] as a function of @xmath113 in @xmath114 units and three values of t @xmath115 from the bottom to the top , respectively ) . for @xmath91 shown in the figure and the scales of @xmath116used ,
three things deserve explanations : first for large enough @xmath117 , the above mentioned functions show almost no dependence on @xmath91and @xmath117 at fixed t ; secondly , the ratio @xmath118/$ ] @xmath119\ $ ] is around two , for big enough t and @xmath120 which is a landmark of the symmetry changing from unitary to simpletic ; thirdly , one observes a very rapid decaying on the dependence on @xmath7 , which can be traced to the saturation behavior of the cooperon as we will see . recall that on the metallic side , an analogous behavior has been described which identifies two fundamental contributions to the field effect : the cooperon and the diffuson@xcite@xcite .
these contributions can be distinguished by the way they enclose the magnetic flux ; while the cooperon is sensitive to ( @xmath121 , the diffuson only responds to field changes @xmath41 . in the insulating regime , a mechanism similar to the cooperon which saturates is associated with a positive magneto - conductance ( mc )
this has been observed as a general effect@xcite@xcite .
a semi quantitative explanation for the behavior of the functions @xmath122\ $ ] and @xmath123 $ ] and their ratio can be found with the help of the cooperon and diffuson analogs . to achieve this goal
, we consider the moments of process of @xmath124 and @xmath125 ( further below it will become clear why ) . in the former case
they are given by @xmath126^{n}}$ ] .
recall from equations [ enss ] and [ enssso ] that this product can be visualized as a set of n paths , each one defined by the respective term in the product .
these paths eventually intersect each other at some values of their length . in order to have nonzero contributions after disorder average
, the paths must pair up , as a consequence of th chosen distribution of the energies @xcite .
neutral paths ( field independent ) are formed by pairing @xmath127 and @xmath128 at the same field ( phase cancels ) . on the other hand ,
charged paths ( field sensible ) are formed by pairing either @xmath129 and @xmath130 or @xmath131 and @xmath132 . in the absence of paired path intersections ,
self interference kills charged paths ( their contribution decays exponentially fast ) .
nevertheless , if intersections are considered , one can have path exchanges for short distances , yielding a magnetic field coupling which is analogous to the magneto - conductance , the source of the initial decaying of the correlation function .
there are three possible diagrams at a paired path crossing , two of which are depicted in fig.7 . without so the spin indexes can be ignored ,
one obtains @xcite : a ) one partner from @xmath131 pairs with one from @xmath132 while one from @xmath133 and one from @xmath130 follow a different path .
such a combination encloses @xmath134 and is therefore called _ cooperon_-like .
b ) one partner is taken from @xmath129 and the other from @xmath130 on the same path , while one from @xmath133 and @xmath132 follow another .
such a combination encloses only @xmath41 and is called _ diffuson_-like .
finally , one can have combination c ) where one partner comes from @xmath131 and the other from @xmath133 , leaving @xmath132 and @xmath130 to pair up .
the latter combination is called uncharged and encloses no field .
note that all previous cases satisfy overall neutrality so that the contributions are real as expected .
the contribution of the replica cooperon and diffuson are the same at zero field and there is an additional contribution from the uncharged diagram .
further progress is achieved using the replica argument .
the replica - moment argument @xcite , maps the @xmath135-th moment problem onto the problem of @xmath136 bosons with contact interaction .
these interactions renormalize due to the diagrams above , making path interactions field dependent .
the @xmath136 boson system can be solved using the bethe ansatz and has ground state energy @xmath137 , where @xmath138 is a function of @xmath139 , such that one has @xmath126^{n}}=$]@xmath140 @xmath141 valid at fixed @xmath135 asymptotically for @xmath142 . on the other hand ,
the @xmath135-th moment can be expressed as a cumulant expansion valid at fixed @xmath53 asymptotically for @xmath143 , @xmath144^{n}}$ ] @xmath145\}$ ] , where @xmath146 $ ] are the cumulants of process p .
the subtleties concerning both limits have been discussed by kardar @xcite , who finds a nonextensive correction subleading term proportional to @xmath147 .
we therefore obtain : @xmath148=(\rho _ { coop}(2b+\delta b)+\rho _ { diff}(\delta b))t^{2/3}+\ln a(b,\delta b ) \label{cooperondiff}\ ] ] here @xmath149 @xmath150 , we have separated the path interaction in terms of the cooperon and diffuson contributions .
we check numerically this important prediction and fig.6 shows the expected scaling with an exponent near @xmath151 . beyond the saturation field of the average log - conductance , the cooperon term on the right hand side saturates ( the same for the variance on the right of eqn .
[ nonlinear ] which depend on @xmath7 ) and the correlation function only depends on @xmath41 .
this behavior is summarized in fig[fig5 ] . in the case of so , it can be shown@xcite , that the only non - zero paired averages are @xmath152 , @xmath153 , @xmath154 , thus so averaging brings a factor of @xmath155 and forces the neutral paths to have parallel spins while the spin of the two partners of charged paths must be antiparallel . as a consequence , one finds the cooperon diagrams cancel in pairs as concluded for the case of the magneto- conductance@xcite(fig.7 ) such that no exponential corrections to the conductance occurs due to the cooperon . for the diffuson
there are only two combinations possible for incoming spin indexes(all up and all down ) , such that in this case @xmath156=$]@xmath157 @xmath158 .
this nice result explains why there are fluctuations with so , even if there are not exponential corrections to the magneto - conductance .
we find again numerically the predicted @xmath147 scaling ( see fig.6 ) .
for large @xmath117 and t one finds the ratio of the variances approaches @xmath159 ( i.e. , 1/(2 times @xmath160 ) ) , in agreement with the numerical results . now from eqn .
[ nonlinear ] one obtains : @xmath161t^{2/3}+s(b,\delta b)\ } \label{correqt}\ ] ] where @xmath162 , are logarithmic corrections from the prefactors .
@xmath163 defines the magneto - conductance , @xmath164^{n}}=$]@xmath165\}=$]@xmath166 @xcite . from eqn.[correqt
] apart from the predicted @xmath147 scaling , one can gain qualitative and quantitative understanding of the decaying behavior of the correlation function for small @xmath41 .
with the explicit field dependence of the cooperon and diffuson given by : @xmath167[fluctuations ] and @xmath168\rho(b=0)t^{2/3}\]][magnectoconductance ] one obtains : @xmath169{\rm var}_{d}[f(0,t]t^{2/3}}\]][correqt2 ] here , we have ignored the logarithmic corrections from eqn.[correqt ] . for @xmath170 and @xmath171 one has @xmath172 $ ] , with @xmath173=\rho(b=0)t^{2/3}$ ] , in accordance with eqn.[nonlinear ] .
now , for small @xmath41 and @xmath174 , one can assume that @xmath175@xcite : @xmath176{\rm var}_{d}[f(0,t]t^{2/3 } \label{correqt3}\ ] ] this behavior is is qualitatively seen even for @xmath177 in fig.[fig3 ] , where for given @xmath7 one observes a faster decay with increasing @xmath53 ( recall that @xmath178 $ ] ) .
this is even more clearly to see with so . in that case one
has : @xmath179t^{2/3}\label{correqtso}\ ] ] @xmath180 - 2)\rho_{mc}^{spinor}t^{2/3}\label{correqtso1}\ ] ] again an expansion in small @xmath41 : @xmath181 on the other hand , without so , numerical calculations ( fig.[fig3 ] ) , shows a very slow decaying , such that for the field range less than @xmath182 , there are no enough ensembles to form , such that even if the field correlation function decays , and a decorrelation field can be defined , it is not possible to define a large enough number of samples such that averaging over the field , could be equivalent to averaging over disorder ( samples ) , a matter that has caused some confusion . with so the decaying of the correlation function is dominated by an even much bigger field scale unknown to us .
although the replica results are valid quantitatively for @xmath183 dimensions , nevertheless , @xmath184 dimensions also has a bound state , although weaker , with a smaller positive mc . therefore , our results apply qualitatively to three dimensional hops@xcite .
summarizing the results of this work , we have found that the log - conductance in the dir of mesoscopic samples is non - ergodic in the mean square sense and the fluctuations of the log - conductance are non - ergodic in the sense that sample to sample fluctuations are larger than magnetic field fluctuations . without so due to the ever increasing magneto - conductance with the field@xcite , only quasi stationarity can be achieved for fields larger than a certain hopp dependent field @xmath185 , whereas with so due to the small magneto - conductance and its rapid saturation@xcite , @xmath185 is negligible , so that the process in that case can be regarded as stationary independent of t. the field @xmath185 is obtained numerically by varying @xmath94 until one observes that @xmath186 . this different behavior with and without
so turns out to have a remarkable influence on the behavior of the two variances that define the sample to sample or disorder fluctuations and the field fluctuations , as their values depend on @xmath94 .
it is only when the averaging interval is taken as @xmath187$],i.e .
, when quasi - stationarity is achieved , that the proper comparison of the variance can be made . in experiments the hopp values were small , and therefore @xmath185 , was small .
however , in order to compare the magnitude of the fluctuations defined by eqn.[vardis ] and eqn.[varfield ] with experiments one must be aware that their values are sensitive to [ @xmath185,@xmath96 ] , an important point indicated by our results .
the experiments of @xcite were done without so ; therefore , in order to make a a rigorous comparison with our results we have to know the experimentally taken interval [ @xmath185,@xmath96 ] for the evaluation of this quantities .
for the small values of t , according to our results , the influence of @xmath94 is small and one should expect @xmath189 and indeed we find the ratio of eqn.[vardis ] and eqn.[varfield ] predicts a value similar to their experiments . on the other hand ,
the scaling of the disorder fluctuations with the hopp length t , as shown in fig.4 , agrees with the experimental values obtained by orlov et tal@xcite(see fig.45 in the review of kramer and mackinnon ) .
it is interesting to observe that both the variance with disorder and the one with field appears to have the same functional form .
a more careful comparison can be done by taking into account all predictions of this work . in the case with so
, experimental studies of fluctuations are lacking . to our knowledge
the only work where field fluctuations in the strongly insulating regime have been observed in samples with so is the work of hernandez and sanquer @xcite . here
our predictions give a clear qualitative way of differentiating the cases with and without so and highlight a way of evaluation of experimental data . from the theoretical point of view
, we have derived two important objects : the @xmath190 and the @xmath191 which are the weak localization analogs in the dir .
they allow us to explain some features of the correlation function like the qualitative behavior for small @xmath41 , and numerically the weak decaying behavior on the scale @xmath192 , that according to the slutski s theorem , is responsible for the non - ergodic behavior as we have found . in order to experimentally observe persistent diffuson fluctuations
, one has to explore a range of parameters so that there is a saturation in the average behavior while the wave function shrinkage@xcite is still a negligible effect .
this range can be defined by the condition @xmath193 , where @xmath194 is the bohr radius .
@xmath195 is the scale for the orbital shrinkage to be important@xcite , i.e. , when the cyclotron radius becomes of the order of the mean free path @xmath196 .
these conditions have been met in ref .
@xcite and @xcite . furthermore , according to references@xcite the magnetic field can not induce geometric fluctuations due to changes in the identity of the hop@xcite .
this finding holds in both two and three dimensions , and therefore we expect the insulating cooperon and diffuson fluctuations should be seen experimentally also in three dimensions . in experiments , one should be aware about the condition of mesoscopic sample , discussed in section [ sec : nssmodel ] .
otherwise , trivial self - averaging of spatially different oriented hops can wash out the possibility to observe fluctuations
. there will be a non - ergodic to ergodic behavior in the case when one relaxes the condition of dir with @xmath197 such that there are many impurities within @xmath198 . in this case
there will be a diffusing behavior within the length scale @xmath198 so that two overlapping random processes are at work , one ergodic for the diffusing scale @xmath199 and the other non - ergodic for the larger scale @xmath53 .
this question is under study @xcite .
finally , we want to stress that the fundamental point of considering correlations in the random processes @xmath200 due to the crossing of paths can not be relaxed .
this approach permitted us to use the full power of the replica theory , which again helped to semi - quantitative and qualitative predictions .
theories like the independent path approximation , where such crossing is neglected , miss the very crucial objects defined by the cooperon and the diffusson .
one of us ( rr ) wants to express his gratitude to the head of department of ica1 , prof .
hans j. herrmann , for the hospitality and support .
also ( rr ) one to express his gratitude to dr .
maheboob alam for helping me in technical details with xmgr and making helpful corrections in the written english
. author correspondence email : @[email protected] [ sec : agradecimientos ] | we give a detailed and rigorous picture of the mesoscopic conductance fluctuations in the deep insulating regime ( dir ) within the nguyen , spivak and shklovskii model including spin - orbit coupling ( so ) .
without so , we find that fluctuations of the log - conductance are persistent above a saturation field @xmath0 , where one has that the log - conductance is approximately a stationary random process .
in contrast , in the so case the saturation field @xmath1 is negligible and the stationarity is well realized .
we find non - vanishing disorder fluctuations of the field average of the log - conductance as a quantitative measure of the lack of ergodicity in the mean square sense . to this fact
, a weak decaying behavior of the correlation function , even weaker in the case of so is established on the relevant field scale of the model , one flux quantum per plaquette .
this finding corroborate the behavior of the fluctuations of the field average of the log - conductance and permit us to invoke slutski s theorem to conclude that the whole stochastic process defined by the log - conductance is non - ergodic in the mean square sense in both cases . as a consequence the commonly used criterion to test the ergodicity based on the equivalence of the variance in disorder and the variance in the field is not fulfilled . using the replica approach , we derive the weak localization analogs of the cooperon and diffuson which permits us to analyze in qualitative form the decaying behavior of the correlation function .
our predictions agree qualitatively and semi - quantitative with experiments in the dir . |
scene flow is a three - dimensional motion field of the surface in world space , or in other words , it shows the three - dimensional displacement vector of each surface point between two frames . as most computer vision issues are , scene flow estimation is essentially an ill - posed energy minimization problem with three unknowns .
prior knowledge in multiple aspects is required to make the energy function solvable with just a few pairs of images .
hence , it s essential to fully make use of information from the data source and to weigh different prior knowledge for a better performance .
the paper attempts to reveal clues by providing a comprehensive literature survey in this field .
scene flow is first introduced by vedula in 1999 @xcite and has made constant progress over the years .
diverse data sources has emerged thus scene flow estimation do nt need to set up the complicated array of cameras .
the conventional framework derived from optical flow field @xcite has extended to this three - dimensional motion field estimation task , while diverse ideas and optimization manners has improved the performance noticeably .
the widely concerned learning based method has been utilized for scene flow estimation @xcite , which brings fresh blood to this integrated field . moreover , a few methods have achieved real - time estimation with gpu implementation at the qvga(@xmath0 ) resolution @xcite , which insure a promising efficiency .
the emergence of these methods stands for the fact that scene flow estimation will be widely utilized and applied to practice soon in the near future .
the paper is organized as follows .
section [ sec : background ] illustrates the relevant issues , challenges and applications of scene flow as a background .
section [ sec : taxonomy ] provides classification of scene flow in terms of three major components . emerging datasets that are publicly available and
the diverse evaluation protocols are presented and analyzed in section [ sec : evaluation ] .
section [ sec : discussion ] arises few questions to briefly discuss the content mentioned above , and the future vision is provided .
finally , a conclusion is presented in section [ sec : conclusion ] .
we provide relevant issues , major challenges and applications as the background information for better understanding this field .
scene flow estimation is an integrated task , which is relevant to multiple issues .
firstly , optical flow is the projection of scene flow onto an image plane , which is the basis of scene flow and has made steady progress over the years . the basic framework and
innovations of scene flow estimation mainly derives from optical flow estimation field .
secondly , in a binocular setting , scene flow can be simply acquired by coupling stereo and optical flow , which makes the stereo matching an essential part for scene flow estimation .
most scene flow estimation methods with promising performance are initialized with a robust optical flow method or a stereo matching method . and
the innovation in scene flow mostly derived from these two fields .
hence , we provides the changes and trend in the relevant issues as heuristic information .
optical flow is a two - dimensional motion field .
the global variational horn - schunck(h - s ) method and the local total - least - square(tls ) lucas - kanade(l - k ) method have led the optical flow field and scene flow field over the years @xcite .
early works was studied and categorized by barron and otte with quantitative evaluation models @xcite .
afterwards , brox implemented the coarse - to - fine strategy to deal with large displacement @xcite , while sun studied the statistics of optical flow methods to find the best way for modeling @xcite .
baker proposed a thorough taxonomy of current optical flow methods and introduced the middlebury dataset for evaluation @xcite , and comparisons between error evaluation methodologies , statistics and datasets are presented as well .
currently , optical flow estimation has reached to a promising status . a segmentation - based method with the approximate nearest neighbor field to handle large displacement ranks the top of middlebury dataset in terms of both endpoint error(epe ) and average angular error(aae )
currently @xcite , where epe varies from 0.07@xmath1 to 0.41@xmath1 in different data and aae varies from 0.99@xmath2 to 2.39@xmath2 .
a similar method reached promising results as well @xcite .
moreover , there are a variety of methods which achieve top - tier performance and solve different problems respectively .
rushwan utilized a tensor voting method to preserve discontinuity @xcite .
xu introduced a novel extended coarse - to - fine optimization framework for large displacement @xcite , while stoll combines the feature matching method with variational estimation to keep small displacement area from being compromised @xcite .
he also introduced a multi - frame method utilizing trilateral filter @xcite . to handle non - rigid optical flow , li proposed a laplacian mesh energy formula which combines both laplacian deformation and mesh deformation @xcite .
stereo matching is essential to scene flow estimation under binocular setting .
a stereo algorithm generally consists of four parts:(1 ) matching cost computation , ( 2 ) cost aggregation , ( 3 ) estimation and optimization and ( 4 ) refinement .
it is categorized into local methods and global methods depending on how the cost aggregation and computation are performed .
local methods suffer from the textureless region , while global methods is computationally expensive .
a semi - global - matching(sgm ) method combines local smoothness and global pixel - wise estimation and leads to a dense matching result at low runtime @xcite , which is commonly utilized as the modification .
a comprehensive review is presented by scharstein in 2001 @xcite .
the upper rank algorithms of middlebury stereo dataset and kitti stereo dataset @xcite are mainly occupied by unpublished papers , indicating the rapid development in this field .
learning methods are utilized with promising efficiency and accuracy @xcite . besides
, zhang proposed a mesh - based approach considering the high speed of rendering and ranks the top among the published papers @xcite , while segmentation - based methods are proven to tackle the textureless problem @xcite .
the complex scene and the limited image capture approach post challenges in diverse ways , which are discussed as follows .
what happens when an unstoppable force meets an immovable object ?
the relationship between accuracy and efficiency is just like the omnipotence paradox . to achieve better accuracy ,
sufficient and complicated prior knowledge is obliged , while in terms of efficiency , the data need to be listed down to a reasonable scale and the calculation scheme should be as simple as possible .
we can not only consider the enhancement of efficiency and accuracy , but also value the trade - off between these two .
the trade - off is discussed in section [ sec : discussion ] , and the performance is illustrated in figure [ fig : accuracy - efficiency ] .
occlusion is common in a complex scene with multiple moving objects .
it occurs between views and frames as figure [ fig : occlusion ] illustrates .
it violates the data consistency assumption and may lead to mismatching on account of missing information of the occluded object .
besides , occlusion may perturb the consistency between frames and affect the multi - frames tracking method . [ htbp ] the occlusion between views can be handled well under the multi - view stereopsis with abundant prior knowledge , while temporal constraint may provide robust temporal coherence and prediction to alleviate occlusion between frames .
large displacement occurs frequently when an object is moving at a high speed or under a limited frame - rate .
moreover , articulated motion may lead to large displacement as well .
this kind of problem is hard to tackle on account that the scene flow algorithms normally assume the constancy and smoothness within a small region , large displacement may make the solution to energy function trapped into a local minimum which leads to enormous errors propagated by iteration procedure .
brox implemented the coarse - to - fine method along with a gradient constancy assumption to alleviate the impact caused by large displacement in the optical flow field @xcite .
currently , several matching algorithms have been introduced to handle this issue specifically and achieved promising results @xcite .
brightness constancy does nt obey the illumination - varying environment .
however , this issue is common in an outdoor scene , e.g. , drifting clouds that block the sunlight , sudden reflection from a window , and lens flares .
furthermore , it will be a disaster at night when lights start to flash .
in the optical flow field , additional assumptions such as gradient constancy and some more complicated constraints have been added to make it more robust to the illumination changes @xcite .
schuchert specifically studied range flow estimation under varying illumination @xcite . in his paper , pre - filtering and changes of brightness model
improve the accuracy .
gotardo introduced an albedo consistency assumption as a revision @xcite .
a relighting procedure was proposed as a key element to handle the multiplexed situation in his paper as well .
the lack of texture may make the scene flow estimation still an ill - posed problem , which is a challenge for discovering consistency .
it is also a challenge to stereo matching , which may lead to enormous errors in the binocular - based scene flow estimation .
the textureless region is still a major distribution of the estimation error . to overcome this problem ,
different scene representations have been utilized .
for example , popham introduced a patch - based method @xcite .
the motion of each patch does nt only rely on the texture information , but utilizes the motion from neighbor patches .
this makes it more robust for a textureless region . as a solution to the occlusion issue ,
segmentation - based method is valid because it assumes uniform motion among the small regions to deal with the ambiguousness @xcite .
scene flow estimation is a comprehensive problem .
motion information reveals the temporal coherence between two moments . in a long sequence
, scene flow can be utilized to get the initial value for the next frame and serve as a constraint in its relevant issue fields .
scene flow can not only profit from its relevant issues , but also facilitate them mutually .
gotardo captured three - dimensional scene flow to provide delicate geometric details @xcite , while liu utilized scene flow as a soft constraint for stereo matching and a prediction for next frame disparity estimation @xcite .
ghuffar combined local estimation and global regularization in a tls framework and utilized scene flow for segmentation and trajectory generation @xcite . beyond that
, scene flow can be a valuable input or mobile robotics and autonomous driving field , which consist of multiple task such as obstacle avoidance and scene understanding .
frank first fused optical flow and stereo by means of kalman filter for obstacle avoidance @xcite .
alcantarilla combined scene flow estimation with the visual slam to enhance the robustness and accuracy @xcite .
herbst got object segmentation with the rgb - d scene flow estimation result @xcite , aiming to achieve autonomous exploration of indoor scenes .
menze utilized scene flow to reason objects by regarding the scene as a set of rigid objects @xcite .
autonomous driving could make use of both the geometric information that represents distance and the scene flow information that represents motion for multiple tasks . in addition
, scene flow can be utilized to serve as a feature as the histogram of optical flow(hof ) @xcite or the motion boundary histogram(mbh ) @xcite feature descriptors for object detection and recognition , e.g. , facial expression , gesture , and body motion recognition .
it may enrich the information in the descriptor with additional depth dimension and can be applied for motion like rotation or dolly moves that optical flow ca nt handle .
for instance , in 2009 , furukawa recorded the motion model of the facial expression using scene flow estimation@xcite .
scene flow estimation is viewed as an ill - posed problem , which includes three main steps : data acquisition , energy function modeling , energy minimization and optimization .
the general taxonomy is depicted in figure [ fig : overall ] .
the energy function consists of data term and regularization as equation [ eq : energy ] illustrates .
@xmath3 data terms derive from different data sources assuming brightness constancy(bc ) or gradient constancy(gc ) as local constraint . while there are three unknown parameters ,
regularization terms need to be added to regularize the ill - posed problem and provide spatial coherence .
theoretically , the more regularization terms are , the more robust and accurate the estimation is .
however , miscellaneous regularization terms may lead to redundancy , intractability and over - fitting , and that s why the design and solution of regularization term are key to a method .
hence , in this section , existing methods are categorized in terms of three fundamental properties that distinguish the major algorithms : _ scene representation _ presented the diverse representations for both scene and scene flow .
_ data source _ describes the major data acquisition manner and the corresponding data term choices .
_ calculation scheme _ mainly discuss the idea for estimation and optimization manner , including diverse choices of regularization terms and implement . over the years , diverse ways to represent
the scene have emerged with different emphasis , which can be broadly categorized into _
depth / disparity _ , _ point cloud _ , _ mesh _ and
_ patch_. the convenient way to represent the scene is to couple color image and depth map as color - d information , where d stands for depth information with rgb - d data or disparity information under a binocular setting .
scene flow under this sort of representation is known as 2.5d scene flow or 2d parameterization of scene flow .
it consists of optical flow component which is measured in pixels , and disparity or depth change component which is measured in pixels or @xmath4 .
the binocular - based scene flow @xcite can be presented as @xmath5 , where @xmath6 is the 2d optical flow , and the @xmath7 stands for disparity change .
likewise , in terms of rgb - d scene flow @xcite , the motion field consists can be presented @xmath8 , where @xmath9 stands for the depth change .
particularly , disparity value can be converted into depth value as equation [ eq : disparitytodepth ] illustrates .
@xmath10 where @xmath11 is the focal length of the camera , and @xmath12 is the camera baseline value . to truly present the three - dimensional scene ,
the image pixel need to be projected into the scene space .
the projection is illustrated in figure [ fig:3d - mapping ] and presented in equation [ eq:3d - mapping ] .
@xmath13 where @xmath14 is the image pixel @xmath15 , @xmath16 is the projection @xmath17 from a pixel value @xmath14 and a depth value @xmath18 to a 3d point @xmath19 . @xmath20 and @xmath21 stands for the camera focal length , and @xmath22 and @xmath23 are the principle points .
formulation [ eq:3d - mapping ] can also be presented as : @xmath24 where matrix @xmath25 is known as the camera projection matrix .
hence , scene flow under point cloud representation @xcite can be presented as @xmath26 , which truly reveal the three - dimensional displacement .
meshes represent a surface as a set of planar polygons , e.g. , triangle , which connected to each other as is shown in figure [ fig : mesh ] .
this representation is a efficient way for rendering , and it occupies less memory .
mesh is essentially sort of point cloud representation as the vertex can be viewed as a point in the three - dimensional point cloud .
scene flow estimation methods with a mesh representation @xcite are only under a multi - view setting and the geometry estimation is given simultaneously .
the motion of vertice is solved with a point cloud methods , while the rest part are solved by interpolating along the meshes . in terms of patch representation , the surface
is represented by collections of small planar or sphere patches .
each patch is six - dimensional in terms of three - dimensional patch center position and three - dimensional patch direction .
patch representation is similar to mesh representation with different emphasis , where mesh representation focuses on the precision and deformable property of each vertex , and patch representation values the local consistency in terms of rigidity and motion within a small neighborhood region .
early paper viewed patch as a surface element(surfel ) under a multi - view setting@xcite . a few binocular - based scene flow methods utilized patches to fit the surface of the scene @xcite on account that this kind of patch - based methods are common in stereo matching field .
in addition , hornacek uniquely exploited a pair of rgb - d data to seek patch correspondences in the 3d world space and leads to dense body motion field including both translation and rotation @xcite . over the years
, scene flow has been estimated under three main kinds of data source : a calibrated multi - camera system , the binocular stereo camera or the rgb - d camera which consists of both rgb color information and depth information .
moreover , the emerging light field has been applied into the scene flow estimation with promising performance @xcite .
data terms under different data sources differs from each other .
hence , in this section , scene flow estimation methods are categorized into these four kinds : _ multi - view stereopsis _ , _ binocular setting _ , _ rgb - d data _ and _ light field data_.
most of the algorithms in the early 2000s assume a multi - view system , with multiple cameras set in a complex calibrated scene .
multi - view scene flow estimation is usually along with 3d geometry reconstruction simultaneously .
ample data sources and diverse prior knowledge ensure the robustness of the estimation , and occlusion issue can be handled well .
however , it is commonly at a high computational cost with an intricate full - view scene to deal with .
vedula proposed two choices for regularization and distinguished three scenarios in 1999 @xcite , which guides the multi - view scene flow estimation till now . in his paper ,
a multi - view scene flow can be estimated from one optical flow and the known surface geometry .
the equation is formulated in equation [ eq : vedula ] .
@xmath27 where @xmath19 is the three - dimensional scene point , @xmath15 is the two - dimensional image pixel , @xmath28 is the scene flow , @xmath29 is the optical flow , and @xmath30 is the inverse jacobian which can be estimated from the surface gradient @xmath31 .
afterwards , zhang proposed two systems for estimation @xcite , where ims assumed each small patch undergoes 3d affine motion , and egs used segmentation to keep the boundary .
these papers modeled energy function with multiple constraint , which provided a basic estimation process . similarly , pons presented a common variational framework with local similarity criteria constraint @xcite .
henceforth , different scene representations were introduced to describe the surface @xcite .
diverse multi - frame tracking methods mentioned for sparse estimation are utilized as well to build the temporal coherence @xcite .
moreover , letouzey added an rgb - d camera into the multi - view system with a mesh representation @xcite , aiming to enrich the geometry information with the depth data constraint .
binocular setting is regarded as a basic and simplified version of multi - view system , while the difference between the two is that binocular scene flow estimation is usually along with disparity estimation between two views but not the full - view 3d geometry knowledge .
the relevance between views and frames is illustrated in figure [ fig : binocular - a ] , and figure [ fig : binocular - b ] depicts the basic data terms in a binocular setting which consists of stereo consistency terms in time @xmath32 and @xmath33 along with optical flow consistency terms in both views .
the specific formulation is presented in equation [ eq : binoculardataterms ] .
@xmath34 where [ eq : binoculardataterms ] @xmath35 @xmath36 and @xmath37 are the optical flow consistency terms that assume the brightness of the same pixel stay constant between frames . similarly , @xmath38 and @xmath39 are the stereo consistency terms that assume brightness constancy between views . moreover , @xmath40 is the cross term to constrain the constancy between both frames and views .
most binocular - based methods fused stereo and optical flow estimation into a joint framework @xcite . on the contrary , others decoupled motion from disparity estimation to estimate scene flow with stereo matching method replaceable at will @xcite , and basha utilized a point cloud scene representation as a three - dimensional parametrization version of scene flow @xcite .
moreover , local rigidity prior was presented along with segmentation prior and achieved promising results @xcite .
specifically , valgaerts introduced a variational framework for scene flow estimation under an uncalibrated stereo setup by embedding an epipolar constraint @xcite , which makes it possible for scene flow estimation under two arbitrary cameras . in 2016 , richardt has made it a reality to compute dense scene flow from two handheld cameras with varying camera settings @xcite .
scene flow was estimated under a variational framework with a daisy descriptor @xcite for wide - baseline matching .
table [ tab : binoculardataterm ] enumerates some typical methods under a binocular setting with diverse choices of data terms .
most methods chose optical flow consistency terms in both views and stereo consistency terms in both time @xmath32 and time @xmath33 @xcite , and few methods only take parts of terms mentioned above @xcite .
cross term was utilized in @xcite .
moreover , huguet @xcite and hung @xcite utilized additional gradient constancy assumption besides intensity to enhance robustness against illumination changes , which turns the image intensity value @xmath41 in energy function into image gradient @xmath42 . additionally , extra rgb constancy terms @xmath43 and @xmath44 are taken in hung s paper as well , which extends gray value intensity into three - channel information . however , it is proposed that image gradient is sensitive to noise and is view dependent @xcite .
hence , the necessity of additional assumptions like gradient constancy remains further research to balance the pros and cons .
.typical methods under the binocular setting [ cols="<,<,<,<",options="header " , ] we calculated the statistics of the ground truth optical flow in terms of average magnitude and the max magnitude to show the challenging level and large displacement situation of each dataset , as figure [ fig : datasetstatistic ] presents . where the bigger the average magnitude and standard deviation are , the more challenging the dataset is , and the bigger the max magnitude is , the more a robust large displacement handling is required .
[ htbp ] we can see clearly that the up - to - date dataset is more challenging than datasets published before . along with information provided in table [ tab : dataset ] , we provide suggestion for datasets with different purpose as follows : + * comprehensiveness * taken multiple issues , e.g. , categories of ground truth data , image resolution , challenging level , naturalism , scale , popularity , into consideration , we recommended mpi sintel dataset and freiburg dataset for their comprehensive property . +
* challenging * flyingthings3d subset of freiburg dataset shows the characteristic of large displacement , complex occlusion and diverse changes between frames , which is really challenging for scene flow estimation .
monkaa subset shows the similar characteristic which is recommended as well .
+ * public popularity * middlebury , kitti(2012&2015 ) and mpi sintel dataset provide evaluation protocols and online ranking that can evaluate the performance of a method conveniently .
moreover , the evaluation can be compared with top tier methods in optical flow estimation field and stereo matching field to indicate the superiority of scene flow estimation . +
* multi - view and rgb - d data source * for multi - view stereopsis , basha rotating sphere and kitti2012/2015 provide multi - view extension for evaluation .
to be noted , basha provides only point cloud scene flow ground truth for quantitative analysis , while the ground truth of kitti2012/2015 is sparse . in terms of rgb - d scene flow estimation ,
mpi sintel dataset and basha rotating sphere dataset provides depth ground truth so that disparity - to - depth conversion is no need .
in addition , long range of distance may make the depth map transferred from disparity map unclear for visualization , which make image - based algorithm hard to work .
the provided depth visualization map is a good option .
+ * large scale for learning * freiburg dataset is currently the only dataset with @xmath45 order of scale that is designed for training optical flow , disparity , and scene flow
. protocols usually varies from each other based on different datasets .
the previous datasets like middlebury and rotating sphere usually use rmse / nrmse and aae for evaluation , while newly - introduced datasets like kitti , mpi sintel and freiburg utilize epe for evaluation .
we recommend researchers to use epe as a overall protocol , while rmse and aae can be supplementary means that reveal error distribution and angular error . +
* should we evaluate the 3d error ?
* as equation [ eq : errorprojection ] presents , the disparity in time @xmath33 or the disparity change do have influence on scene flow estimation .
hence , we highly recommended that for 2d parametrization scene flow estimation methods , the accuracy of disparity @xmath33 or the disparity change should be evaluated and provided .
considering the fact that equation [ eq : nrmse - wedel ] and [ eq : aae - wedel ] is rarely utilized , and rotating sphere of basha is the only dataset that provide three - dimension point cloud ground truth , we recommend researchers to provide epe for optical flow , disparity in time @xmath32 and time @xmath33 as a common protocol for scene flow evaluation . with 17 years of development
, scene flow estimation has reached a promising status , while there are still many issues remains to be solved . in this section , we discuss the limitation of existing datasets , and present a brief vision on modification of algorithms .
the survey of existing evaluation methodologies reveals problems and limitations as elaborated below : + * size * on account that real - time scene flow has been achieved with gpu implementation , high resolution data can be introduced for more challenging work . + * ground truth * point cloud is the real three - dimensional parametrization representation for scene flow , which reveals the difference between scene flow and optical flow significantly .
hence , the ground truth for scene flow under the point cloud representation is necessary .
meanwhile , occlusion , textureless region and discontinuity region ground truth are essential for evaluation due to the fact that errors mainly exist near these regions . + *
data source * current datasets mainly focus on scene flow under the binocular setting , while multi - view extension and specific datasets for rgb - d and light field based scene flow is required .
otherwise , the properties of missing data in rgb - d cameras and the abilities like refocusing of light field cameras will be neglected with current datasets .
+ * protocol * a protocol for evaluating performances under different datasets remains vacant .
moreover , the three - dimensional protocol is nt been applied due to the limitation of point cloud ground truth . by checking the error map provided by kitti benchmark
, it s clear that inaccuracy mainly exists in the boundaries of objects .
since this is a common issue for all computer vision tasks , edge - preserving and reasonable filtering is the first priority .
gpu implementation has shown a great efficiency improvement , and the duality - based optimization has proved to enhance the efficiency of global variational methods without accuracy sacrifice . these kind of methods may be a routine in the future for better efficiency . with the development of a robust and efficient estimation between two frames ,
some papers have studied motion estimation under a long sequence @xcite .
the multi - fames estimation with temporal prior knowledges deserves more attention .
a robust temporal constraint can benefit the methods with a better initial value or a better feature @xcite to match .
the challenges like varying illumination and occlusion can be handled with the help of it .
the emerging learning based methods and light field technique has brought fresh blood to scene flow estimation .
learning method with cnn shows an upward tendency in the relevant issues of scene flow estimation like stereo matching and optical flow with promising accuracy and computational cost @xcite . with the help of the up - to - date large scale training dataset @xcite
, learning - based method has a profound potential to achieve an accurate and fast estimation .
light field camera provides more data than existing data source , which brings diverse possibilities for this field .
similar to the emergence of rgb - d cameras , this new source of data may lead to a new attractive branch . on account of the fact that scene flow estimation relies highly on texture and intensity information
, application will suffer in the night or an insufficient illumination circumstance .
moreover , the car headlights and lighting on the building that are frequent in the autonomous driving scene may interfere motion estimation significantly .
hence , the scene flow estimation with insufficient illumination is worth studying .
this paper presents a comprehensive and up - to - date survey on both scene flow estimation methods and the evaluation methodologies for the first time after 17 years since scene flow was introduced .
we have discussed most of the estimation methods so researchers could have a clear view of this field and get inspired for their studies of interest .
the representative methods are highlighted so the differences between these methods are clear , and the similarities between top - tier methods can be seen as a tendency for modification .
the widely used benchmarks have been analyzed and compared , so are multiple evaluation protocols .
this paper provides sufficient information for researchers to choose the appropriate datasets and protocols for evaluating performance of their algorithms .
there are still ample rooms for future research on accuracy , efficiency and multiple challenges .
we wish our work could arise public interest in this field and bring it to a new stage .
this work was supported by projects of national natural science foundation of china [ 61401113 ] ; and natural science foundation of heilongjiang province of china [ lc201426 ] .
n. mayer , e. ilg , p. husser , p. fischer , d. cremers , a. dosovitskiy , t. brox , a large dataset to train convolutional networks for disparity , optical flow , and scene flow estimation , arxiv preprint arxiv:1512.02134 . c. rabe , t. muller , a. wedel , u. franke , dense , robust , and accurate motion field estimation from stereo image sequences in real - time , in : european conference on computer vision ( eccv ) , springer - verlag , 2010 , pp . 582595 .
m. jaimez , m. souiai , j. gonzalez - jimenez , d. cremers , a primal - dual framework for real - time dense rgb - d scene flow , in : ieee international conference on robotics and automation ( icra ) , 2015 , pp .
98104 . o. u. n. jith , s. a. ramakanth , r. v. babu , optical flow estimation using approximate nearest neighbor field fusion , in : ieee international conference on acoustics , speech and signal processing ( icassp ) , 2014 , pp .
6736577 .
h. a. rashwan , m. a. garca , d. puig , variational optical flow estimation based on stick tensor voting , ieee transactions on image processing a publication of the ieee signal processing society 22 ( 7 ) ( 2013 ) 25892599 . c. zhang , z. li , y. cheng , r. cai , h. chao , y. rui , meshstereo : a global stereo model with mesh alignment regularization for view interpolation , in : ieee international conference on computer vision ( iccv ) , 2015 , pp . 20572065 .
f. alcantarilla , j. j. yebes , j. almaz , x00e , l. m. bergasa , on combining visual slam and dense scene flow to increase the robustness of localization and mapping in dynamic environments , in : ieee international conference on robotics and automation ( icra ) , 2012 , pp .
12901297 .
a. wedel , c. rabe , t. vaudrey , t. brox , u. franke , d. cremers , efficient dense scene flow from sparse or dense stereo data , in : european conference on computer vision ( eccv ) , vol .
5302 lncs , 2008 , pp . 739751 .
y. zhang , c. kambhamettu , integrated 3d scene flow and structure recovery from multiview image sequences , in : ieee conference on computer vision and pattern recognition ( cvpr ) , vol . 2 , 2000 , pp .
674681 .
d. ferstl , c. reinbacher , g. riegler , r. m , x00fc , ther , h. bischof , atgv - sf : dense variational scene flow through projective warping and higher order regularization , in : international conference on 3d vision , vol . 1 , 2014 , pp .
285292 .
j. park , t. h. oh , j. jung , y .- w .
tai , i. s. kweon , a tensor voting approach for multi - view 3d scene flow estimation and refinement , in : european conference on computer vision ( eccv ) , vol .
7575 lncs , 2012 , pp . 288302 .
r. l. carceroni , k. n. kutalakos , multi - view scene capture by surfel sampling : from video streams to non - rigid 3d motion , shape and reflectance , in : ieee international conference on computer vision ( iccv ) , vol . 2 , 2001 , pp .
6067 vol.2 .
j. p. pons , r. keriven , o. faugeras , g. hermosillo , variational stereovision and 3d scene flow estimation with statistical similarity measures , in : ieee international conference on computer vision ( iccv ) , 2003 , pp .
597602 vol.1 .
j. p. pons , r. keriven , o. faugeras , multi - view stereo reconstruction and scene flow estimation with a global image - based matching score , international journal of computer vision 72 ( 2 ) ( 2007 ) 179193 .
l. valgaerts , a. bruhn , h. zimmer , j. weickert , c. stoll , c. theobalt , joint estimation of motion , structure and geometry from stereo sequences , in : european conference on computer vision ( eccv ) , vol .
6314 lncs , 2010 , pp .
568581 .
x. zhang , d. chen , z. yuan , n. zheng , dense scene flow based on depth and multi - channel bilateral filter , in : x. zhang , d. chen , z. yuan , z. hu ( eds . ) , asian conference on computer vision ( accv ) , 2012 , pp .
140151 .
m. jaimez , m. souiai , j. st , x00fc , ckler , j. gonzalez - jimenez , d. cremers , motion cooperation : smooth piece - wise rigid scene flow from rgb - d images , in : international conference on 3d vision ( 3dv ) , 2015 , pp .
6472 .
m. w. tao , s. hadap , j. malik , r. ramamoorthi , depth from combining defocus and correspondence using light - field cameras , in : ieee international conference on computer vision ( iccv ) , 2013 , pp .
673680 .
v. lempitsky , s. roth , c. rother , fusionflow : discrete - continuous optimization for optical flow estimation , in : computer vision and pattern recognition , 2008 .
cvpr 2008 .
ieee conference on , ieee , 2008 , pp .
z. lv , c. beall , p. f. alcantarilla , f. li , z. kira , f. dellaert , a continuous optimization approach for efficient and accurate scene flow , in : european conference on computer vision , springer , 2016 , pp .
757773 .
a. dosovitskiy , p. fischer , e. ilg , h. p , xe , usser , c. hazirbas , v. golkov , p. v. d. smagt , d. cremers , t. brox , flownet : learning optical flow with convolutional networks , in : ieee international conference on computer vision ( iccv ) , 2015 , pp .
27582766 .
t. vaudrey , c. rabe , r. klette , j. milburn , differences between stereo and motion behavior on synthetic and real - world stereo sequences , in : international conference of image and vision computing new zealand ( ivcnz ) , 2008 , pp .
16 .
j. gall , c. stoll , e. d. aguiar , c. theobalt , b. rosenhahn , h. p. seidel , motion capture using joint skeleton tracking and surface estimation , in : ieee conference on computer vision and pattern recognition ( cvpr ) , 2009 , pp . | this paper is the first to review the scene flow estimation field to the best of our knowledge , which analyzes and compares methods , technical challenges , evaluation methodologies and performance of scene flow estimation .
existing algorithms are categorized in terms of scene representation , data source , and calculation scheme , and the pros and cons in each category are compared briefly . the datasets and evaluation protocols are enumerated , and the performance of the most representative methods is presented .
a future vision is illustrated with few questions arisen for discussion .
this survey presents a general introduction and analysis of scene flow estimation . |
several cosmological observations show that our universe is expanding with an acceleration@xcite .
this fact can be interpreted as a dominance of the energy of the unknown nature , so called _ dark energy _
( de)@xcite .
the main feature of this energy consists of negative pressure that leads to an accelerated expansion .
the standard cosmological scenario implies that order of 75% of the total energy density is present in the form of de .
there are several observational data based indications that de is highly spatial uniform and isotropic , as well as that the de became dominant recently .
definitely the nature of de is one of major puzzles of modern cosmology@xcite .
a lot of theories of de have been proposed@xcite .
the simplest model of de is the @xmath1cdm model , called a _ concordance model _
, that assumes that the accelerated expansion of the universe is driven by the presence of a cosmological constant@xcite .
this model fits well the cosmological observations , but the @xmath1cdm model has the coincidence and the fine tuning still unexplained problems@xcite . instead of the considering the cosmological constant model
there were several models proposed in which de is a dynamical quantity and in these models de is associated with a _ dynamical scalar field _ . for the @xmath1cdm model
the equation of state parameter @xmath2 ( @xmath3 is a pressure and @xmath4 is an energy density of the de ) is a constant and it equals to minus one , whilst for the dynamical scalar field models the equation of state parameter is a time varying function@xcite .
depending on the value of the equation of state parameter at present , the time dependent de models are divided into the phantom models@xcite ( @xmath5 ) and the _ quintessence models_@xcite ( @xmath6 ) .
the quintessence models are subdivided into two classes : the thawing models and the freezing ( or tracking ) ones.@xcite in the tracking or freezing ( slow roll ) quintessence model the form of the potential allows the attractor in the late - time evolution of the scalar field be insensitive to the initial conditions , and allows the scalar field energy density to track the matter energy density in the matter domination epoch and then the radiation energy density in the radiation domination epoch , remaining subdominant during these epochs . and only at late times , the scalar field becomes dominant and starts behaving like a component with a negative pressure driving the acceleration of the universe .
thus the quintessence models can clarify the coincidence problem .
in this paper we have investigated the freezing quintessence model with an inverse power law ratra - peebles potential@xcite : @xmath7 , @xmath8 is a model parameter , defining the steepness of the potential ; @xmath9 is a scalar field amplitude . in order to distinguish between different dynamical de models commonly constraint of energy equation of state @xmath10
is used , because different models of de give different low of @xmath10 .
recent supernova legacy survey three year sample ( snls3 ) combining with other data on cmb , bao and hubble constant measurement gives rise to @xmath11 for constant of @xmath12 in standard @xmath13 models @xcite .
the bao measurements the values of the equation of state parameter @xmath10 ( @xmath14 is a redshift ) and its redshift derivative @xmath15 is the primary goal of the ongoing de experiments such as snls3 , vipers or boss , but only the next generation of the large scale redshift surveys at @xmath16 and beyond this limit of the redshift like euclid@xcite , wfirst or bigboss@xcite will be able to provide the data to distinguish the de models from each other .
we can get many information about the dynamical de models analyzing the growth of the matter perturbations which were obtained from the redshift space distortion ( rsd ) surveys .
the classical quintessence models are not clustered , but they affect the rate of the matter evolution , therefore the different de models predict the different growth rate history @xcite .
there are a lot of observational growth rate data@xcite , but all these ongoing and future experiments are dedicated to the measurements in the range of the redshifts @xmath17 .
the main goal of our research is the estimation of the sensitivity of the bao and the rsd data to the cosmological parameters , especially to the values @xmath18 and the @xmath19 in the range of the redshifts @xmath20 .
also we have explored what volume and number of the galaxies will be necessary to compete with the other surveys in the range of the redshifts @xmath20 . in this paper we will develop this ideas in quintessence model with ratra - peebles potential , that was well studied in many papers@xcite .
this paper is organized as follows : + the introduction is presented in the sec .
i. in the sec .
ii we have considered a theory of the growth of matter perturbations for the ratra - peebles @xmath9cdm model . in the sec .
iii we have derived the responses of measured quantities to the de model parameter @xmath8 . in the sec .
iv we evaluated the errors of bao and rsd measurements .
our discussions and conclusions are presented in the sec .
the influence of the scalar field ( of the ratra - peebles potential ) on growth of structure was well investigated in many papers@xcite .
further we will follow the paper of o. avsajanishvili et .
al.@xcite we use the linear perturbation equation for computation of the matter s
overdensity@xcite @xmath21 : @xmath22 where @xmath23 is small perturbations in homogeneous universe expanding with the hubble , @xmath24 and @xmath25 are the density and overdensity respectively , @xmath26 is average density of the universe .
a prime designates the differentiation with respect to a scale factor @xmath27 , @xmath28 where @xmath14 is a redshift ; @xmath29 - the normalized value of the hubble parameter @xmath30 to a hubble constant @xmath31 . the first fridmann equation in the case of the spatially - flat universe : @xmath32 where @xmath33 total energy density of all matter components normalized to @xmath34 , @xmath35 and @xmath36 the radiation and the matter density parameters today .
we have neglected the radiation term in the calculations because it is not relevant during the late stage expansion of the universe .
we have applied the fiducial values@xcite : @xmath37 , @xmath38 .
the linear growth factor is defined as : @xmath39 where @xmath40 - a value of the density contrast at present .
the initial conditions are imposed as @xmath41 and @xmath42 , where @xmath43 .
the logarithmic derivative from the linear growth factor is called the growth rate : @xmath44 this function is parametrized as@xcite : @xmath45 where @xmath46 the @xmath47 parameter is called a growth index . suggesting the correctness of general relativity
, the value of @xmath47 parameter can be evaluated as@xcite : @xmath48 where @xmath49 is a value of the equation of state parameter today and @xmath50 . for the @xmath1cdm model @xmath51 .
in order to estimate the precision of the constraints of the value of the @xmath8 parameter , we have found the response of the different cosmological tests on the change of the @xmath8 parameter .
for this purpose we have considered three kinds of cosmological tests which differ by their response on the change of the background dynamics and of the growth rate of the large scale structure : 1 . _
the evolution of the hubble parameter @xmath52_. + the values of the @xmath52 can be obtained from the bao ( along the line of sight ) , sn ia , the weak lensing data .
the luminosity distance and the angular size distance_. + which values can be found from the bao , sn ia data .
( in our analysis is nt relevant what kind of distance is used and further we will use the common designation for distance @xmath53 ) .
3 . _ the linear growth factor @xmath54 or the growth rate function @xmath55_. + which was obtained from the rsd data ( the bias factor is nt sensitive to the background changing ) .
the difference between the measured values of the functions @xmath52 , @xmath53 , @xmath55 and the corresponding values of ones evaluated for the @xmath1cdm model may be by the evidence of the existence the quintessence . in order to illustrate this we have considered the quantity @xmath56 for the @xmath1cdm model : @xmath57 where @xmath58 may be one from the values @xmath59 , or @xmath53 or @xmath55 . when the value of the @xmath8 parameter is small , the value of the @xmath56 is the relative difference of the quantity @xmath58 between the @xmath1cdm model and the ratra - peebles @xmath9cdm model .
we also consider a quantity similar to eq .
[ eq : xi ] for @xmath60 : @xmath61 , @xmath53 , @xmath55 to the presence of the quintessence with a small @xmath8 parameter ( top panel ) and to the change of @xmath62 parameter ( bottom panel ) . ] as one can see from fig .
[ fig : alpha ] , the best response to @xmath63 can be reached for the redshift @xmath64 for @xmath52 and @xmath55 measurements and for the redshift @xmath65 for the distance @xmath53 measurement . in the range of the redshifts
@xmath20 the distance measurement of the @xmath53 is the most sensitive . however , these conclusions are based on the assumption that the present day values of the parameters @xmath36 and @xmath66 are evaluated perfectly well . since this is not the case , to verify the existence of the quintessence is necessary to investigate the differences between the measured values and the corresponding values of ones for the @xmath1cdm model .
in general , this problem should be solved by the mcmc simulations of observations for a set of different cosmological models , but for the better understanding it is instructive to analyze the two limiting cases .
in the first case two quantities @xmath67 and @xmath68 ( @xmath52 , or @xmath53 , or @xmath55 ) are measured at a single redshift @xmath14 . in the second case , the same quantity @xmath58 is measured at two different redshifts @xmath69 and @xmath70 . in order to analyze the first case , we have considered a fiducial model with @xmath71 and @xmath72 .
we can apply the taylor series expansion near the fiducial model for two measurements @xmath67 and @xmath68 : @xmath73 @xmath74 eliminating @xmath75 and obtaining the value of the @xmath8 parameter : @xmath76 thus , when the quantities @xmath67 and @xmath68 are known with the fractional errors @xmath77 and @xmath78 respectively , the error for the calculation of the @xmath8 parameter will be : @xmath79 similarly , when the same quantity @xmath58 is measured at two redshifts @xmath69 and @xmath70 , we have obtained : @xmath80 similar equations ( [ eq : xi]-[eq : err2z ] ) can be obtained from the well known bayesian analysis with the gaussian distributions and uniform priors or with the help of fisher matrix , which is , in or terms : @xmath81
the question of the accuracy of the measurements of the cosmological parameters has been intensively studied in the recent few years@xcite . in order to evaluate the relative errors of the bao measurements
we have used the fitting formula@xcite : @xmath82 where @xmath83 gpc@xmath84 , @xmath85 mpc@xmath86 , @xmath87 number density of the galaxies in @xmath88 mpc@xmath86 , @xmath89 a galaxy bias factor at the scale of the bao measurements , @xmath90 and @xmath91 for the distance @xmath53 and the hubble parameter @xmath52 , respectively . in order to reach the desired precision we have selected a pair of the quantities : @xmath92 and @xmath87 . on fig .
[ fig : vn ] we have shown the relation between these quantities for the precisions of the @xmath53 measurements of 1% and 5% for the redshifts @xmath93 and @xmath94 , respectively . for generality we have used the combination @xmath95 .
the bias depends on the kind of galaxies .
for example , from the planck mission results of measuring the cosmic infrared background anisotropy@xcite , for the submillimeter galaxies @xmath96 . as a function of survey volume @xmath92 needed to achieve precision of distance measurement @xmath53 of 1% or 5% at @xmath93 and @xmath94 by the means of bao observations
. ] concerning the rsd measurements , the precision can be estimated by the formula from ref .
: @xmath97 where @xmath98 gpc@xmath84 and @xmath99 .
the number density and volume of a survey needed to reach the desired precision with the rsd measurements is shown in fig .
3 . as a function of survey volume @xmath92 needed to achieve precision of the growth factor @xmath55 measurement of 1% or 5% by the means of rsd . ]
we substitute the estimates eq .
[ eq : epsbao ] and eq .
[ eq : epsrsd ] for a hypothetical survey which covers all the sky and a logarithmic interval in the redshift in eq .
[ eq : err2q ] and eq .
[ eq : err2z ] and analyze the results .
we have assumed three choices of the number densities of the galaxies : @xmath100 for the bao measurements and @xmath101 for the rsd , in other words we have assumed @xmath102 .
parameter from the bao measurements of the @xmath53 and the @xmath52 functions at a single logarithmic redshift interval as a function of the redshift for number densities of the galaxies @xmath103 ( solid line ) , @xmath104 ( dashed line ) , @xmath105 ( dotted line ) . ] for the case of the measuring pair of quantities at the same redshift , the resulting error of the @xmath8 parameter is the smallest in the case of @xmath106 pair , and for the other two pairs the resulting error is higher on the order of the magnitude for the redshifts @xmath107 for the same galaxy number density .
the values of the @xmath108 for different number densities of the galaxies are shown on fig .
[ fig : d_h ] .
parameter from the bao measurements of the @xmath53 function in two logarithmic redshift intervals .
the number densities of the galaxies @xmath103 ( solid line ) , @xmath104 ( dashed line ) , @xmath105 ( dotted line ) and the first redshift interval is centered at @xmath109 , @xmath110 and @xmath111 respectively , while the position of the second one is varied . ]
parameter from the bao measurements of the @xmath52 function in two logarithmic redshift intervals .
the number densities of the galaxies @xmath103 ( solid line ) , @xmath104 ( dashed line ) , @xmath105 ( dotted line ) . the first redshift interval is centered at @xmath112 , @xmath113 and @xmath114 , respectively , while the position of the second one is varied . ]
secondly , the measurement of the of a single quantity at two different redshifts @xmath69 , @xmath70 gives the most precise estimation when the @xmath52 function or the @xmath53 function is considered .
for the @xmath55 function the value of the @xmath108 is higher on the order of the magnitude .
we have varied both @xmath69 and @xmath70 and found the position of the global minimum of @xmath115 , @xmath116 , and then we fix @xmath69 and plot @xmath117 on figs .
( [ fig : d_z][fig : h_z ] ) . note that when @xmath118 there is insuffitient information to measure @xmath8 and the error tends to infinity , thus , on figs .
( [ fig : d_z][fig : h_z ] ) the spikes are seen at @xmath119 . from the table [ tab :
cat ] one can see that the number densities @xmath120 will be available for the next generation catalogues constructed by the wfirst and by the euclid missions .
for such number densities the parameters of these surveys are optimal for the constraining of the @xmath8 parameter , ( see the figs .
( [ fig : d_h][fig : h_z ] ) ) .
however , when the observations of a larger number of the galaxies at the high redshifts will become available and the density @xmath121 could be reached , the sensitivity of the constraints on the @xmath8 parameter can be increased by 6 - 8 times by moving from the median redshift @xmath65 to the redshift @xmath122 .
the actual number density of galaxies in a survey depends on the luminosity function of galaxies and the detection limit of the survey . from the observed shape of the luminosity function@xcite at @xmath123
we conclude that in order to increase the number density of galaxies by one order of magnitude , the detection limit must be pushed down by 34 magnitudes .
our estimates can be compared with the other predictions of the constraints on @xmath8 by future missions , e.g. the prediction for euclid@xcite which gives @xmath124 when the combined data of cmb and bao measurements are used .
we use equation ( [ eq : err2z ] ) to estimate @xmath108 . to take into account cmb measurements , we take the relative error in matter density @xmath125 and we place this measurement at @xmath126 . we also take into account that euclid survey will cover roughly half of the sky . from our approach
we have found @xmath127 which is close to the result of amendola et al .
we have analyzed the observational manifestations of ratra - peebles model of dark energy at high redshifts , @xmath20 , with the help of a simple approach similar to fisher matrix .
our approach allows to find the precision of constraints on two model parameters , matter density @xmath60 and spectral index of quintessence potential @xmath8 , from the accuracy of measurements of two observables .
we have considered three observables : the hubble parameter , the angular size distance and the growth rate of density perturbations .
for these three observables we have considered two kinds of measurements : in the first case two different observables are measured at the same redshift . in the second case
the same observable is measured at two different redshifts . in order to make predictions on the accuracy of measurements of observables in future surveys we have taken from literature the approximations of the precision of bao and rsd measurements .
we have considered all possible combinations of pairs of three observables listed above measured at a single redshift and also measurements of each observable at two different redshifts and found that at @xmath20 bao measurements of are more sensitive to @xmath8 than rsd .
we have found that when the number density of galaxies in future large surveys at these redshifts will approach to @xmath128 , bao measurements alone at @xmath20 will allow to improve constraints on parameter @xmath8 in comparison with the measurements at @xmath129 .
this will require severe increase of the sensitivity of telescopes , so the missions aimed at measuring de parameters from the large scale structure at @xmath20 need to be technically complicated than currently planned missions such as euclid .
we thank for useful comments from olga avsajanishvili . this research was supported by basic research program p-7 of the presidium of the russian academy of sciences and grant of the president of the russian federation for support of leading scientific schools of the russian federation nsh-6595.2016.2 .
the work of sergey pilipenko is also partially supported by unk fian .
0 d. h. weinberg et al . , arxiv:1201.2434[astro - ph.co ] ; s. perlmutter et al . , _ astrophys .
j. _ * 517 * 565 ( 1999 ) .
a. g. riess et al .
, _ astron .
j. _ * 116 * 1009 ( 1998 ) .
a. g. riess et al .
, _ astrophys .
j. _ * 659 * 98 ( 2007 ) .
a. g. riess et al .
, _ astrophys .
j. _ * 607 * 665 ( 2004 ) .
a. albrecht et al . , astro - ph/0609591 ; m. hicken et al . , _ astrophys .
j. _ * 700 * 1097 ( 2009 ) .
r. kessler et al . , _ astrophys .
j. suppl . _ * 185 * 32 ( 2009 ) .
d. huterer , arxiv:1010.1162 [ astro-ph.co ] . j. frieman , m. turner and d. huterer , _ ann .
astrophys . _ * 46 * 385 ( 2008 )
. e. v. linder , arxiv:1009.1411[astro - ph.co ] ; e. v. linder , arxiv:1004.4646 [ astro-ph.co ] ; v. n. lukash and v. a. rubakov , _ phys .
_ * 51 * 283 ( 2008 ) . w. m. wood - vasey et al . , _ astrophys . j. _ * 666 * 694 ( 2007 ) .
r. j. scherrer and a. a. sen , _ phys .
d _ * 78 * 067303 , ( 2008 ) .
r. r. caldwell and e. v. linder , _ phys .
lett . _ * 95 * 141301 ( 2005 ) .
t. chiba , a. de felice and s. tsujikawa , arxiv:1210.3859 [ astro-ph.co ] ; r. de putter and e. v. linder , _ jcap _ * 0810 * 042 ( 2008 ) .
i. zlatev , l. -m .
wang and p. j. steinhardt , _ phys .
_ * 82 * 896 ( 199 . ) .
a. masiero , m. pietroni and f. rosati , _ phys .
d _ * 61 * 023504 ( 2000 ) .
p. j. steinhardt , l. -m .
wang and i. zlatev , _ phys .
d _ * 59 * 123504 ( 1999 ) .
r. r. caldwell , r. dave and p. j. steinhardt , _ phys .
_ * 80 * 1582 ( 1998 ) .
k. shi , y. f. huang and t. lu , _ phys .
lett . b _ * 717 * 299(2012 ) .
l. samushia , w. j. percival and a. raccanelli , _ mon . not .
_ * 420 * 2102 ( 2012 ) . c. di porto , l. amendola and e. branchini , arxiv:1101.2453 [ astro-ph.co ] ; y. gong , _ phys . rev .
d _ * 78 * 123010 ( 2008 ) .
a. pavlov , l. samushia and b. ratra , _ astrophys .
j. _ * 760 * 19 ( 2012 ) .
a. b. belloso , j. garcia - bellido and d. sapone , _ jcap _ * 1110 * 010 ( 2011 ) .
x. fu , p. wu and h. w. yu , _ eur . phys .
j. c _ * 68 * 271 ( 2010 ) .
s. lee and k. -w .
ng , _ phys .
d _ * 82 * 043004 ( 2010 ) .
s. lee and k. -w .
ng , arxiv:0905.1522 [ astro-ph.co ] . l. samushia _ et al .
_ , _ mon . not .
_ * 410 * 1993(2011 ) . c. di porto and l. amendola ; _ phys .
d _ * 77 * 083508 ( 2008 ) .
z. girones et al . , _ j. cosmology astropart .
phys . _ * 11 * 4 ( 2010 ) ; k. hirano , z. komiya and h. shirai , _ prog .
_ * 127 * 1041 ( 2012 ) .
j. -q .
xia ; _ phys .
d _ * 79 * 103527 ( 2009 ) . c. blake _ et al .
_ , _ mon . not .
soc . _ * 415 * 2876 ( 2011 ) ; c. blake _ et al .
_ , _ mon . not .
soc . _ * 406 * 803 ( 2010 ) .
l. -m .
wang and p. j. steinhardt , _ astrophys .
j. _ * 508 * 483 ( 1998 ) .
e. v. linder , _ phys .
d _ * 72 * 043529 ( 2005 ) . c. blake , d. parkinson , b. bassett , et al .
, _ mon . not .
_ * 365 * 255 ( 2006 ) | now creation of big catalogs of galaxies for measurement of baryon acoustic oscillation ( bao ) is actively conducted . existing and planned in the near future surveys
are directed on the range of red shifts of z@xmath02
. however , some popular models of dark energy ( de ) give the maximum deviation from @xmath1cdm at z>2 therefore we investigated sensitivity of hypothetical high redshift surveys to the model of de .
we have found that with the increase of the number density of detected galaxies at z>2 high redshift observations may give better constraints of de parameters . |
there are various models of collective opinion formation in which agents modify their opinions according to interaction with other agents @xcite . opinion formation is a dynamic process : for example , interaction between agents makes their opinions approach each other .
an important problem in opinion dynamics is to examine when an agreement ( i.e. , consensus ) among all the agents occurs .
complete agreement is rarely observed in the real world @xcite .
however , it is an established fact that opinion dynamics under the voter model , a classical opinion model in statistical physics and probability theory , inevitably reaches agreement in finite populations @xcite .
the majority rule model has a similar feature @xcite .
partly motivated by this discrepancy , various extensions of voter and majority rule models and different models of collective opinion formation have been proposed to account for the disagreement in finite populations .
examples include the deffuant model @xcite , language competition models @xcite , voter - like models on adaptive networks @xcite , voter model under partisan bias ( the assumption that agents naturally prefer one opinion ) @xcite , and variations of axelrod s cultural dynamics ( see @xcite for references ) .
theoretical models have also been proposed in social sciences to explain disagreement in the context of polarization .
for example , prior beliefs or initially received signals can cause disagreement between agents , even if they receive the same public signals from then on @xcite .
although there is a plethora of studies addressing the problem of agreement and disagreement in opinion dynamics , we propose a model incorporating two factors that are relevant to human behavior : bayesian belief updating and confirmation bias .
bayesian belief updating is commonly used in studies of the decision making of agents receiving uncertain information @xcite .
the confirmation bias is a psychological bias inherent in humans , in which an agent inclined towards an opinion tends to misperceive incoming signals as supporting the agent s belief @xcite . a non - bayesian model with the confirmation bias
was previously proposed for explaining the influences of media and interactions between agents @xcite .
we are not the first to study opinion formation under the bayesian updating and confirmation bias . in the framework of single agent opinion formation
, rabin and schrag showed that the confirmation bias triggers overconfidence and can cause the individual to hold incorrect beliefs , even if it receives a series of external signals suggesting the true state of the world @xcite .
orlan studied the bayesian dynamics of agents subjected to the confirmation bias , interacting through the mean field @xcite . the model yields agreement or disagreement depending on the parameter values . in this study , motivated by the rabin - schrag model @xcite
, we propose a model of collective opinion formation with a confirmation bias .
we model direct peer - to - peer interactions between agents ( not through the mean field ) and their effects on the bayesian updating of each agent . to study the pure effects of interactions among agents
, we do not assume that agents receive signals from the environment as in previous studies @xcite .
we numerically simulate the model to reveal the conditions under which the populations of agents agree and disagree , depending on the values of parameters such as the strength of the confirmation bias , fidelity of the signal , and the system size .
our model modifies the bayesian decision - making model proposed by rabin and schrag @xcite in two main ways .
first , we consider a well - mixed population of bayesian agents that interact with each other ; rabin and schrag focused on the case of the single agent .
second , agents do not receive external signals from the environment in our model . in the rabin - schrag model , such an external signal , which represents the correct " answer in the binary choice situation ( i.e. , the true state of nature ) ,
is assumed . by making the two changes , we concentrate on collective opinion formation by bayesian agents , whereby there are two possible alternative opinions of equal attractiveness .
we label the @xmath0 agents @xmath1 and denote the opinion of agent @xmath2 ( @xmath3 ) by @xmath4 , where a and b are the alternative opinions .
we assume that agents are not perfectly confident in their opinions . to model this factor
, we adopt the bayesian formalism used by rabin and schrag @xcite .
we denote by @xmath5 the strength of the belief ( hereafter , simply the belief ) with which agent @xmath2 believes in opinion a. a parallel definition is applied to @xmath6 .
it should be noted that @xmath7 , @xmath8 , and @xmath9 . if @xmath10 , agent @xmath2 is indifferent to either opinion .
we update the agent s belief as follows . the time @xmath11 starts from @xmath12 . upon every updating of an agent s belief , we add @xmath13 to @xmath11 such that the belief of each agent is updated once per time unit on average . in an updating event , we select an agent @xmath2 to be updated with equal probability @xmath13 .
agent @xmath2 refers to agent @xmath14 s opinion for updating @xmath2 s belief @xmath5 , where @xmath14 ( @xmath15 ) is selected with equal probability @xmath16 from the population .
agent @xmath14 imparts a signal @xmath17 , where @xmath18 and @xmath19 correspond to @xmath14 s opinions a and b , respectively .
we assume that the probabilities that agent @xmath14 imparts @xmath20 and @xmath21 are given by @xmath22 and @xmath23 respectively , where @xmath24 represents the reliability of the signal , and @xmath25 .
if @xmath14 is confident in its own opinion and the transformation from @xmath14 s belief [ i.e. , @xmath26 to @xmath14 s output signal ( i.e. , @xmath18 or @xmath19 ) is reliable , signals @xmath18 and @xmath19 are likely to indicate opinions a and b , respectively . in the limit @xmath27 , @xmath28 and @xmath29 . if @xmath30 , @xmath31 such that @xmath32 does not convey any information about @xmath14 s belief .
we implicitly assume that all the agents share the same value of @xmath33 and that they know this fact when performing the bayesian update , as described below .
when agent @xmath14 imparts signal @xmath34 , agent @xmath2 is assumed to perceive a subject signal @xmath35 , where @xmath36 and @xmath37 correspond to @xmath38 and @xmath39 , respectively .
the flow of the signal conversion is depicted in fig .
[ fig : signals ] .
if agent @xmath2 is not subject to the confirmation bias , @xmath36 and @xmath37 are equal to @xmath18 and @xmath19 , respectively .
otherwise , agent @xmath2 may misinterpret the signal imparted by agent @xmath14 , depending on the prior exposure of agent @xmath2 to other signals . following rabin and schrag @xcite
, we define @xmath40 = \pr[\sigma=\beta | s = b,\pr(x_i={\rm a } ) \le 1/2 ] = 1 \label{eq : s_to_sigma_bias_1}\end{aligned}\ ] ] and @xmath41 = \pr[\sigma=\beta | s = a,\pr(x_i={\rm a } ) < 1/2 ] = q , \label{eq : s_to_sigma_bias_2}\end{aligned}\ ] ] where @xmath42 ( @xmath43 ) parameterizes the strength of the confirmation bias .
equation states that an agent preferring opinion a misinterprets an arriving @xmath19 signal as @xmath38 ( i.e. , @xmath44 ) with probability @xmath42 .
if @xmath45 , the confirmation bias is absent , and @xmath20 and @xmath21 are always converted to @xmath44 and @xmath46 , respectively . if @xmath47 , the agent perceives the signal that is consistent with its current preference [ i.e. , @xmath36 if @xmath48 and @xmath37 if @xmath49 , irrespective of the signal imparted by agent @xmath14 ( i.e. , @xmath18 or @xmath19 ) . the other conditional probabilities can be readily derived from eqs .
( [ eq : s_to_sigma_bias_1 ] ) and ( [ eq : s_to_sigma_bias_2 ] ) . for example , eq . ( [ eq : s_to_sigma_bias_1 ] ) implies @xmath50 = 1 - \pr[\sigma=\alpha|s = a , \pr(x_i={\rm a})>1/2 ] = 0 , \label{eq : s_to_sigma_bias_3}\end{aligned}\ ] ] and eq .
( [ eq : s_to_sigma_bias_2 ] ) implies @xmath51 = 1- \pr[\sigma=\beta|s = a , \pr(x_i={\rm a})<1/2 ] = 1-q .
\label{eq : s_to_sigma_bias_4}\end{aligned}\ ] ] then , by using the bayes theorem , we update agent @xmath2 s belief @xmath52 on the basis of the old belief @xmath5 [ @xmath53 and the perceived signal ( i.e. , @xmath36 or @xmath37 ) . the perceived signal may be different from the received signal ( i.e. , @xmath18 or @xmath19 ) because of their confirmation bias [ eqs . and ] .
we assume that agents are not aware that they may be subject to the confirmation bias .
agents use the subjective conditional probabilities given by @xmath54 and @xmath55 to perform the bayesian update . the posterior belief @xmath52 is given by @xmath56
it should be noted that @xmath57 .
then , we increment the time by @xmath13 such that each agent is updated once per unit time on average .
iterative application of eq .
( [ eq : pia_update ] ) leads to @xmath58 and @xmath59 where @xmath60 ( @xmath61 ) is the accumulated number of signals @xmath44 ( @xmath46 ) that agent @xmath2 has perceived .
the state of each agent @xmath2 is uniquely determined by @xmath62 , which is consistent with basic bayesian theory @xcite .
unless otherwise stated , we set @xmath63 and assume a neutral initial condition @xmath64 ( @xmath65 ) , or , equivalently , @xmath66 ( @xmath67 ) . the agents exchange signals and
update their beliefs , possibly under a confirmation bias .
after a transient , the agents believe in either opinion with a strong confidence , i.e. , @xmath68 or @xmath69 .
we halt a run when @xmath70 is satisfied for all @xmath2 for the first time , where @xmath71 is the threshold .
in other words , a run continues if at least one agent @xmath2 has the @xmath72 value smaller than @xmath71 .
we first consider the case without a confirmation bias ( i.e. , @xmath45 ) .
we investigate the dynamics of the mean belief @xmath73 at time @xmath11 by drawing a return map , i.e. , @xmath74 as a function of @xmath75 @xcite .
the return map for @xmath76 , @xmath77 , and @xmath78 based on @xmath79 runs is shown in fig .
[ fig : meanp_ba ] . because @xmath80 when
@xmath81 and @xmath82 when @xmath83 , the dynamics is in accordance with majority rule behavior .
all @xmath79 runs finished with an agreement of opinion a [ i.e. , @xmath84 for all @xmath2 ] or opinion b [ i.e. , @xmath86 for all @xmath2 ] .
each case occurred approximately half the time .
we turn on the confirmation bias to examine the possibility that it induces disagreement among agents . at least for large @xmath42 ( i.e. , @xmath87 )
, disagreement is expected to be reached because the first perceived signal would determine the final belief of each agent and is equally likely to be @xmath36 and @xmath37 for many agents . in the following numerical simulations , we measured the degree of disagreement , which we defined as follows .
we determined that agreement was reached in a run if the final signs of @xmath62 were the same for @xmath3 .
otherwise , we said that disagreement was reached .
we denoted the fraction of runs that finished with disagreement by @xmath88 .
we set @xmath78 and the number of runs to @xmath79 . in figs .
[ fig : r_bm](a ) and [ fig : r_bm](b ) , @xmath88 is shown as a function of @xmath42 and @xmath33 for @xmath89 and @xmath90 , respectively .
first , @xmath88 monotonically increases with @xmath42 and decreases with @xmath33 for both @xmath89 and @xmath76 .
it should be noted that disagreement occurred in at least one run in the regions right to the solid fractured lines in fig .
[ fig : r_bm ] .
second , @xmath88 for @xmath89 [ fig .
[ fig : r_bm](a ) ] is smaller than @xmath88 for @xmath76 [ fig .
[ fig : r_bm](b ) ] for all the @xmath42 and @xmath33 values . therefore , disagreement seems to be a likely outcome of the model for large @xmath0 , particularly for large @xmath42 and small @xmath33 .
when @xmath76 , perfect agreement , i.e. , @xmath91 , is realized only for @xmath42 close to zero . in other words ,
even a small degree of confirmation bias elicits disagreement among the agents . to obtain analytical insights into the model
, we performed an annealed approximation for @xmath89 by averaging out fluctuations of the dynamics for different times and runs .
the configuration of the population is specified by @xmath92 .
the stochastic dynamics of the model can be mapped to a random walk on the two - dimensional lattice ; a walker is initially located at @xmath93 and randomly hops to one of the four neighboring lattice points in each time step .
we defined @xmath94 , @xmath95 , @xmath96 , and @xmath97 as the probabilities that the walker located at @xmath98 moves to @xmath99 , @xmath100 , @xmath101 , and @xmath102 , respectively .
the four probabilities are given by @xmath103 \dfrac{\pr(s = a | m_2)}{2 } & ( m_1 = 0 ) , \\[6pt ] \dfrac{(1-q ) \pr(s = a | m_2)}{2 } & ( m_1 \le -1 ) , \end{cases } \label{eq : fr}\end{aligned}\ ] ] @xmath104 @xmath105 \dfrac{\pr(s = a | m_1)}{2 } & ( m_2 = 0 ) , \\[6pt ] \dfrac{(1-q ) \pr(s = a | m_1)}{2 } & ( m_2 \le -1 ) , \end{cases } \label{eq : fu}\end{aligned}\ ] ] and @xmath106 where @xmath107 is the probability that agent @xmath14 with @xmath108 imparts signal @xmath20 . @xmath94 and @xmath96 increase with @xmath109 and @xmath110 , and @xmath95 and @xmath97 decrease with @xmath109 and @xmath110 . in the following , we study the mean dynamics of the random
walk driven by the drift terms . because the transition probability of the random walk is symmetric with respect to the lines @xmath111 and @xmath112 , we focus on the region given by @xmath113 .
we define @xmath114 and @xmath115 , which are not integers in general , as the values satisfying @xmath116 and @xmath117 , respectively .
they are given by @xmath118^{-1}}{\ln \dfrac{\theta}{1-\theta}}. \label{eq : r2_fr = fl}\end{aligned}\ ] ] note that @xmath114 and @xmath115 exist if and only if @xmath119 , i.e. , @xmath120 first , we consider the case @xmath121 .
we partition the upper quadrant of the lattice ( given by @xmath113 ) into five regions : region @xmath69 ( @xmath122 ) , region @xmath123 ( @xmath124 ) , region @xmath125 ( @xmath126 ) , region @xmath127 ( @xmath128 ) , and region @xmath129 ( @xmath130 ) , as shown in fig .
[ fig : schematicview_fr - fl_fu - fd](a ) .
we obtain from the condition @xmath131 @xmath132 in region @xmath69 , @xmath133 in region @xmath123 , @xmath134 in region @xmath125 , @xmath135 in region @xmath127 , and @xmath136 in region @xmath129 .
the probability flow of the walker after the annealed approximation , i.e. , ( @xmath137 ) inferred from eqs .
- is shown schematically in fig .
[ fig : schematicview_fr - fl_fu - fd](a ) . if the walker is in the second quadrant ( i.e. , regions @xmath125 , @xmath127 , and @xmath129 ) where the two agents disagree with each other , the random walker is likely to eventually escape and enter the first quadrant ( i.e. , region @xmath69 ) where the two agents agree with each other .
in fact , fig . [
fig : vectorplot](a ) , which shows the actual probability flow , indicates that the agreement necessarily occurs .
therefore , agreement is the expected outcome when @xmath138 .
second , if @xmath139 , regions @xmath127 and @xmath129 are absent because @xmath140 and @xmath141 diverge . regions @xmath69 and @xmath123 , in which inequalities and are satisfied , respectively , are the same as those in the case @xmath138 .
region @xmath125 , in which inequality is satisfied , is modified to @xmath142 . the probability flows are schematically shown in fig .
[ fig : schematicview_fr - fl_fu - fd](b ) . @xmath143 and @xmath144 are satisfied in region @xmath125 .
therefore , once the walker is deep in the second quadrant , it is likely to move toward @xmath145 and @xmath146 , which implies that two agents finally disagree . the actual probability flow shown in fig .
[ fig : vectorplot](b ) is consistent with this prediction . the transition line @xmath147 is shown by the dashed line in fig .
[ fig : r_bm](a ) .
it accurately predicts the parameter region in which disagreement can occur , i.e. , the region right to the solid line .
the same transition line is also derived for the rabin - schrag model , which is concerned with a single agent subjected to a confirmation bias @xcite . in their model ,
the agent forms a belief by repetitively receiving a stochastic signal @xmath34 from nature , according to @xmath148 .
rabin and schrag calculated the probability that the agent eventually misunderstands the state of the nature ( i.e. , a or b ) , starting from neutral belief .
this probability is equal to zero when @xmath149 and positive when @xmath150 ( see proposition @xmath127 in @xcite ) .
our results obtained in this section are consistent with theirs because disagreement in our model roughly corresponds to misunderstanding in the rabin - schrag model . in general , there are @xmath151 disagreement configurations , as distinguished by the number of agents that finally believe in opinion a , which ranges from @xmath69 to @xmath151 . to distinguish different disagreement configurations , we examined the fraction of agents that believed in the minority opinion at the end of a run .
we averaged this fraction over the runs ending with disagreement .
we called this quantity the average size of the minority .
figures [ fig : r_minority_minoritymin](a ) and [ fig : r_minority_minoritymin](b ) show the average size of the minority for @xmath152 and @xmath76 , respectively .
the black regions indicate the parameter values for which the average size of the minority is undefined because all @xmath79 runs end with agreement .
when @xmath42 is small , the average size of the minority monotonically decreases with @xmath42 and monotonically increases with @xmath33 for both @xmath152 and @xmath90 .
therefore , small @xmath42 and large @xmath33 values allow only balanced disagreement configurations , in which the numbers of the agents believing in the opposite opinions are close to @xmath153 . however , the average size of the minority increases when @xmath42 is large .
this is particularly the case for @xmath76 [ fig .
[ fig : r_minority_minoritymin](b ) ] .
this increase occurs for the following reason . with a strong confirmation bias
, agents end up with an opinion consistent with a small number of signals perceived in the early stages , and both signals are equally likely to be observed in the early stages under neutral initial conditions . in the extreme case in which @xmath47 , agents reinforce the opinion that is consistent with their first perceived signal .
therefore , unbalanced disagreement configurations are rarely realized when @xmath42 is large .
figures [ fig : r_bm ] and [ fig : r_minority_minoritymin ] suggest that the agreement is unlikely to be reached in a large population .
to examine the effect of the population size , we defined @xmath154 as the value of @xmath42 such that a threshold number of runs among @xmath155 runs end with agreement . for a given @xmath33 value
, we determined @xmath154 by the bisection method .
the number of agreement runs may not monotonically change in @xmath42 because the number of runs is finite .
therefore , the bisection method does not perfectly work in general .
however , we corroborated that the following results were negligibly affected by the lack of monotonicity .
the dependence of @xmath154 on @xmath0 is shown in fig .
[ fig : qcr_vs_n](a ) for three threshold values .
for example , the results for the threshold value @xmath90 ( shown by circles ) indicate that at least @xmath90 runs among the @xmath155 runs end up with disagreement when @xmath156 .
we set @xmath77 and @xmath78 .
to explore the possibility of disagreement in large populations , we set @xmath33 close to @xmath69 .
it should be noted that fig .
[ fig : r_bm ] indicates that the probability of disagreement is small for a large @xmath33 value . in fig .
[ fig : qcr_vs_n](a ) , @xmath154 quickly decreases for @xmath157 and gradually decreases for @xmath158 .
disagreement often occurs for large @xmath0 unless @xmath42 is small .
nevertheless , fig .
[ fig : qcr_vs_n](a ) suggests that the range of @xmath42 for which agreement always occurs survives for diverging @xmath0 . in generating fig .
[ fig : qcr_vs_n](a ) , we used an initial condition in which all the agents had a neutral belief [ i.e. , @xmath159 . to check the effect of the initial condition ,
we investigated the dependence of @xmath154 on @xmath0 under two other initial conditions . in the bimodal initial condition , we initially set @xmath160 ( @xmath161 ) and @xmath162 ( @xmath163 ) .
we assumed that @xmath0 was even for this initial condition .
in the so - called most unbalanced initial condition , we set @xmath164 and @xmath162 ( @xmath165 ) . the numerical results for the two initial conditions are shown in figs .
[ fig : qcr_vs_n](b ) and [ fig : qcr_vs_n](c ) .
the parameter values @xmath77 , @xmath78 are the same as those used in fig . [
fig : qcr_vs_n](a ) .
the transition point @xmath154 decreases with @xmath0 more rapidly with the bimodal initial condition [ fig .
[ fig : qcr_vs_n](b ) ] than with the neutral initial condition [ fig .
[ fig : qcr_vs_n](a ) ] .
this result is intuitive : the bimodal initial condition paves the way to disagreement .
in contrast , @xmath154 under the most unbalanced initial condition is almost constant near @xmath166 irrespective of @xmath0 .
therefore , disagreement is highly unlikely unless the confirmation bias is strong ( i.e. , @xmath42 is greater than @xmath166 ) .
the results shown in fig . [
fig : qcr_vs_n ] suggest that the eventual behavior of the model strongly depends on the initial condition even after the results are averaged over runs .
our numerical results are summarized as follows . when the confirmation bias is absent ( i.e. , @xmath45 ) , the opinion dynamics under the bayesian update rule leads to the complete agreement among agents .
the behavior of the model is similar to majority rule dynamics ( fig .
[ fig : meanp_ba ] ) . when the confirmation bias is present , disagreement is a likely outcome , particularly for a strong confirmation bias ( i.e. , large @xmath42 ) .
disagreement is also more likely for a lower fidelity of the signal ( i.e. , @xmath167 ) and a larger system size .
the transition line separating the parameter region in which both agreement and disagreement can occur and that in which only agreement occurs is approximately given by @xmath168 when @xmath89 .
this line is identical to the one determined by rabin and schrag for their model for a single agent s decision making @xcite . finally , the behavior of the model strongly depends on the initial condition .
our model and results are different from orlan s @xcite , although orlan s model employs multiple agents that perform the bayesian updates under a confirmation bias .
first , the belief of each agent is binary in orlan s model , whereas our model introduces an infinite range of discrete beliefs , as in @xcite .
second , interaction between agents is introduced differently in the two models . in orlan s model
, each agent refers to the global fraction of agents believing in one of the two opinions . in our model
, agents refer to other opinions by peer - to - peer interaction , i.e. , by receiving a binary signal that is correlated with the belief of the sender .
third , the stochastic dynamics of orlan s model is ergotic when the collective opinion does not reach agreement .
the collective opinion obeys a stationary distribution , irrespective of the initial condition .
in contrast , in all our simulations , the stochastic dynamics of our model was nonergotic , such that the final configuration depended on the initial condition in a wide parameter region .
in social science studies of polarization , several authors analyzed bayesian models in which different agents receiving a series of common signals end up in disagreement .
the proposed mechanisms governing disagreement include different initial beliefs or factors that affect perception of later incoming signals @xcite , different update rules @xcite , and ambiguity aversion @xcite .
these models and ours are different in three major ways .
first , a ground truth opinion corresponding to the state of nature is assumed in these models but not in ours .
second , public signals commonly received by different agents are assumed in these models but not in ours .
third , the agents do not have direct peer - to - peer interaction in these models , but they do in ours
. models with interacting bayesian agents , which show disagreement ( reviewed in ref .
@xcite ) , are also different from our model in the first respect .
it should be noted that zimper and ludwig discussed confirmation bias with their bayesian model @xcite .
however , they derived a confirmation bias from their model , rather than assuming one , such that their results pertaining to confirmation bias were also distinct from ours . extending our model to the case of networks
is straightforward .
for example , we can select a recipient of the signal with probability @xmath13 and then select the sender with equal probability among the neighbors of the recipient on the network .
another possible update rule is to select the sender first and then the recipient among the sender s neighbors .
yet another possibility is to select a link with equal probability and designate one of the two agents as sender and the other as recipient . on heterogeneous networks ,
the results may depend on the update rule because it is the case in the voter model @xcite .
extension of the model to the case of confirmation bias heterogeneity may also be interesting .
neurological evidence shows that different individuals have different confirmation bias strengths @xcite .
the strength of the confirmation bias and the position of the node in a social network may be correlated and affect the dynamics .
it is also straightforward to extend the model to the case of multiple opinion cases .
these and other extensions , along with the study of analytically tractable models that capture the essence of the present study , warrant future work .
we thank mitsuhiro nakamura , taro takaguchi , and shoma tanabe for critical reading of the manuscript .
this work is supported by grants - in - aid for scientific research [ grant
23681033 and innovative areas systems molecular ethology " ( grant no .
20115009 ) ] from mext , japan . , @xmath45 , and @xmath77 .
we recorded the values of @xmath169 ) for @xmath170 for @xmath79 runs and divided the recorded pairs into @xmath171 classes . the @xmath172th class ( @xmath173 )
was composed of the pairs satisfying @xmath174 .
we obtained the mean value @xmath175 for the @xmath172th class by averaging @xmath74 over all the pairs contained in the @xmath172th class .
finally , we plotted @xmath175 against @xmath176 for @xmath173 .
the diagonal @xmath177 is also shown as a guide . ] .
( a ) @xmath89 .
( b ) @xmath76 .
solid lines represent the boundary between @xmath91 and @xmath178 . the dashed line in ( a ) represents @xmath168 .
the dashed line is not drawn in ( b ) because this theoretical estimate is valid only for @xmath89 . in ( a )
, the two lines almost overlap each other .
the initial belief of each agent was assumed to be neutral [ i.e. , @xmath179 , @xmath67 ] . ] .
vector @xmath183 is shown by an arrow of proportional size at each position of the random walker @xmath184 .
( a ) @xmath185 and @xmath186 , which satisfies @xmath138 .
( b ) @xmath187 and @xmath186 , which satisfies @xmath188 .
the size of the vectors is manually normalized for clarity , independently for the two panels . ] and ( b ) @xmath76 .
the initial belief of each agent is assumed to be neutral [ i.e. , @xmath179 , @xmath67 ] .
the black region represents the case where all the @xmath79 runs end with agreement such that the average size of the minority is undefined . ]
99ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop ( , , ) @noop ( , , ) @noop ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.100.108702 [ * * , ( ) ] link:\doibase 10.1103/physreve.77.016102 [ * * , ( ) ] link:\doibase 10.1103/physreve.74.056108 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.158701 [ * * , ( ) ] @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop `` , '' @noop * * , ( ) @noop * * , ( ) @noop ( , , ) @noop ( ) @noop ( , , ) @noop * * , ( ) @noop * * ( 5 ) , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop `` , '' @noop * * , ( ) @noop * * , ( ) | we propose a collective opinion formation model with a so - called confirmation bias .
the confirmation bias is a psychological effect with which , in the context of opinion formation , an individual in favor of an opinion is prone to misperceive new incoming information as supporting the current belief of the individual .
our model modifies a bayesian decision - making model for single individuals [ m. rabin and j. l. schrag , q. j. econ . *
114 * , 37 ( 1999 ) ] for the case of a well - mixed population of interacting individuals in the absence of the external input .
we numerically simulate the model to show that all the agents eventually agree on one of the two opinions only when the confirmation bias is weak .
otherwise , the stochastic population dynamics ends up creating a disagreement configuration ( also called polarization ) , particularly for large system sizes .
a strong confirmation bias allows various final disagreement configurations with different fractions of the individuals in favor of the opposite opinions .
pacs numbers : : 87.23.ge , 02.50.ey , 02.50.le 0 |
completion of the knowledge of the generalized nuclear force , which includes not only the nucleon - nucleon ( nn ) interaction but also hyperon - nucleon ( yn ) and hyperon - hyperon ( yy ) interactions , brought the deeper understanding of atomic nuclei , structure of neutron stars and supernova explosions .
however it is hard to know the properties of the yn and yy interactions because their scattering data in free - space are scarce . recently
a method to extract the @xmath8 potential through the nbs wave function from lattice qcd simulations has been proposed in @xcite .
the obtained potential is found to have desirable features , such as attractive well at long and medium distances , and the central repulsive core at short distance @xcite .
further applications have been done in refs .
@xcite . in this work ,
we focus on the @xmath0 , @xmath1 b - b system to seek the @xmath9 interaction and to see the su(3)@xmath10 breaking effects of b - b interaction from lattice qcd simulation .
the @xmath11 baryon - baryon state consists of the @xmath9 , @xmath12 and @xmath13 components in terms of low - lying baryons .
mass differences of these components are quite small , and it causes the contamination of nbs wave function from excited states .
in sucn situation the source operator should be optimized to extract the energy eigen states through the variational method @xcite .
the equal - time nbs wave function @xmath14 for an energy eigen state with @xmath15 is extracted from the four point function , @xmath16 where @xmath17 is diagonalized wall - source operator .
the transition potential matrix of 3-states coupled channel equation can be acquired in a particle basis or a su(3 ) irreducible representation ( ir ) basis .
they are connected by unitary trandformation ( see in appendix b in ref .
the non - diagonal part of potential matrix in ir basis is a good measure of the su(3 ) breaking effect .
.hadron masses in unit of [ mev ] are listed . [ cols="^,^,^,^,^,^,^,^",options="header " , ] in this calculation we employ the 2 + 1-flavor full qcd gauge configurations of japan lattice data grid(jldg)/international lattice data grid(ildg ) .
they are generated by the cp - pacs and jlqcd collaborations with a renormalization - group improved gauge action and a non - perturbatively @xmath18 improved clover quark action at @xmath19 , corresponding to lattice spacings of @xmath20 @xcite .
we choose three ensembles of the @xmath21 lattice which means the spatial volume of about @xmath22 .
quark propagators are calculated from the spatial wall source at @xmath23 with the dirichlet boundary condition in temporal direction at @xmath24 .
the numerical computation is carried out at kek supercomputer system , blue gene / l .
the hadron masses are shown in table [ tab : gconf ] .
in figure [ fig : potall ] we compare the results of potential matrix in the ir basis calculated in different configuration sets .
we found the growth of repulstive core in the @xmath25 potential with decreasing the light quark mass .
the @xmath26 and @xmath27 transition potential are consistent with zero within error bar . on the other hand ,
it is noteworthy that the @xmath28 transition potential which is not allowed in the su(3 ) symmetric world is strengthen as the su(3)@xmath10 breaking gets larger .
we have investigated the @xmath29 bb state , which is known as the @xmath9 , @xmath12 and @xmath13 coupled state , from lattice qcd .
we have found a small transition potential between the singlet and octet state in terms of the su(3 ) ir basis .
such transition can not be allowed in the su(3 ) symmetric world .
this method could greatly assist us to complete the knowledge of not only the generalized nuclear force but also the interaction of hadrons including mesons , baryons and quarks .
* acknowledgements * : this work was supported by the large scale simulation program no.0923(fy2009 ) of high energy accelerator research organization ( kek ) , grant - in - aid of the ministry of education , science and technology , sports and culture ( nos . 20340047 , 22540268 , 19540261 ) and the grant - in - aid for scientific research on innovative areas ( no .
2004:20105001 , 20105003 ) .
k. murano , n. ishii , s. aoki and t. hatsuda , pos * lattice2009 * ( 2009 ) 126 .
y. ikeda et al . , arxiv:1002.2309 [ hep - lat ] .
t. inoue et al .
[ hal qcd collaboration ] , arxiv:1007.3559 [ hep - lat ] . c. michael , nucl .
b * 259 * ( 1985 ) 58 .
m. luscher and u. wolff , nucl .
b * 339 * ( 1990 ) 222 . | we investigate baryon - baryon interactions with strangeness @xmath0 and isospin @xmath1 system from lattice qcd . in order to solve this system ,
we prepare three types of baryon - baryon operators ( @xmath2 , @xmath3 and @xmath4 ) for the sink and construct three source operators diagonalizing the @xmath5 correlation matrix . combining of the prepared sink operators with the diagonalized source operators , we obtain nine effective nambu - bethe - salpeter ( nbs ) wave functions .
the @xmath5 potential matrix is calculated by solving the coupled - channel schrdinger equation .
the flavor @xmath6 breaking effects of the potential matrix are also discussed by comparing with the results of the @xmath6 limit calculation .
our numerical results are obtained from three sets of @xmath7 flavor qcd gauge configurations provided by the cp - pacs / jlqcd collaborations .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
the credit valuation adjustment ( cva ) is , by definition , the difference between the risk - free portfolio value and the true portfolio value that takes into account default risk of the counterparty . in other words , cva is the market value of counterparty credit risk .
after the financial crisis in 2007 - 2008 , it has been widely recognized that even major financial institutions may default .
therefore , the market participants has become fully aware of counterparty credit risk . in order to reflect the counterparty credit risk in the price of over - the - counter ( otc ) derivative transactions ,
cva is widely used in the financial institutions today .
although duffie - huang @xcite has already introduced the basic idea of cva in 1990 s , several people reconsidered the theory of cva related to collateralized derivatives ( cf .
@xcite ) and also efficient numerical calculation methods appeared(cf .
@xcite ) .
there are two approaches to measuring cva : unilateral and bilateral ( cf .
@xcite ) . under the unilateral approach
, it is assumed that the bank that does the cva analysis is default - free .
cva measured in this way is the current market value of future losses due to the counterparty s potential default .
the problem with unilateral cva is that both the bank and the counterparty require a premium for the credit risk they are bearing and can never agree on the fair value of the trades in the portfolio .
therefore , we have to consider not only the market value of the counterparty s default risk , but also the bank s own counterparty credit risk called debit value adjustment ( dva ) in order to calculate the correct fair value . bilateral cva ( it is calculated by netting unilateral cva and dva ) takes into account the possibility of both the counterparty default and the own default .
it is thus symmetric between the own company and the counterparty , and results in an objective fair value calculation .
mathematically , unilateral cva and dva are calculated in the same way , and bilateral cva is the difference of them .
so we focus on the calculation of unilateral cva in this paper .
cva is measured at the counterparty level and there are many assets in the portfolio generally .
therefore , we have to be involved in the high dimensional numerical problem to obtain the value of cva .
this is one of the reasons why cva calculation is difficult . on the other hand
, each payoff usually depends only on a few assets .
we will focus on this property and suggest an efficient calculation methods of cva in the present paper .
let us consider the portfolio consist of the contracts on one counterparty .
let @xmath0 be @xmath1-valued stochastic processes @xmath2 we think that @xmath3 is an underlying process .
we consider the model that the macro factor is determined by @xmath4 and the payoff of each derivative at maturity @xmath5 is the form of @xmath6 let @xmath7 be the final maturity of all the contracts in the portfolio .
let @xmath8 be the default time of the counterparty , @xmath9 be its hazard rate process , @xmath10 be the process of loss when the default takes place at time @xmath11 , and @xmath12 be the discount factor process from @xmath11 to @xmath13 .
we assume that @xmath14 is the function of @xmath15 and that @xmath16 and @xmath17 are the function of @xmath18 let @xmath19 be total value of all contracts in the portfolio at time @xmath11 under the assumption that counterparty is default free .
then @xmath19 is given by @xmath20,\ ] ] where @xmath21 denotes the expectation with respect to the risk neutral measure .
then unilateral cva on this portfolio is the restructuring cost when the counterparty defaults .
so unilateral cva is given by @xmath22 \nonumber\\ & = e[\int_0^t l(t)\exp(-\int_0^t \lambda(s)ds ) \lambda(t)d(0,t ) ( \tilde{v}_0(t ) \vee 0 ) dt ] \nonumber\\ & = e[\int_0^t l(t)\exp(-\int_0^t \lambda(s)ds ) \lambda(t ) ( v_0(t ) \vee 0 ) dt ] \label{defcva},\end{aligned}\ ] ] where @xmath23,\ ] ] and @xmath24 is a function such as @xmath25 since @xmath26 is a function of @xmath27 we denote it by @xmath28 then cva is given by the following form .
@xmath29 \vee 0 ) dt ] \label{defcva2}.\end{aligned}\ ] ] now we prepare the mathematical setting .
let @xmath30 be fixed , @xmath31 , @xmath32 @xmath33 and @xmath34 .
+ let @xmath35 @xmath36 @xmath37 be the borel algebra over @xmath35 and @xmath38 be the wiener measure on @xmath39 let @xmath40 @xmath41 be given by @xmath42 @xmath43 then @xmath44 is a @xmath45-dimensional brownian motion .
let @xmath46 @xmath47 let @xmath48 @xmath49 @xmath50 here @xmath51 denotes the space of @xmath52-valued smooth functions defined in @xmath53 whose derivatives of any order are bounded .
we regard elements in @xmath54 as vector fields on @xmath55 now let us consider the following stratonovich stochastic differential equations .
@xmath57 where @xmath58 + let @xmath59 and @xmath60 be @xmath61 then we have @xmath62 there is a unique solution @xmath63 to this equation
. then @xmath64 also satisfies the solution to the following stratonovich stochastic differential equation . @xmath65 where @xmath66 is @xmath67 we assume that vector fields @xmath68 satisfy condition ( ufg ) stated in the section [ vec ] .
let @xmath69 be defined by ( [ defe ] ) in section [ vec ] . by @xcite ,
if @xmath70 the distribution law of @xmath71 under @xmath38 has a smooth density function @xmath72 for @xmath73 + let @xmath74 we assume that the underlying asset process is @xmath75 .
we also assume that @xmath76 let @xmath77 denotes the space of functions on @xmath52 given by @xmath78 where @xmath79 .
+ @xmath80 denotes the space of lipschitz continuous functions on @xmath81 and we define a semi - norm @xmath82 on @xmath83 by @xmath84 let @xmath85 be the linear subspace of @xmath80 spanned by @xmath86 + + we define linear operators @xmath87 by @xmath88 , \quad f\in { lip}({\bf r}^n),\ ] ] and @xmath89 by @xmath90 , \quad f \in { lip}({\bf r}^{\tilde{n}_m}).\ ] ] we remind that @xmath91 is represented by @xmath92 we assume that @xmath93 \times { \bf r}^n \to [ 0,\infty)$ ] satisfies the following two conditions .
+ ( 1 ) @xmath94 is differentiable in @xmath11 and there is an integer @xmath95 and a constant @xmath96 such that @xmath97}| \frac{\partial}{\partial t}g(t , x)| \leqq c_1(1+|x|^{n_1 } ) , \quad x \in { \bf r}^n.\ ] ] ( 2 ) @xmath94 is @xmath98-times coninuously differentiable in @xmath99 and there is an integer @xmath100 and a constant @xmath101 such that @xmath97 } |\frac{\partial^{\alpha}}{\partial x^{\alpha}}g(t , x)| \leqq c_2(1+|x|^{n_2}),\quad x \in { \bf r}^n\ ] ] for any multi index @xmath102 + we assume that a discounted payoff functions @xmath103 in equation ( [ defcva2 ] ) belong to @xmath104 under the assumptions above , cva @xmath105 is given by @xmath106 \vee0\}dt ] .\end{aligned}\ ] ] we will introduce numerical calculation methods by monte carlo simulation for @xmath105 .
+ + let @xmath107 be a probability space , and @xmath108 @xmath109 be continuous stochastic processes such that each probability law on @xmath110 of @xmath111 under @xmath112 is the same as that of @xmath113 for all @xmath114 @xmath115 and that @xmath116 @xmath109 are independent .
+ let us define projections @xmath117 by @xmath118 and define @xmath119 by @xmath120 we define random linear operators ( stochastic mesh operators ) @xmath121 on @xmath122 by @xmath123 let @xmath124 denotes the set of partitions @xmath125 such that @xmath126 and that @xmath127 let @xmath128 we define estimators @xmath129 in the following .
@xmath130 . \label{app2}\end{aligned}\ ] ] our main results are following .
[ main2 ] let @xmath131 let @xmath132 be a sequence and suppose that there is a constant @xmath133 such that @xmath134 then there exists a constant @xmath135 such that @xmath136 \leqq c_1 ( l^{-\frac{1}{1+\alpha_0}}+|\delta|)\ ] ] for any @xmath137 and @xmath138 [ main1 ] let @xmath139 and let @xmath132 be a sequence such that there is a constant @xmath140 such that @xmath141 suppose that there are constants @xmath142 $ ] and @xmath143 such that @xmath144 for all @xmath145.$ ] + then then there exists a constant @xmath146 and @xmath147 @xmath148 such that @xmath149 and @xmath150 for any @xmath151 and @xmath138 [ actual ] let @xmath152 be @xmath153 then by theorem [ main2 ] , we see that @xmath154 and @xmath155 theorem [ main1 ] shows that the estimation of @xmath156 may be better than @xmath157 .
we can compute the estimators @xmath158 practically in the following way .
+ first , we generate a family of independent paths @xmath159 next , by using @xmath160 , we compute @xmath161 then our estimator @xmath162 is @xmath163 we used the same paths for monte carlo simulation and construction of stochastic mesh operator . for @xmath164 , we generate another independent family of independent paths @xmath165 and we compute @xmath166 in the above computations of @xmath162 and @xmath164 , we use the values of @xmath167 , only .
+ as for the computation of @xmath168 we do not use @xmath169 explicitly .
we use @xmath169 only to judge whether @xmath170 or not .
so the approximation has no error when the signs of @xmath171 and @xmath172 are the same , even if there are large differences between them .
let @xmath173 and @xmath174 for @xmath175 , let @xmath176 if @xmath177 @xmath178 @xmath179 and let @xmath180 @xmath181 also , for each @xmath182 @xmath183 we define vector fields @xmath184},$ ] @xmath185 inductively by @xmath186 } = v_i , \quad i = 0,1,\ldots , d,\ ] ] @xmath187 } = [ v_{[\alpha ] } , v_i ] , \qquad i = 0,1,\ldots , d.\ ] ] here @xmath188 @xmath189 for @xmath177 @xmath178 and @xmath190 we assume that a system of vector fields @xmath191 satisfies the following condition ( ufg ) .
( ufg ) there is an integer @xmath192 and there are functions @xmath193 @xmath194 @xmath195 satisfying the following .
@xmath196 } = \sum_{\beta \in { \cal a}^{*}_{\leqq \ell_0 } } \varphi_{\alpha,\beta}v_{[\beta ] } } , \qquad \alpha \in { \cal a}^{*}.\ ] ] a system of vector fields @xmath197 also satisfies the @xmath198 condition . _ proof .
_ we prove following by induction on @xmath199 @xmath200 } ( f \circ \pi_m ) = ( \tilde{v}^{(m)}_{[\alpha ] } f)\circ \pi_m , \quad f \in c_b^{\infty}({\bf r}^{\tilde{n}_m } ) , \end{aligned}\ ] ] for any @xmath201 and @xmath202 + it is trivial in the case of @xmath203 by the assumption for induction , @xmath204 } ( f \circ \pi_m ) = ( { v}_{[\alpha ] } v_i- v_i { v}_{[\alpha ] } ) ( f \circ \pi_m)\ ] ] @xmath205 } ( ( \tilde{v}^{(m)}_{i } f)\circ \pi_m ) - v_i ( ( \tilde{v}^{(m)}_{[\alpha]}f ) \circ \pi_m)\ ] ] @xmath206}(\tilde{v}^{(m)}_{i } f ) ) \circ \pi_m - ( \tilde{v}^{(m)}_{i } ( \tilde{v}^{(m)}_{[\alpha]}f ) ) \circ \pi_m.\ ] ] so we have ( [ inductive ] ) . from ( ufg ) condition , we have @xmath207 } ( f \circ \pi_m)= \sum _ { \beta \in { \cal a}^{*}_{\leqq \ell_0 } } \varphi _ { \alpha , \beta } { v}_{[\beta ] } ( f \circ \pi_m)\ ] ] @xmath208}^{(m)}f ) \circ \pi_m.\ ] ] let @xmath209 be @xmath210 then @xmath211}f=(v_{[\alpha]}^{(m)}f)\circ \pi_m \circ j_m\ ] ] @xmath212}^{(m)}f ) \circ \pi_m ) \circ j_m = \sum _ { \beta \in { \cal a}^{*}_{\leqq \ell_0 } } ( \varphi _ { \alpha , \beta } \circ j_m ) \tilde{v}_{[\beta]}^{(m)}f.\ ] ] so we have our assertion .
+ + let @xmath213 @xmath214 @xmath215 be a @xmath216 symmetric matrix given by @xmath217}(\tilde{x}_m)\tilde{v}^j_{m,[\alpha ] } ( \tilde{x}_m ) , \qquad i , j=1,\ldots , \tilde{n}_m.\ ] ] let @xmath218 and @xmath219 by @xcite , we see that if @xmath220 the distribution law of @xmath71 under @xmath38 has a smooth density function @xmath221 for @xmath222 moreover , by @xcite we see that @xmath223 @xmath224 we have @xmath225 by @xcite .
in this section , we use the notation in @xcite . we have the following lemma similarly to the proof of @xcite lemma 8 ( 3 ) . [ rev ] for any @xmath226 let @xmath227 and @xmath228 then @xmath229}f ) ( x(t , x ) ) ] = t^{-\|\alpha\| / 2}e^{\mu}[\phi_{\alpha}(t , x ) f(x(t , x))],\ ] ] and @xmath97 , x \in { \bf r}^n , p\in ( 1 , \infty ) } e[|\phi_{\alpha}(t , x)|^p ] < \infty.\ ] ] + let @xmath230 be a smooth function such that @xmath231 let @xmath232 and @xmath233 be @xmath234 then for any @xmath235 @xmath236 [ rev2 ] if @xmath237 then @xmath238 _ proof .
_ let @xmath239 then for any @xmath235 @xmath240 and @xmath241 we have @xmath242 then @xmath243 is a cauchy sequence in @xmath244 because @xmath245 \}}\|d\phi(t , x)\|_h , \quad n\geqq m\geqq 1.\end{aligned}\ ] ] because @xmath246 is a closed operator , we have @xmath247 for any @xmath248 .
so we have the assertion .
+ let us denote @xmath249 [ disclem2 ] let @xmath250 .
then there exists a @xmath251 such that @xmath252\\ & \leqq c \|\nabla f\|_{\infty } \int_{s}^t(r^{-1/2}+(t - r)^{-1/2})dr , \end{aligned}\ ] ] for any @xmath253 and any @xmath254 .
_ let @xmath255 be @xmath256=(p_{t - t}f)(x(t , x^*)).\ ] ] @xmath255 is a @xmath257-martingale .
let @xmath258 let @xmath259 by it formula , @xmath260 note that , @xmath261 @xmath262 @xmath263 and @xmath264 so we have @xmath265\ ] ] @xmath266 dr\ ] ] @xmath267dr+ \sum_{j=1}^d \int_s^t e^{\mu}[|(v_jg)(v_j ( p_{t - r}f))(x(r , x^*))|]dr.\ ] ] now by the definition of @xmath268 and @xmath269 , we have , @xmath270dr \leqq \int_s^t e^{\mu}[|m(r)|^2]^{1/2}e^{\mu}[|(l_t g)(x(r , x^*))|^2]^{1/2}dr\ ] ] @xmath271}e^{\mu}[|(l_t g)(x(r , x^*))|^2]^{1/2 } \int_s^t e^{\mu}[|m(r)|^2]^{1/2}dr.\ ] ] by burkholder s inequality , @xmath272^{1/2}dr \leqq e^{\mu}[\langle m \rangle_t]^{1/2 } ( t - s ) \leqq \sup_{r\in ( s , t ) } \|v_j ( p_{t - r}f)\|_{\infty}(t - s).\ ] ] on the other hand , we have , @xmath273dr\ ] ] @xmath274dr\ ] ] @xmath275 for any @xmath253 and any @xmath254 .
+ on the other hand @xmath276 @xmath277 @xmath278 notice that @xmath279 we have @xmath280\ ] ] @xmath281\ ] ] @xmath282 where @xmath283,\ ] ] @xmath284.\ ] ] let @xmath285 then by lemma [ rev2 ] , @xmath286 let @xmath287 be defined by the formula of lemma [ rev ] . then we have @xmath288,\ ] ] and @xmath289 , x\in { \bf r}^n } e^{\mu}[|\phi_{g , i}(t , x)|^p ] < \infty.\ ] ] then there exists a constant @xmath251 such that @xmath290 also we have @xmath291\|v_j^2 ( p_{t - r}f)\|_{\infty},\ ] ] for any
@xmath253 and any @xmath292 . + + let vector field @xmath293 be represented by @xmath294 then we have @xmath295 where @xmath296 and @xmath297.\ ] ] moreover , we have @xmath298 then by corollary 9 of @xcite , since @xmath299 and there is a constant @xmath251 such that @xmath300 and @xmath301 for any @xmath302 and any @xmath292 .
+ + so we have @xmath303\ ] ] @xmath304 letting @xmath305 we have our assertion .
let @xmath250 .
there exists a constant @xmath306 such that @xmath307\ ] ] @xmath308 for any @xmath309 and any @xmath254 .
_ for @xmath310 , there exists @xmath311 such that @xmath312 and @xmath313 , for any @xmath314 .
so we obtain the result from lemma [ disclem2 ] .
+ [ disclem1 ] let @xmath315 and @xmath316 there exists a constant @xmath306 such that @xmath317\nonumber \\
\leqq & c(\|\nabla h\|_{\infty}+ \|\nabla^2 h\|_{\infty } ) ( t - t),\label{disc1}\\ \text{and } \nonumber\\ & e^{\mu } [ |g(t , x(t , x))(p^{(m)}_{t - t}(h\vee 0))(\tilde{x}^{(m)}(t,\tilde{x}_m^*))-(h\vee 0)(\tilde{x}^{(m)}(t,\tilde{x}_m^*))| ] \nonumber \\ \leqq & c(\|\nabla h\|_{\infty}+
\|\nabla^2 h\|_{\infty } ) ( t - t ) \label{disc2}.\end{aligned}\ ] ] for any @xmath318 _ proof . _ ( [ disc1 ] ) follows from it s formula .
so we show ( [ disc2 ] ) .
+ let @xmath319 are as defined in ( [ defphi ] ) .
let @xmath320 by it s formula @xmath321
@xmath322 @xmath323 @xmath324 so we have @xmath325-\bar{\varphi}_k(h(\tilde{x}^{(m)}(t,\tilde{x}_m^*))\ ] ] @xmath326 ds\ ] ] @xmath327 ds\ ] ] notice that @xmath328 then we have @xmath329-\bar{\varphi}_k(h(\tilde{x}^{(m)}(t,\tilde{x}_m^*))|]\ ] ] @xmath330\|\tilde{l}_m h\|_{\infty}(t - t)\ ] ] @xmath331 ds.\ ] ] on the other hand , we have @xmath332 @xmath333 @xmath334 + let @xmath335 and @xmath336 be defined by the formula of lemma [ rev ] .
then it follows that @xmath337\|_{\infty}\ ] ] @xmath338 \|(\varphi_k \circ h ) \tilde{v}^{(m)}_{i } h \|_{\infty } \leqq c s^{-1/2 } \| \tilde{v}^{(m)}_{i } h \|_{\infty},\ ] ] and @xmath339 \|_{\infty } \leqq c\| ( \tilde{v}^{(m)}_{i})^2 h \|_{\infty}.\ ] ] so we have @xmath340 ds\ ] ] @xmath341 @xmath342 letting @xmath343 we have the assertion .
let @xmath344 and @xmath345 .
there exists a constant @xmath251 such that @xmath346 \leqq c(t - t),\end{aligned}\ ] ] for any @xmath347 _ proof . _
notice that @xmath348 and lemma [ disclem1 ] is valid for @xmath349 . on the other hand , for @xmath350
, we have the expression that @xmath351 @xmath352 so our assertion follows from lemma [ disclem1 ] .
let @xmath353 be given by @xmath354.\ ] ] let @xmath355 be such that @xmath356 then we have @xmath357 is as follows .
@xmath358\ ] ] let @xmath359 @xmath360 be sub @xmath361-algebra of @xmath37 given by @xmath362 , \ \ell = 1,2,\ldots \ } .\ ] ] + [ time disc ] there exists a constant @xmath306 such that @xmath363 .
_ let @xmath364 then by lemma [ disclem2 ] , there is a constant @xmath251 such that + @xmath365 \\ - & e^{\mu } [ g(t_i , x(t_i , x^*))(p_{t_{k}-t_i } \tilde{f}_{k})(x(t_i , x^ * ) ) \vee 0]| dt\\ \leqq & c\sum_{k=1}^k \sum_{i = i_{(k-1)}}^{i_{(k)}-1 } \int_{t_i}^{t_{i+1 } } dt \int_{t_i}^t ( r^{-1/2}+(t_k - r)^{-1/2})dr\\ \leqq & c|\delta| \sum_{k=1}^k \sum_{i = i_{(k-1)}}^{i_{(k)}-1}\int_{t_i}^{t_{i+1 } } ( r^{-1/2}+(t_k - r)^{-1/2})dr . \end{aligned}\ ] ] so we have @xmath366 so the assertion follows .
to estimate the stochastic mesh operator , we use the following estimation of transition kernel @xmath367 obtained by proposition 8 of @xcite .
[ heatkernel ] let @xmath368 be given by @xmath369 then for any @xmath370 and @xmath315 there is a @xmath251 such that @xmath371 , \ x , y \in e_m,\ ] ] and @xmath372 , \ x , y \in e_m.\ ] ] in particular , for any @xmath373 and @xmath374 there is a @xmath251 such that @xmath375 , \ x , y \in e_m.\ ] ] let @xmath376 from proposition 13 , 21 and proposition 15 ( 1 ) of @xcite , we have the followings .
[ basic ] let @xmath377 and @xmath378 then we have @xmath379 = ( p^{(m)}_{s , t}f)(x ) , \qquad \nu_s^{(m)}-a.e.x \in e_m.\ ] ] and @xmath380 \leqq \frac{1}{l } \int_{e_m } \frac{p^{(m)}(t - s , x , y)^2 |f(y)|^2}{q_{s , t}^{(m , l,\omega)}(y)}\ ; dy .\ ] ] [ key ] let @xmath381 then there exists a @xmath382 such that @xmath383\right)^{1/2 } \\ & \leqq c l^{-(1-\delta)/2}(t_k - t)^{-(1+\delta)(\tilde{n}+1)\ell_0/4 } ( \int_{e_m } f(y)^2 ( 1+|y|^2)^{-\tilde{n}_m}dy)^{1/2}.\end{aligned}\ ] ] for any @xmath384 any @xmath315 and any @xmath385 [ exset ] let @xmath386 @xmath387 } z_l^{(m , k)}(s;\delta).\end{aligned}\ ] ] then we have the followings .
+ @xmath388 for any @xmath381 , and @xmath248 , there is a @xmath389 such that @xmath390^{1/p}\leqq c_{p,\delta}\varepsilon^{-5\ell_0}l^{-p\delta^2/2 + 1/p}\ ] ] for any @xmath391 , k=1,\ldots , k,$ ] and @xmath392 + @xmath393 let @xmath394 and @xmath395 if @xmath396 , and @xmath397 , then @xmath398 for any @xmath399 , \quad k=1,\ldots , k.$ ] and @xmath392 now we introduce the following sets and functions .
let @xmath400 @xmath401 be given by @xmath402 and @xmath403 let @xmath404 \times e \times \omega \to [ 0 , \infty ) , i=1,2,3,$ ] be the measurable functions given by @xmath405 let @xmath406 be the transition kernel of @xmath407 [ d12 ] let @xmath381 .
then there exists a constant @xmath382 such that @xmath408|g(t , x)| p(t , x^ * , dx ) \leqq c l^{-(1-\delta)^3 } 1 _ { [ 0 , t_k- \varepsilon ) } ( t ) , \label{d1 } \\ & ( \int_{e_m } e^p[\ d_{2,\varepsilon , l}^{(m , k)}(t , x)^2 ] p(t , x^ * , dx))^{1/2}\nonumber \\
\leqq & c l^{-(1-\delta)/2 } ( t_k - t)^{-(1+\delta)(\tilde{n}+1)\ell_0/4 } 1 _ { [ 0 , t_k- \varepsilon ) } ( t),\label{d2}\\ \text{and } \nonumber\\ & \int_{e_m } d_{3,\varepsilon , l}^{(m , k)}(t , x ) |g(t , x)| p(t , x^ * , dx ) \leqq c ( t_k - t)1_{[t_k - \varepsilon , t_k ) } ( t ) . \label{d3}\end{aligned}\ ] ] for any @xmath409 , l \geqq 1,$ ] @xmath315 and @xmath410 _ proof .
_ equation ( [ d3 ] ) follows from lemma [ disclem1 ] .
so we will show ( [ d1 ] ) and ( [ d2 ] ) . note that if @xmath411 , both sides are @xmath412 in ( [ d1 ] ) and ( [ d2 ] ) .
so we will consider the case @xmath413 . by proposition [ basic ] , we have @xmath414 |g(t , x)| p(t , x^ * , dx)\\ = & \int_{e } e^p[|(q_{t , t_k , \varepsilon}^{(m)}(1-\varphi_{m , k , l})f_{m , k})(x ) -(p_{t_k - t}^{(m ) } ( 1-\varphi_{m , k , l})f_{m ,
k})(\pi_m(x ) ) | ] |g(t , x)| p(t , x^ * , dx ) \\ \leqq & \int_{e } ( e^p[(q_{t , t_k , \varepsilon}^{(m ) } |(1-\varphi_{m , k , l})f_{m , k}| ) ( \pi_m(x))]\\ & \qquad \qquad \qquad \qquad \qquad \qquad+ ( p_{t_k - t}^{(m ) } |(1-\varphi_{m , k , l})f_{m , k}| ) ( \pi_m(x ) ) ) |g(t , x)| p(t , x^ * , dx)\\ \leqq & 2 \int_{e } ( p_{t_k - t}^{(m ) } ( 1-\varphi_{m , k , l})|f_{m , k}|)(\pi_m(x ) ) g(t , x ) p(t , x^ * , dx).\end{aligned}\ ] ] using hlder s inequality for @xmath415 @xmath416 @xmath417 @xmath418 @xmath419 @xmath418 @xmath420 @xmath421 we used @xmath422 in the last inequality .
so we have equation ( [ d1 ] ) .
+ next we will show equation ( [ d2 ] ) . noting that from @xmath423 and @xmath137 , @xmath424 since propsition [ exset ] , @xmath425 . and by proposition [ basic ] , we have @xmath426\\ \leqq & 1_{b^{(m , k)}(t , \delta , l ) } \frac{1}{l } \int_{e_m } \frac{|\varphi_{m , k , l}f_{m ,
k}(\tilde{y}_m)|^2p^{(m)}(t_k - t , \tilde{x}_m,\tilde{y}_m)^2}{q_{t , t_k}^{(m , l,\omega)}(\tilde{y}_m)}d\tilde{y}_m\\ \leqq & \frac{2}{l } \int_{e_m } \frac{|\varphi_{m , k , l}f_{m , k}(\tilde{y}_m)|^2}{p^{(m)}(t_k , \pi_m(x^*),\tilde{y}_m ) } p^{(m)}(t_k - t , \tilde{x}_m,\tilde{y}_m)^2 dy.\end{aligned}\ ] ] then we have @xmath427 p(t , x^ * , dx))^{1/2 } \\ \leqq & ( \int_{e } e^p[1_{b^{(m , k)}(t , \delta , l)}e^p [ |(q_{t , t_k , \varepsilon}^{(m)}\varphi_{m , k , l}f_{m , k})(x ) -(p_{t_k - t}^{(m ) } \varphi_{m , k , l}f_{m , k})(x ) |^2| \mathcal{f}_t ] ] p(t , x^ * , dx))^{1/2}\\ \leqq & ( \frac{2}{l } \int_{e_m } \frac{|\varphi_{m , k , l}f_{m , k}(y)|^2}{p^{(m)}(t_k , \pi_m(x^*),\tilde{y}_m ) } ( \int_{e } p^{(m)}(t_k - t , \tilde{x}_m,\tilde{y}_m)^{(1-\delta)+(1+\delta ) } p(t , x^ * , dx))d\tilde{y}_m)^{1/2}\\ \leqq & ( \frac{2}{l } \int_{e_m } \frac{|\varphi_{m , k , l}f_{m , k}(y)|^2}{p^{(m)}(t_k , \pi_m(x^*),\tilde{y}_m)^{\delta } } ( \int_{e_m } p^{(m)}(t , \pi_m(x^ * ) , \tilde{x}_m ) p^{(m)}(t_k - t , \tilde{x}_m,\tilde{y}_m)^{(1+\delta)/\delta}d\tilde{x}_m)^{\delta}d\tilde{y}_m)^{1/2}.\end{aligned}\ ] ] let @xmath428 from lemma [ heatkernel ] , there exists a constant @xmath251 such that @xmath429 we set @xmath430 as @xmath431}\max_{\substack{m=1,\ldots , m,\\ k=1,\ldots , k}}(\int_{e_m } h_m(x)^{-(\tilde{n}_m+1)\ell_0(1+\delta)/\delta}(1+|x|^2)^{q(1+\delta)/\delta}p^{(m)}(t , \tilde{x}_m^*,x)dx)^{\delta/2}\\ & \quad \quad \times ( \int_e|g(t , x)|dx)^{1/2}.\end{aligned}\ ] ] @xmath430 is bounded by proposition 3 of @xcite . then since @xmath432 , we have @xmath433^{1/2}|g(t , x)| p(t , x^ * , dx)\\ \leqq & \frac{c_1}{l } \int_{e_m } p^{(m)}(t_k , \tilde{x}_m^*,\tilde{y}_m)^{-\delta } |\varphi_{m , k , l}f_{m , k}(\tilde{y}_m)|^2(1+|\tilde{y}_m|^2)^{-q(1+\delta)}d\tilde{y}_m ( t_k - t)^{-(1+\delta)(n+1)\ell_0/2 } , \\
\leqq & c_1l^{-(1-\delta)}(t_k - t)^{-(1+\delta)(\tilde{n}_m+1)l_0/2 } \int_{e_m } |f_{m , k}(\tilde{y}_m)|^2(1+|\tilde{y}_m|^2)^{-q(1+\delta)}d\tilde{y}_m .\end{aligned}\ ] ] since @xmath434 and @xmath24 is lipschitz continuous , @xmath435 so we have the assertion . +
let @xmath436 and @xmath437 let @xmath438 and @xmath439 be @xmath440 [ dd ] there exists a constant @xmath382 such that @xmath441 and , @xmath442 for any @xmath384 .
_ let us take @xmath443 as @xmath444 if @xmath445 $ ] and @xmath446 then @xmath447 so notice that @xmath448 so we have @xmath449 because @xmath450 for @xmath451 .
+ on the other hand , @xmath452 and @xmath453 so we have equation ( [ sum1 ] ) .
+ next we show equation ( [ sum2 ] ) .
@xmath454 where @xmath455 @xmath456 @xmath457 @xmath458 note that ( [ disceps ] ) and @xmath459 we have for @xmath460 , @xmath461 @xmath462 we have the followings similarly .
@xmath463 so we obtain ( [ sum2 ] ) .
[ convergence ] there exists a constant @xmath306 such that @xmath464 \leqq c\left ( l^{-(1-\delta)/2 } \hat{e}\left(\varepsilon , ( 1+\delta)(\tilde{n}+1)\ell_0/4 \right ) + \varepsilon^2\right ) , l\geqq 1.\ ] ] _ proof .
_ @xmath464\ ] ] @xmath465 @xmath466|.\ ] ] then by schwartz s inequality , @xmath467|\ ] ] @xmath468\ ] ] @xmath469 @xmath470 \right)^{1/2 } 1_{[0 , t_k-\varepsilon)}\ ] ] @xmath471^{1/2 } + c(t_k - t)1_{[t_k-\varepsilon , t_k)}\ ] ] by proposition [ key ] @xmath464\ ] ] @xmath472 by proposition[dd ] , we have the assertion .
[ improve ] let @xmath473 and @xmath474 .
then we have @xmath475 _ proof . _
if @xmath476 , then @xmath477 so we see that @xmath478 @xmath479 @xmath480 let @xmath481 suppose that there is @xmath142 $ ] and @xmath482 such that @xmath483 @xmath484 then there exists a constant @xmath251 , @xmath485 , such that @xmath486 and @xmath487 @xmath488 for @xmath489 _ proof .
_ in this proof , we denote @xmath490 by @xmath491 for simplicity .
let @xmath492 let @xmath493 be given by @xmath494 and let @xmath495 be given by @xmath496 then @xmath497 @xmath498|,\ ] ] @xmath499 since @xmath500 is @xmath501 measurable .
+ applying lemma [ improve ] to @xmath502 and @xmath503 we have @xmath504 where @xmath505,\\ & i_2=1_{\tilde{b}_{\varepsilon } } \sum_{i=0}^{n-1}(t_{i+1}-t_i ) \theta e^{\mu } [ |g(t_i , x(t_i))||1_{\{f_{p , i}(x(t_i)| \leqq \theta \}}].\end{aligned}\ ] ] by hlder s inequality , @xmath506^{\delta } \mu ( |f_{p , i}(x(t_i)| \leqq \theta)^{1-\delta } \leqq c \theta \mu ( |f_{p , i}(x(t_i)| \leqq \theta).\ ] ] from the assumption , we have @xmath507 next we will estimate @xmath508 .
@xmath509 @xmath510 .\end{aligned}\ ] ] @xmath508 is dominated by @xmath511 where @xmath512,\ ] ] @xmath513 @xmath514,\ ] ] @xmath515 @xmath516.\ ] ] @xmath517\ ] ] from proposition [ d12 ] , @xmath518=\sum_{i=0}^{n-1 } ( t_{i+1}-t_i ) \sum_{\substack{m=1,\ldots , m,\\ k ; t_k \geqq t_{i+1 } } } \int_{e } e^p[d_{1,\varepsilon , l}^{(m , k)}(t,\tilde{x}_m ) ] @xmath519 @xmath520 + next , we will estimate @xmath521 . by hlder s inequality @xmath522^{1/2}\ ] ] @xmath523^{\delta/2}e^{\mu}[1_{\{\sum_{m=1}^m \sum_{k : t_k\geqq t_{i+1 } } ( d_{1,\varepsilon , l}^{(m , k)}(t_i , \pi_m(x(t_i ) + d_{3,\varepsilon , l}^{(m , k)}(t_i , \pi_m ( x(t_i ) ) > \theta/2\}}]^{(1-\delta)/2},\ ] ] @xmath524^{1/2}\ ] ] @xmath525)^{(1-\delta)/2}.\ ] ] so we have @xmath526 \leqq c\sqrt{\frac{2}{\theta } } \sum_{i=0}^{n-1}(t_{i+1}-t_i ) e^p[\sum_{\substack{m=1,\ldots , m,\\k : t_k\geqq t_{i+1 } } } \int_{e_m } d_{2,\varepsilon , l}^{(m , k)}(t_i , x)^2 p^{(m)}(t_i , \tilde{x}_m^ * , x)dx ] ^{1/2}\\ & \times e^p [ \sum_{\substack{m=1,\ldots , m,\\k : t_k\geqq t_{i+1 } } } \int_{e_m } ( d_{1,\varepsilon , l}^{(m , k)}(t_i , x ) + d_{3,\varepsilon , l}^{(m , k)}(t_i , x ) ) p^{(m)}(t_i , \tilde{x}_m^ * , x)dx ] ^{(1-\delta)/2}\\ \leqq & c \theta^{-(1-\delta)/2 } \sum_{i=0}^{n-1}(t_{i+1}-t_i ) \left ( \sum_{k : t_k\geqq t_{i+1 } } \phi^{(k)}(t_i , \varepsilon ; l^{-(1-\delta)^2/2 } , ( 1-\delta^2)(\tilde{n}+1)\ell_0/4 , 0 , 0)\right ) \\ & \times \left(\sum_{k : t_k\geqq t_{i+1 } } \phi^{(k)}(t_i , \varepsilon ; l^{-(1-\delta)^4/2 } , 0 , 1 , ( 1-\delta)/2)\right).\end{aligned}\ ] ] by proposition [ dd ] , @xmath527\leqq c\theta^{-1/2 } \left ( l^{-(1-\delta)^4 } \hat{e}\left(\varepsilon , ( 1-\delta^2)(\tilde{n}+1)\ell_0/4\right ) + l^{-(1-\delta)^2/2}\varepsilon^{3(1-\delta)/2 } \right).\ ] ] similarly , we have @xmath528 \leqq c \theta^{-(1-\delta)/2 } \sum_{i=0}^{n-1}(t_{i+1}-t_i ) e[\sum_{\substack{m=1,\ldots , m,\\k : t_k\geqq t_{i+1 } } } \int_{e_m } d_{2,\varepsilon , l}^{(k)}(t_i , x)^2 p^{(k)}(t_i , \tilde{x}_k^ * , x)dx ] ^{(1-\delta)}\\ \leqq & c \theta^{-(1-\delta)/2 } \sum_{i=0}^{n-1}(t_{i+1}-t_i ) ( \sum_{k : t_k\geqq t_{i+1 } } \phi^{(k)}(t_i , \varepsilon ; l^{-(1-\delta)^2/2 } , ( 1-\delta^2)(\tilde{n}+1)\ell_0/4 , 0,0 ) ) ^2\\ \leqq & c\theta^{-1 } l^{-(1-\delta)^2 } \hat{e}\left(\varepsilon,(1-\delta^2)(\tilde{n}+1)\ell_0/2 \right).\end{aligned}\ ] ] it follows easily that @xmath529\leqq c\varepsilon^2.\ ] ] notice that @xmath530 we have @xmath531 \leqq c\left(\theta^{\gamma+1}+\theta^{-1 } l^{-(1-\delta)^2/2 } \left ( l^{-(1-\delta)^2/2 } \hat{e}\left(\varepsilon,(1-\delta^2)(\tilde{n}+1)\ell_0/2 \right)+\varepsilon^{3(1-\delta)/2 } \right ) \right).\ ] ] in particular if we take @xmath532 as @xmath533 then we have @xmath531 \leqq c\left ( l^{-(1-\delta)^2/2 } \left ( l^{-(1-\delta)^2/2 } \hat{e}\left(\varepsilon,(1-\delta^2)(\tilde{n}+1)\ell_0/2 \right)+\varepsilon^{3(1-\delta)/2 } \right ) \right)^{(1+\gamma)/(2+\gamma ) } .\ ] ] let @xmath534 be @xmath535 from proposition [ exset ] , we have @xmath536 @xmath537 and @xmath487 @xmath538 theorem [ main2 ] and theorem [ main1 ] follow from the theorem [ convergence ] and theorem [ improve ] .
let @xmath539 be 1 dimensional brownian motion .
let @xmath540 let @xmath541 be @xmath542\vee0 ) \ dt ] = \frac{2}{3 \sqrt{2\pi}}.\ ] ] let @xmath357 be the discretization of @xmath541 , such that @xmath543.\ ] ] we approximate @xmath541 as remark [ actual ] , where @xmath544 let @xmath545 be i.i.d sample paths of @xmath546 .
we compute @xmath547 and @xmath157 by using of paths @xmath160 .
@xmath548 let @xmath549 be another i.i.d sample paths of @xmath546 .
we compute @xmath156 by @xmath550 we have @xmath551 when we take @xmath552 we have @xmath553 we also take @xmath554 and @xmath555 we replicate 100 estimators of @xmath556 for each @xmath557 .
let `` average @xmath558 '' denote the average and `` standard dviation @xmath558 '' denote the unbiased standard deviation of these 100 estimators of @xmath559 we show the numerical result in table [ nume ] , we show graph of `` average @xmath560 '' and @xmath357 in figure [ fig1 ] and graph of `` standard deviation @xmath561 '' in figure [ fig2 ] . we see in figure [ fig1 ] that both @xmath157 and @xmath156 are close to @xmath357 , but we see in figure [ fig2 ] that @xmath156 is more stable than @xmath562 + 99 avramidis , a.n . , and p. hyden , efficiency improvements for pricing american options with a stochastic mesh , in proceedings of the 1999 winter simulation conference , pp . 344 - 350 broadie , m. , and p. glasserman , a stochastic mesh method for pricing high - dimensional american options j. computational finance , 7 ( 4 ) ( 2004 ) , 35 - 72 .
duffie , d. , m. huang , 1996 , swap rates and credit quality , journal of finance , vol .
51 , no . 3 , 921 fujii , f. , a. takahashi derivative pricing under asymmetric and imperfect collateralization , and cva carf working paper f-240 , december 2010 glasserman , p. , `` monte carlo methods in financial engineering '' springer , 2004 , berlin .
gregory , j. , counterparty credit risk and credit value adjustment : a continuing challenge for global financial markets , second editio , john wiley & sons , 2012 .
kusuoka , s. , malliavin calculus revisited , j. math .
tokyo 10(2003 ) , 261 - 277 .
kusuoka , s. , and d.w.stroock , applications of malliavin calculus ii , j. fac .
tokyo sect .
32(1985),1 - 76 .
kusuoka , s. , and y. morimoto , stochastic mesh methods for h " ormander type diffusion processes , preprint .
labordere , h . cutting cva s complexity , risk magazine , july , 2012 , pp .
liu , g. , and l.j .
hong , revisit of stochastic mesh method for pricing american options , operations research letters 37(2009 ) , 411 - 414 .
shigekawa , i. , `` stochastic analysis '' , translation of mathematical monographs vol.224 , ams 2000 . | in this paper , the author considers the numerical computation of cva for large systems by mote carlo methods .
he introduces two types of stochastic mesh methods for the computations of cva . in the first method ,
stochastic mesh method is used to obtain the future value of the derivative contracts . in the second method ,
stochastic mesh method is used only to judge whether future value of the derivative contracts is positive or not .
he discusses the rate of convergence to the real cva value of these methods .
jel classification : c63 , g12 mathematical subject classification(2010 ) 65c05 , 60g40 keywords : computational finance , option pricing , malliavin calculus , stochastic mesh method , cva |
the concept of entropy plays a central role in information theory , and has found a wide array of uses in other disciplines , including statistics , probability and combinatorics .
the _ ( differential ) entropy _ of a random vector @xmath6 with density function @xmath7 is defined as @xmath8 where @xmath9 .
it represents the average information content of an observation , and is usually thought of as a measure of unpredictability . in statistical contexts , it is often the estimation of entropy that is of primary interest , for instance in goodness - of - fit tests of normality @xcite or uniformity @xcite , tests of independence @xcite , independent component analysis @xcite and feature selection in classification @xcite .
see , for example , @xcite and @xcite for other applications and an overview of nonparametric techniques , which include methods based on sample spacings ( in the univariate case ) , histograms and kernel density estimates , among others .
the estimator of @xcite is particularly attractive as a starting point , both because it generalises easily to multivariate cases , and because , since it only relies on the evaluation of @xmath0th - nearest neighbour distances , it is straightforward to compute . to introduce this estimator ,
let @xmath10 be independent random vectors with density @xmath7 on @xmath2 .
write @xmath11 for the euclidean norm on @xmath2 , and for @xmath12 , let @xmath13 denote a permutation of @xmath14 such that @xmath15 .
for conciseness , we let @xmath16 denote the distance between @xmath17 and the @xmath0th nearest neighbour of @xmath17 .
leonenko estimator of the entropy @xmath18 is given by @xmath19 where @xmath20 denotes the volume of the unit @xmath21-dimensional euclidean ball and where @xmath22 denotes the digamma function .
in fact , this is a generalisation of the estimator originally proposed by @xcite , which was defined for @xmath23 . for integers
@xmath0 we have @xmath24 where @xmath25 is the euler mascheroni constant , so that @xmath26 as @xmath27 .
this estimator can be regarded as an attempt to mimic the ` oracle ' estimator @xmath28 , based on a @xmath0-nearest neighbour density estimate that relies on the approximation @xmath29 the initial purpose of this paper is to provide a detailed description of the theoretical properties of the kozachenko leonenko estimator .
in particular , when @xmath3 , we show that under a wide range of choices of @xmath0 ( which must diverge to infinity with the sample size ) and under regularity conditions , the estimator satisfies @xmath30 this immediately implies that in these settings , @xmath31 is efficient in the sense that @xmath32 the fact that the asymptotic variance is the best attainable follows from , e.g. , @xcite .
when @xmath33 , we show that no longer holds but the kozachenko
leonenko estimator is still root-@xmath1 consistent provided @xmath0 is bounded .
moreover , when @xmath34 , it turns out a non - trivial bias means that the rate of convergence is slower than @xmath35 in general , regardless of the choice of @xmath0 .
these results motivate our second main contribution , namely the proposal of a new entropy estimator , formed as a weighted average of kozachenko leonenko estimators for different values of @xmath0 .
we show that it is possible to choose the weights in such a way as to cancel the dominant bias terms , thereby yielding an efficient estimator in arbitrary dimensions , given sufficient smoothness .
there have been several previous studies of the kozachenko
leonenko estimator , but results on the rate of convergence have until now confined either to the case @xmath23 or ( very recently ) the case where @xmath0 is fixed as @xmath1 diverges .
the original @xcite paper proved consistency of the estimate under mild conditions in the case @xmath23 .
@xcite proved that the mean squared error of a truncated version of the estimator is @xmath36 when @xmath23 and @xmath37 under a condition that is almost equivalent to an exponential tail ; @xcite showed that the bias vanishes asymptotically while the variance is @xmath36 when @xmath23 and @xmath7 is compactly supported and bounded away from zero on its support . very recently , in independent work and under regularity conditions , @xcite derived the asymptotic normality of the estimator when @xmath23 , confirming the suboptimal asymptotic variance in this case .
previous works on the general @xmath0 case include @xcite , where heuristic arguments were presented to suggest the estimator is consistent for general @xmath21 and general fixed @xmath0 and has variance of order @xmath38 for @xmath37 and general fixed @xmath0 . the only work of which we are aware in which @xmath0 is allowed to diverge with @xmath1 is @xcite , where the estimator was shown to be consistent for general @xmath21 .
other very recent contributions include @xcite and @xcite . in this paper , we substantially weaken regularity conditions compared with these previous works , and show that with an appropriate choice of @xmath0 , efficient entropy estimators can be obtained in arbitrary dimensions , even in cases where the support of the true density is the whole of @xmath2 .
such settings present significant new challenges and lead to different behaviour compared with more commonly - studied situations where the underlying density is compactly supported and bounded away from zero on its support . to gain intuition , consider the following second - order taylor expansion of @xmath39 around a density estimator @xmath40 : @xmath41 when @xmath7 is bounded away from zero on its support
, one can estimate the ( smaller order ) second term on the right - hand side , thereby obtaining efficient estimators of entropy in higher dimensions @xcite ; however , when @xmath7 is not bounded away from zero on its support such procedures are no longer effective . to the best of our knowledge ,
therefore , this is the first time that a nonparametric entropy estimator has been shown to be efficient in multivariate settings for densities having unbounded support .
( we remark that when @xmath37 , the histogram estimator of @xcite is known to be efficient under fairly strong tail conditions . )
although the focus of this work is on efficient estimation , our results have other methodological implications : first , they suggest that prewhitening the data , i.e. replacing @xmath17 with @xmath42 before computing the estimator , where @xmath43 denotes the sample covariance matrix , can yield substantial bias reduction .
second , they yield asymptotically valid confidence intervals and hypothesis tests in a straightforward manner .
the outline of the rest of the paper is as follows . in section
[ sec : bias ] , we give our main results on the bias of kozachenko leonenko estimators , and present examples to understand the strength of the conditions .
we also discuss prewhitening the estimators to improve performance .
section [ sec : var ] is devoted to the study of the variance of the estimators and their asymptotic normality . in section [ sec : weighted ] , we introduce our weighted kozachenko
leonenko estimators , and show both how to choose the weights to achieve efficiency , and how to construct confidence intervals for entropy .
proofs of main results are presented in section [ sec : proof ] with auxiliary material and detailed bounds for various error terms deferred to the appendix .
we conclude the introduction with some notation used throughout the paper . for @xmath44 and @xmath45 , let @xmath46 be the closed euclidean ball of radius @xmath47 about @xmath48 .
we write @xmath49 , @xmath50 and @xmath51 for the operator norm , frobenius norm and determinant , respectively , of @xmath52 . for an @xmath53 dimensional array @xmath54
we write @xmath55 . for a smooth function @xmath56
, we write @xmath57 and @xmath58 for its first , second , and @xmath59 derivatives respectively , @xmath60 for its laplacian , and @xmath61 for its uniform norm . we write @xmath62 to mean that there exists @xmath63 , depending only on @xmath21 and @xmath7 , such that @xmath64 .
in this section we study the bias of the kozachenko leonenko estimator . for @xmath65 , we write @xmath66 so that @xmath67 to gain intuition about the bias , we introduce for @xmath44 and @xmath68 , the sequence of distribution functions @xmath69 where @xmath70 further , for @xmath68 , define the limiting distribution function @xmath71 where @xmath72 . that this is the limit distribution for each fixed @xmath0 follows from a poisson approximation to the binomial distribution .
we therefore expect that @xmath73 although we do not explicitly use this approximation in our asymptotic analysis of the bias , it motivates much of our development .
it also explains the reason for using @xmath74 in the definition of @xmath31 , rather than simply @xmath0 .
we will work with the following conditions : * ( a1)*(@xmath75 ) : : @xmath7 is bounded , and writing @xmath76 and @xmath77 , we have that @xmath7 is @xmath78 times continuously differentiable and there exist @xmath79 and a borel measurable function @xmath80 such that for each @xmath81 and @xmath82 , @xmath83 and @xmath84 as @xmath85 , for each @xmath86 . *
( a2)*(@xmath87 ) : : @xmath88 .
condition * ( a1)*(@xmath75 ) is discussed in section [ sec : assumptions ] below .
we are now in a position to state our main result on the bias of the kozachenko
leonenko estimator .
[ biasthm ] suppose * ( a1)*(@xmath75 ) and * ( a2)*(@xmath87 ) hold for some @xmath89 .
let @xmath90 denote any deterministic sequence of positive integers with @xmath91 as @xmath92 for some @xmath86 . then *
if @xmath93 and @xmath94 , we have @xmath95 uniformly for @xmath96 as @xmath92 . * if @xmath97 or @xmath98 with either @xmath99 or @xmath100 $ ] , then for every @xmath101 we have @xmath102 uniformly for @xmath96 , as @xmath92 .
it is important to note that the conditions * ( a1)*(@xmath75 ) with @xmath103 together with * ( a2)*(@xmath87 ) for some @xmath94 ensure that the integral in part ( i ) is finite when @xmath98 ; cf .
proposition [ prop : moment ] in section [ sec : biasproof ] .
the following corollary , which is a straightforward extension of theorem [ biasthm](i ) , simplifies the expression for the dominant contribution to the bias when @xmath27 as @xmath92 .
[ cor : easycor ] assume the conditions of theorem [ biasthm](i ) , and let @xmath104 denote any deterministic sequence of positive integers with @xmath105 and @xmath106 as @xmath92 .
then @xmath107 uniformly for @xmath108 as @xmath92 .
the proof of theorem [ biasthm ] is given in section [ sec : biasproof ] , but we present here some of the main ideas .
first , note that @xmath109 where @xmath110 denotes the density of a @xmath111 random variable at @xmath112 , with @xmath113 .
for @xmath114 and @xmath115 , define @xmath116 . since @xmath117 is a continuous , non - decreasing function of @xmath47
, we can define a left - continuous inverse for @xmath112 by @xmath118 so that @xmath119 if and only if @xmath120 .
we use the approximation @xmath121 for small @xmath122 , which is formalised in lemma [ lemma : hxinvbounds](ii ) in section [ sec : biasproof ] .
the proof of theorem [ biasthm](i ) can be seen as justifying the use of the above approximation in the following : @xmath123 note that @xmath124 , which is negligible compared with the main @xmath125 contribution . in case ( ii ) , the integral is not necessarily finite , so the approximation above is not as helpful , though the proof of the theorem is similar in the two cases .
these heuristics make clear that the function @xmath126 plays a key role in understanding the bias .
this function is in general complicated , though some understanding can be gained from the following simple example , where it can be evaluated explicitly , leading to an exact expression for the bias , even though * ( a1)*(@xmath75 ) does not hold for any @xmath127 .
consider the uniform distribution , @xmath128 $ ] .
for @xmath129 , we have @xmath130 it therefore follows that @xmath131 condition * ( a1)*(@xmath75 ) imposes smoothness and tail constraints on the unknown density @xmath7 . in particular
, it is reminiscent of more standard hlder smoothness conditions , though it also requires that the partial derivatives of the density vary less where @xmath7 is small . roughly speaking
, it also means that the partial derivatives of the density decay about as fast as the density itself in the tails of the distribution .
moreover , it can be shown that if * ( a1)*(@xmath75 ) holds for a density @xmath132 then it also holds for any density from the location - scale family : @xmath133 in fact , * ( a1)*(@xmath75 ) is satisfied by many standard families of distributions , as indicated by the following examples .
[ eg : normal ] consider the standard @xmath21-dimensional normal distribution , with density function @xmath134 then , for any @xmath135 , @xmath136 for some polynomial @xmath137 of degree @xmath138 with positive coefficients .
moreover , @xmath139 hence , @xmath140 since @xmath141 only when @xmath142 , and since @xmath143 as @xmath144 for all @xmath145 , we may therefore use taylor s theorem and the equivalence of norms on finite - dimensional vector spaces to see that * ( a1)*(@xmath75 ) is satisfied for any @xmath127 with @xmath146 and @xmath147 .
[ eg : t ] consider the standard @xmath21-dimensional @xmath138-distribution with @xmath148 degrees of freedom , with density function @xmath149 in this case , @xmath150 where @xmath137 is a polynomial of degree @xmath138 with positive coefficients .
moreover , @xmath151 it may also be shown that @xmath152 so we may use taylor s theorem to show that * ( a1)*(@xmath75 ) is satisfied for any @xmath127 with @xmath146 and @xmath80 taken be a sufficiently large constant ( which may depend on @xmath153 and @xmath21 ) .
although * ( a1)*(@xmath75 ) is rather general for densities with @xmath154 for all @xmath44 , a class of such densities for which the condition does not hold is those of the form @xmath155 for large @xmath156 , which have highly oscillatory behaviour in the tails . on the other hand , * ( a1)*(@xmath75 )
apparently precludes points @xmath157 with @xmath158 . to provide some guarantees on the performance of kozachenko leonenko estimators in settings where the density vanishes at certain points
, we now give a very general condition under which our approach to studying the bias can be applied .
[ prop : weakcond ] assume that @xmath7 is bounded , that * ( a2)*(@xmath87 ) holds for some @xmath159 , and let @xmath160 be as in theorem [ biasthm ] .
let @xmath161 , let @xmath162 , and assume further that there exists @xmath163 such that the function on @xmath164 given by @xmath165 is real - valued .
suppose that @xmath166 is such that @xmath167 as @xmath92 , where @xmath168
. then writing @xmath169 , we have for every @xmath101 that @xmath170 uniformly for @xmath96 . to aid interpretation of proposition [ prop : weakcond ] , we first remark that if * ( a1)(@xmath75 ) * holds , then so does , with @xmath171 , where @xmath172 is defined in below .
we can then obtain explicit bounds on the terms in , which enable us to recover theorem [ biasthm](ii ) . for @xmath173 ,
consider the density @xmath174 of the @xmath175 distribution .
then for any @xmath176 , we may take @xmath177\ ] ] to deduce from proposition [ prop : weakcond ] that for every @xmath101 , @xmath178 similar calculations show that the bias is of the same order for @xmath111 distributions with @xmath179 .
finally , we show that we can even obtain a small bias in settings where all derivatives of @xmath7 are zero at a point @xmath180 where @xmath158 .
let @xmath181 .
then for any @xmath176 , we may take @xmath182\ ] ] to deduce from proposition [ prop : weakcond ] once again that @xmath183 for every @xmath101 .
so far , we have assumed that the kozachenko
leonenko estimator @xmath31 is applied directly to @xmath184 ; when @xmath98 and our regularity conditions hold , theorem [ biasthm](i ) then yields that the dominant asymptotic contribution to the bias of @xmath31 is @xmath185 in this section , we consider the possibility of an alternative strategy , namely to make a transformation to the observations , apply the kozachenko
leonenko estimator to the transformed data , and then correct for the transformation in an appropriate way . recall that for any @xmath186 with @xmath187 , we have @xmath188 this motivates us to consider transformed estimators defined as follows : set @xmath189 , and let @xmath190 if @xmath191 is either orthogonal , or a positive scalar multiple of the identity , then @xmath192 and we have gained nothing , so we should look for transformations that are not of this type .
in fact , under the conditions of theorem [ biasthm](i ) , we see that the dominant asymptotic contribution to the bias of @xmath193 is @xmath194 in general , it is hard to compare the magnitudes of these two expressions ( [ eq : originalbias ] ) and ( [ eq : lineartrans ] ) .
however , in certain specific cases , an explicit comparison is possible .
for instance , suppose that we can write @xmath195 for some positive definite @xmath196 , where either @xmath197 for some univariate density @xmath198 , or can be written in the form @xmath199 for some @xmath200 and @xmath201 .
then , as shown in propositions [ prop : product ] and [ prop : spherical ] in appendix [ appendix : conds ] , we have @xmath202 from the expressions above , we see that the linear transformation strategy results in reduced asymptotic bias whenever @xmath203 . in fact , by the am gm inequality ,
@xmath204 is minimised when the eigenvalues of @xmath205 are equal .
one option for a practical data - driven approach is to ` prewhiten ' the data by setting @xmath206 , where @xmath207 is the sample covariance matrix and @xmath208 is the sample mean . in order to investigate the empirical performance of the prewhitened estimator @xmath193 with @xmath206 , we generated samples of size @xmath209 from the distributions @xmath210 and @xmath211 , where @xmath212 figures [ fig : prew1 ] and [ fig : prew2 ] present the mean squared errors of both the prewhitened and orignal kozachenko leonenko estimator based on @xmath213 repetitions , suggesting that the prewhitened estimator can yield significant improvements in performance for appropriate choices of @xmath0 . ] ]
we now study the asymptotic variance of kozachenko leonenko estimators under the assumption that the tuning parameter @xmath0 is diverging with @xmath1 ( which is necessary for efficiency ) .
[ varthm ] assume * ( a1)*(2 ) and that * ( a2)*(@xmath87 ) holds for some @xmath214 .
let @xmath215 and @xmath216 denote any two deterministic sequences of positive integers with @xmath217 , with @xmath218 and with @xmath219 , where @xmath220 then @xmath221 as @xmath92 , uniformly for @xmath222 .
the proof of this theorem is lengthy , and involves many delicate error bounds , so we outline the main ideas here .
first , we argue that @xmath223 where we hope to exploit the fact that @xmath224 .
the main difficulties in the argument are caused by the fact that handling the covariance above requires us to study the joint distribution of @xmath225 , and this is complicated by the fact that @xmath226 may be one of the @xmath0 nearest neighbours of @xmath227 or vice versa , and more generally , @xmath227 and @xmath226 may have some of their @xmath0 nearest neighbours in common .
dealing carefully with the different possible events requires us to consider separately the cases where @xmath228 is small and large , as well as the proximity of @xmath226 to @xmath227 .
finally , however , we can apply a normal approximation to the relevant multinomial distribution ( which requires that @xmath27 ) to deduce the result .
we remark that under stronger conditions on @xmath0 , it should also be possible to derive the same conclusion about the asymptotic variance of @xmath31 while only assuming similar conditions on the density to those required in proposition [ prop : weakcond ] , but we do not pursue this here . a straightforward consequence of theorem [ varthm ] is the asymptotic normality of kozachenko leonenko estimators .
[ clt ] assume that @xmath3 and the conditions of theorem [ varthm ] are satisfied .
if @xmath229 , we additionally assume that @xmath230
. then @xmath231 and @xmath232 as @xmath92 , uniformly for @xmath233 .
as was mentioned in the introduction , the asymptotic variance in theorem [ clt ] is best possible .
although the results of sections [ sec : bias ] and [ sec : var ] reveal that the estimators @xmath31 can not be efficient when @xmath234 , the insights provided motivate us to propose and analyse alternative estimators of @xmath18 that can be written as weighted averages of kozachenko leonenko entropy estimators . in this section we show that , for carefully chosen weights , they have reduced asymptotic bias while maintaining the same asymptotic variance as kozachenko leonenko estimators , and as such can achieve efficiency in higher dimensions . for @xmath235 such that @xmath236 , we define the weighted estimator @xmath237 where @xmath238 . then , under the conditions of theorem [ biasthm](i ) , we will show that for appropriate choices of @xmath239 and for every @xmath101 , @xmath240 as @xmath92 .
this facilitates a choice of @xmath0 and @xmath241 such that @xmath242 , and results in an estimator that has bias @xmath243 for @xmath244 and , under the stronger assumption that @xmath245 , for @xmath246 . in propostion [ weightedbiasprop ] in section [ sec : weightedproof ] we will see that , under stronger assumptions on the smoothness of @xmath7 and for @xmath0 diverging to infinity , the bias of @xmath31 can be expanded in powers of @xmath125 .
it then turns out that the weights @xmath239 we should consider are elements of the following set , where the restrictions on the support of @xmath239 are convenient for our analysis .
let @xmath247 and let @xmath248 for @xmath249 .
a sufficient condition for @xmath250 is that the @xmath251 matrix @xmath252 , with @xmath253 entry @xmath254 is invertible , as then @xmath255 gives the non - zero entries of a vector @xmath256 , where @xmath257 .
now define @xmath258 to have @xmath253 entry @xmath259 .
since @xmath260 as @xmath261 for @xmath262 , we have @xmath263 as @xmath27 .
now , @xmath191 is a vandermonde matrix ( depending only on @xmath21 ) and as such has determinant @xmath264 hence , by the continuity of the determinant and eigenvalues of a matrix , we have that there exists @xmath265 such that , for @xmath266 , the matrix @xmath252 is invertible and @xmath267 where @xmath268 denotes the eigenvalue of a matrix with smallest absolute value .
thus , for each @xmath266 , there exists @xmath256 satisfying @xmath269 .
the following theorem establishes the efficiency of @xmath270 for arbitrary @xmath21 .
the main ideas of the proof are provided in section [ sec : weightedproof ] , with details deferred to appendix [ appendix : weighted ] .
[ weightedclt ] assume * ( a1)*(@xmath75 ) , * ( a2)*(@xmath87 ) , and that the conditions on @xmath271 of theorem [ varthm ] hold .
assume further that @xmath272 for large @xmath1 and that @xmath273 .
* if @xmath274 , @xmath275 and @xmath276 , then @xmath277 uniformly for @xmath233 .
* if @xmath234 , @xmath278 $ ] and @xmath275 , then @xmath279 uniformly for @xmath233 .
theorem [ weightedclt ] generalises theorem [ clt ] , since we can simply that @xmath280 .
we remark that the level of smoothness we require for efficiency in theorem [ weightedclt](i ) , namely @xmath274 , is more than is needed for the two - stage estimator of @xcite in the case where @xmath7 is compactly supported and bounded away from zero on its support , where @xmath281 suffices . as alluded to in the introduction
, the fact that the function @xmath282 is non - differentiable at @xmath283 means that the entropy functional is no longer smooth when @xmath7 has full support , so the arguments of @xcite can no longer be applied and very different behaviour may occur @xcite .
theorem [ weightedclt](i ) of the previous subsection can be used to construct asymptotic confidence intervals for @xmath39 based on @xmath270 in the standard way , given a consistent estimator @xmath284 of @xmath285 ; cf .
also @xcite . since @xmath286 , it is sufficient to find a consistent estimator of @xmath287 , since we may simply use @xmath288 to estimate @xmath289 .
for this purpose , it is natural to consider the estimator @xmath290 an alternative , motivated by in the proof of theorem [ varthm ] , is to define @xmath291 ; as shown in proposition [ consistency ] below , these estimators have the same limiting behaviour . [ consistency ]
assume that the conditions of theorem [ varthm ] hold .
then @xmath292 uniformly for @xmath233 .
the same conclusion holds for @xmath293 .
the following corollary is immediate .
[ cor : conf ] assume that the conditions of theorem [ weightedclt](i ) hold and set @xmath294 .
for any @xmath295 , let @xmath296 denote the @xmath297th quantile of the standard normal distribution .
then @xmath298 as @xmath92 , uniformly for @xmath233 .
corollary [ cor : conf ] guarantees that @xmath299 provides an asymptotically valid confidence interval for @xmath39 , of asymptotically minimal width .
the proof of theorem [ biasthm ] relies on the following two auxiliary results , whose proofs are given in appendix [ appendix : auxiliary ] .
[ prop : moment ] let @xmath7 be a bounded density satisfying * ( a2)*(@xmath87 ) and let @xmath300 be a measurable function with @xmath301 as @xmath85 for every @xmath101 .
then @xmath302 for all @xmath303 $ ] .
recall the definition of @xmath126 in .
the first part of lemma [ lemma : hxinvbounds ] below provides crude but general bounds ; the second gives much sharper bounds in a more restricted region .
[ lemma : hxinvbounds ] a. assume that @xmath6 has a bounded density @xmath7 and that * ( a2)*(@xmath87 ) holds .
then for every @xmath112 and @xmath114 , @xmath304 b. assume * ( a1)*(@xmath75 ) , and let @xmath305 , @xmath166 be such that @xmath306 where @xmath80 is taken from * ( a1)*(@xmath75 ) .
then there exists @xmath63 such that for all @xmath1 sufficiently large , @xmath307 and @xmath308 , we have @xmath309 where @xmath310 and @xmath311 if @xmath75 is an even integer .
moreover , if @xmath103 , then @xmath312 we are now in a position to prove theorem [ biasthm ] .
\(i ) define @xmath313 where @xmath80 is as in assumption * ( a1)*(@xmath75 ) , let @xmath314 and let @xmath315 . recall that @xmath316 and let @xmath317 the proof is based on the fact that @xmath318 and lemma [ lemma : hxinvbounds](ii ) , which allow us to make the transformation @xmath319 . writing @xmath320 for remainder terms to be bounded at the end of the proof , we can write @xmath321 proposition [ prop : moment ] tells us that @xmath322 uniformly for @xmath96 as @xmath92 . noting that @xmath323
, it now remains to bound @xmath324 . henceforth , to save repetition ,
we adopt without further mention the convention that whenever an error term inside @xmath325 or @xmath326 depends on @xmath0 , this error is uniform for @xmath96 ; thus @xmath327 as @xmath92 means @xmath328 as @xmath92 . _
to bound @xmath329_. first note that for any @xmath330 , @xmath331 where the conclusion follows from hlder s inequality as in the proof of proposition [ prop : moment ] in appendix [ appendix : auxiliary ] .
now , from lemma [ lemma : hxinvbounds](i ) , @xmath332 moreover , @xmath333 when @xmath334 , so we deduce that for each @xmath101 , @xmath335 as @xmath92 . since @xmath336 when @xmath94 , this bound on @xmath329 is sufficiently tight . _ to bound @xmath337 . _ for random variables @xmath338 and @xmath339 we have that for every @xmath86 @xmath340 where the final equality follows from standard bounds on the left - hand tail of the binomial distribution ( see , e.g. @xcite , equation ( 6 ) , page 440 ) .
we therefore deduce from and the cauchy
schwarz inequality that for each @xmath145 , @xmath341 ( with further work , it can in fact be shown that @xmath342 , but this is not necessary for our overall conclusion . ) _ to bound @xmath343 .
_ we can write @xmath344 \mathrm{b}_{k , n - k}(s ) \ , ds \,dx \\ & = : r_{31}+r_{32},\end{aligned}\ ] ] say .
now , note that @xmath345 } \sup_{x \in \mathcal{x}_n } \frac{g_*^{d\beta}(x)s}{f(x ) } \rightarrow 0.\ ] ] it follows by lemma [ lemma : hxinvbounds](ii ) that there exists @xmath346 such that for @xmath347 , @xmath96 , @xmath348 and @xmath308 , taking @xmath349 , @xmath350 thus , for @xmath347 and @xmath96 , using the fact that @xmath351 for @xmath352 , @xmath353 \mathrm{b}_{k , n - k}(s ) \,ds \,dx \\ & \leq \frac { \gamma(k+4/d ) \gamma(n)}{2(d+2)^{2}v_d^{4/d } \gamma(k ) \gamma(n+4/d ) \delta_n^{2/d } } \int_{\mathcal{x}_n } g_*(x)^2 f(x)^{1 - 2/d } \ , dx \\ & \hspace{3 cm } + \frac{2c^{2 } \gamma(k+(4 + 2 \eta)/d ) \gamma(n ) } { \gamma(k ) \gamma(n+(4 + 2 \eta)/d ) \delta_n^{(2 + 2 \eta)/d } } \int_{\mathcal{x}_n } g_*(x)^2 f(x)^{1 - 2/d } \,dx.\end{aligned}\ ] ] on the other hand , we also have for @xmath347 and @xmath96 that @xmath354 it follows by proposition [ prop : moment ] and the fact that @xmath355 that @xmath356 _ to bound @xmath357_. consider the random variable @xmath338 .
then , using we conclude that for every @xmath101 , @xmath358 _ to bound @xmath359 .
_ it follows from proposition [ prop : moment ] and the fact that @xmath360 as @xmath92 that for each @xmath101 , @xmath361 _ to bound @xmath362_. by stirling s formula , @xmath363 as required .
\(ii ) the bias calculation in this setting is very similar , but we approximate @xmath364 simply by @xmath365 .
writing @xmath366 for the modified error terms , we obtain @xmath367 here , @xmath368 , and @xmath369 , for every @xmath101 in both cases . on the other hand , @xmath370 for every @xmath101 .
similarly , @xmath371 finally , @xmath372 for every @xmath101 . this concludes the proof . to deal with the integrals over @xmath373
, we first observe that by , for every @xmath101 , @xmath374 moreover , @xmath375 for every @xmath145 . now a simpler form of lemma [ lemma : hxinvbounds](ii ) states that there exists @xmath376 such that for @xmath377 and @xmath308 , we have @xmath378 } \bigl|v_df(x)h_x^{-1}(s)^d - s\bigr| \leq \frac{2dv_d^{-\tilde{\beta}/d}}{d+\tilde{\beta}}s^{1+\tilde{\beta}/d}\frac{c_{n,\tilde{\beta}}(x)}{f(x)^{1+\tilde{\beta}/d}}.\ ] ] it follows from , , and an almost identical argument to that leading to that for every @xmath101 , @xmath379 as required .
since this proof is long , we focus here on the main argument , and defer proofs of bounds on the many error terms to appendix [ appendix : varproof ] .
we use the same notation as in the proof of theorem [ biasthm ] , except that we change the definition of @xmath172 so that @xmath380 .
we write @xmath381 for this newly - defined @xmath172 .
similar to the proof of theorem [ biasthm ] , all error terms inside @xmath325 and @xmath326 that depend on @xmath0 are uniform for @xmath233 . first note that @xmath382 our first claim is that for every @xmath101 , @xmath383\ ] ] as @xmath92 .
the proof of this claim uses similar methods to those in the proof of theorem [ biasthm ] .
in particular , writing @xmath384 for remainder terms to be bounded later , we have @xmath385 \,dx + \sum_{i=1}^4 s_i \nonumber \\ & = \int_\mathcal{x } f(x ) \log^2 f(x ) \,dx + \sum_{i=1}^5 s_i + \frac{1}{k}\{1+o(1)\ } , \end{aligned}\ ] ] as @xmath92 . in appendix
[ appendix : s ] , we show that for every @xmath101 , @xmath386\ ] ] as @xmath92 .
the next step of our proof consists of showing that for every @xmath101 , @xmath387 as @xmath92 .
define @xmath388 so that @xmath389 writing @xmath390 , we therefore have that @xmath391 to deal with the first term in , we make the substitution @xmath392 and let @xmath393 . writing @xmath394 for remainder terms to be bounded later , for every @xmath101 , @xmath395 in appendix [ appendix : t ] , we show that for every @xmath101 , @xmath396 as @xmath92 .
we now deal with the second term in . by similar arguments to those used previously , for every @xmath101 , @xmath397 in appendix [ appendix : u ]
, we show that for every @xmath101 , @xmath398.we also require some further notation .
let @xmath399 denote the conditional distribution function of @xmath225 given @xmath400 .
let @xmath401 , and let @xmath402 so that @xmath403 and @xmath404 for every @xmath101 .
for pairs @xmath405 with @xmath406 and @xmath407 , let @xmath408 , and write @xmath409 write @xmath410 with @xmath411 for @xmath412 , let @xmath413 denote the distribution function of a @xmath414 random vector at @xmath415 , and let @xmath416 denote the standard univariate normal distribution function .
writing @xmath417 for remainder terms to be bounded later , we have @xmath418 the proof is completed by showing in appendix [ appendix : w ] that for every @xmath101 , @xmath419 as @xmath92 .
recall from the introduction that @xmath420 by the central limit theorem applied to @xmath421 , for the first part it suffices by slutsky s theorem to show that @xmath422 as @xmath92 . by theorem [ biasthm ] , under the conditions imposed , we know that @xmath423 as @xmath92 . moreover , from in the proof of theorem [ varthm ] , we have that @xmath424 as @xmath92 .
hence @xmath425 as @xmath92 .
now by similar , but simpler , methods to those employed in the proof of theorem [ biasthm ] , we have that for all @xmath145 , @xmath426 thus , @xmath427 as @xmath92 and the first result follows .
the second part is an immediate consequence of and , together with an application of the cauchy
schwarz inequality .
the first step towards proving theorem [ weightedclt ] is to gain a higher - order understanding of the bias of kozachenko leonenko estimators , which is provided by the following proposition . the proof is given in appendix [ appendix : weighted ] .
[ weightedbiasprop ] suppose that * ( a1)*(@xmath75 ) and * ( a2)*(@xmath87 ) hold for some @xmath428 .
let @xmath90 denote any deterministic sequence of positive integers with @xmath429 as @xmath92 for some @xmath430 .
then there exist @xmath431 , depending only on @xmath7 and @xmath21 , such that for each @xmath101 , @xmath432 as @xmath92 , uniformly for @xmath96 , where @xmath433 if @xmath434 , and @xmath435 if @xmath75 is an even integer
. note that in the setting of theorem [ biasthm](i ) we have @xmath436 proposition [ weightedbiasprop ] provides justification for our choice of @xmath437 .
the following result on the variance of @xmath270 is proved similarly to theorem [ varthm ] , and the proof is given in appendix [ appendix : weighted ] .
[ weightedvarprop ] assume that the conditions of theorem [ varthm ] hold . for @xmath1 sufficiently large , find @xmath438 such that @xmath439 .
then @xmath440 as @xmath92 , uniformly for @xmath233 .
we are now in a position to prove our main result on the theoretical properties of our weighted kozachenko leonenko estimators .
\(i ) by proposition [ weightedbiasprop ] and the fact that @xmath441
, we have @xmath442 under our conditions on @xmath443 . by proposition [ weightedvarprop ]
we have @xmath444 .
the results therefore follow by very similar arguments to those presented in theorem [ clt ] .
\(ii ) by proposition [ weightedbiasprop ] again , we have @xmath445 similarly , by proposition [ weightedvarprop ] , @xmath446 the result follows by an application of chebychev s inequality .
similar to , we may write @xmath447 thus , mimicking arguments in the proofs of theorems [ biasthm ] and [ varthm ] , for any @xmath448 , @xmath449 as @xmath92 . hence , by the cauchy
schwarz inequality , @xmath450 \rightarrow \mathbb{e}\{\log^m f(x_1)\}\end{aligned}\ ] ] as @xmath92 .
thus @xmath451 . similarly , from and cauchy schwarz , @xmath452 as @xmath92 .
we conclude that @xmath453 .
the fact that @xmath454 is also asymptotically unbiased can be deduced from ( and the subsequent bounds on the remainder terms ) in the proof of theorem [ varthm ] .
the first conclusion follows , and the second can be deduced from the first , together with the fact that @xmath455 as @xmath27 . * acknowledgements : * the second author is grateful to sebastian nowozin for introducing him to this problem , and to grard biau for helpful discussions
. 99 beirlant , j. , dudewicz , e. j. , gyrfi , l. , and van der meulen , e. c. ( 1997 ) nonparametric entropy estimation : an overview .
_ , * 6 * , 1739 .
biau , g. and devroye , l. ( 2015 ) _ lectures on the nearest neighbor method_. springer , new york .
cai , t. t. and low , m. g. ( 2011 ) testing composite hypotheses , hermite polynomials and optimal estimation of a nonsmooth functional .
_ , * 39 * , 10121041 .
cressie , n. ( 1976 ) on the logarithms of high - order spacings .
_ biometrika _ , * 63 * , 343355 .
delattre , s. and fournier , n. ( 2016 ) on the kozachenko
leonenko entropy estimator .
available at ` arxiv:1602.07440 ` .
gao , w. , oh , s. and viswanath , p. ( 2016 ) demystifying fixed @xmath0-nearest neighbor information estimators .
available at ` arxiv:1604.03006 ` .
goria , m. n. , leonenko , n. n. , mergel , v. v. and novi inverardi , p. l. ( 2005 ) a new class of random vector entropy estimators and its applications in testing statistical hypotheses . _ j. nonparametr .
_ , * 17 * , 277297 .
gtze , f. ( 1991 ) on the rate of convergence in the multivariate clt .
_ , * 19 * , 724739
. hall , p. and morton , s. c. ( 1993 ) on the estimation of entropy .
_ ann . inst .
_ , * 45 * , 6988 .
ibragimov , i. a. and khasminskii , r. z. ( 1991 ) asymptotically normal families of distributions and efficient estimation .
statist . _ ,
* 19 * , 16811724 .
kozachenko , l. f. and leonenko , n. n. ( 1987 ) sample estimate of the entropy of a random vector .
_ , * 23 * , 95101 .
kwak , n. and choi , c. ( 2002 ) input feature selection by mutual information based on parzen window .
_ ieee trans . pattern anal .
_ , * 24 * , 16671671 .
laurent , b. ( 1996 ) efficient estimation of integral functionals of a density .
_ , * 24 * , 659681 .
lepski , o. , nemirovski , a. and spokoiny , v. ( 1999 ) on estimation of the @xmath456 norm of a regression function .
fields _ , * 113 * , 221253 .
levit , b. ya .
( 1978 ) asymptotically efficient estimation of nonlinear functionals .
_ , * 14 * , 204209 .
miller , e. g. and fisher , j. w. ( 2003 ) ica using spacings estimates of entropy .
_ j. mach .
_ , * 4 * , 12711295 .
mnatsakanov , r. m. , misra , n. , li , s. and harner , e. j. ( 2008 ) @xmath457-nearest neighbor estimators of entropy .
methods statist . _ , * 17 * , 261277 .
paninski , l. ( 2003 ) estimation of entropy and mutual information .
_ neural comput .
_ , * 15 * , 11911253 .
shorack , g. r. and wellner , j. a. ( 2009 ) _ empirical processes with applications to statistics_. .
singh , h. , misra , n. , hnizdo , v. , fedorowicz , a. and demchuk , e. ( 2003 ) nearest neighbor estimates of entropy .
_ , * 23 * , 301321 .
singh , s. and pczos , b. ( 2016 ) analysis of @xmath0 nearest neighbor distances with application to entropy estimation .
available at ` arxiv:1603.08578 ` .
tsybakov , a. b. and van der meulen , e. c. ( 1996 ) root-@xmath1 consistent estimators of entropy for densities with unbounded support . _
_ , * 23 * , 7583 .
vasicek , o. ( 1976 ) a test for normality based on sample entropy .
_ j. r. stat .
_ , * 38 * , 5459 .
fix @xmath303 $ ] , and let @xmath458 .
we first claim that given any @xmath101 , there exists @xmath459 such that @xmath460 . to see this , observe that there exists @xmath461 $ ] such that @xmath462 for @xmath463 .
but then @xmath464 } \delta^\epsilon \sup_{x : f(x ) \geq \delta } g(x ) \leq \max\bigl\{1,c^\epsilon \sup_{x : f(x ) \geq \delta_0 } g(x)\bigr\ } \leq c^\epsilon \delta_0^{-\epsilon}.\ ] ] hence , defining @xmath465 , we have @xmath466 which establishes our first claim .
now choose @xmath467 and let @xmath468 .
then , by hlder s inequality , @xmath469 since @xmath470 .
\(i ) the lower bound is immediate from the fact that @xmath471 for any @xmath45 . for the upper bound ,
observe that by markov s inequality , for any @xmath45 , @xmath472 the result follows on substituting @xmath473 for @xmath112 .
\(ii ) we first prove this result in the case @xmath474 $ ] , giving the stated form of @xmath475 .
let @xmath476 , and let @xmath477 .
now , by * ( a1)*(@xmath75 ) and the mean value theorem , we have for @xmath478 that @xmath479 it is convenient to write @xmath480 then , provided @xmath481 $ ] , we have @xmath482 now , by our hypothesis , we know that @xmath483 as @xmath92 .
hence there exists @xmath346 such that for all @xmath347 , all @xmath307 and all @xmath308 , we have @xmath484 moreover , there exists @xmath485 such that for all @xmath486 , all @xmath307 and all @xmath308 we have @xmath487 finally , we can choose @xmath488 such that @xmath489 and such that @xmath490 for @xmath491 .
it follows that for @xmath492 , @xmath307 and @xmath308 , we have that @xmath481 $ ] and @xmath493 \geq s.\end{aligned}\ ] ] the lower bound is proved by very similar calculations , and the result for the case @xmath474 $ ] follows .
the general case can be proved using very similar arguments , and is omitted for brevity .
let @xmath7 be a density of the form @xmath195 for some positive definite @xmath196 and density @xmath132 , and let @xmath494 .
suppose further that @xmath495 for some invertible @xmath52 .
then @xmath496 thus , provided @xmath497 , we have @xmath498 this is proportional to @xmath499 as a function of @xmath500 if there exists @xmath501 such that @xmath502 where @xmath503 is the kronecker delta . in the next two propositions , we provide two simple , but reasonably general , conditions under which holds .
[ prop : product ] suppose that @xmath98 and that @xmath132 satisfies * ( a1)*(@xmath75 ) and * ( a2)*(@xmath87 ) for some @xmath504 and @xmath94 .
suppose further that @xmath505 for some density @xmath198 on @xmath506
. then holds . for @xmath507
, we have @xmath508 we may integrate by parts to write , for any @xmath509 , @xmath510 since @xmath132 satisfies * ( a1)*(@xmath75 ) for @xmath504 , it must be the case that @xmath198 has a bounded second derivative and hence @xmath198 must satisfy @xmath511 as @xmath512 .
hence , taking @xmath513 and @xmath514 in we see that @xmath515 when @xmath516 , we find that @xmath517 is independent of @xmath518 , as required . [
prop : spherical ] suppose that @xmath98 and that @xmath132 satisfies * ( a1)*(@xmath75 ) and * ( a2)*(@xmath87 ) for some @xmath504 and @xmath94 .
suppose further that @xmath519 for some function @xmath201 and some @xmath200 , where @xmath520
. then ( [ d2int ] ) holds . for @xmath521 ,
@xmath507 and @xmath522 , @xmath523 moreover , for @xmath524 , @xmath525 since this mixed second partial derivative is an odd function of both @xmath526 and @xmath527 when @xmath507 , it follows that @xmath515 finally , when @xmath516 , @xmath517 is independent of @xmath518 , so the claim follows .
_ to bound @xmath532 .
_ we have @xmath533 it therefore follows from lemma [ lemma : hxinvbounds](ii ) that for every @xmath101 , latexmath:[\ ] ] by similar means we can establish the same bound for the approximation of @xmath630 by @xmath685 . to conclude the proof , we write @xmath759 , where @xmath760 and deal with these two regions separately . by slepian s inequality , @xmath761 for all @xmath762 and @xmath138 , so by , for every @xmath145 , @xmath763 by lemma [ lemma : hxinvbounds](ii ) we have , uniformly in @xmath764 , @xmath765 with similar bounds holding for @xmath679 and @xmath766 .
we therefore have that for each @xmath101 , @xmath767 a corresponding lower bound of the same order for the left - hand side of follows from and the fact that @xmath768 uniformly in @xmath412 .
it now follows from , , and that for each @xmath145 @xmath769 as required .
the proof of the proposition is very similar to the proof of theorem [ biasthm ] , with the main difference being in bounding the error corresponding to @xmath343 . in this case
we just need to bound the error in a higher - order taylor expansion of @xmath770 this can be done using lemma [ lemma : hxinvbounds](ii ) ; we omit the details for brevity .
we first study the diagonal terms in the standard expansion of the variance of @xmath270 , and claim that @xmath771 in the proof of theorem [ varthm ] we studied the terms with @xmath772 and showed that , for @xmath518 with @xmath773 , @xmath774 as @xmath92 . for @xmath775 , using similar arguments to those used in the proof of theorem [ biasthm ] , we have that @xmath776 as @xmath92 .
now follows on noting that @xmath777 and each @xmath239 has at most @xmath21 non - zero entries .
we now study the off - diagonal terms in the expansion of the variance of @xmath270 , and claim that @xmath778 as @xmath92 . in light of
, it is sufficient to show that @xmath779 as @xmath92 .
we suppose here that @xmath780 and @xmath781 ; the @xmath772 case is dealt with in the proof of theorem [ varthm ] . we broadly follow the corresponding part of the proof of theorem [ varthm ] , though we require some new ( similar ) notation .
let @xmath782 denote the conditional distribution function of @xmath783 given @xmath400 and let @xmath784 denote the conditional distribution function of @xmath785 given @xmath786 .
let @xmath787 let @xmath788 , and let @xmath789 for pairs @xmath405 with @xmath790 and @xmath791 , let @xmath792 and write @xmath793 also write @xmath794 where @xmath795 .
writing @xmath796 for remainder terms to bounded later , we have @xmath797 as @xmath92 .
the final equality follows by the fact that , for borel measurable sets @xmath798 , @xmath799 _ to bound @xmath802 _ : similar to our work used to bound @xmath609 , we may show that @xmath803 as @xmath92 , for fixed @xmath804 . also , @xmath805 as @xmath92 . using these facts and very similar arguments to those used to bound @xmath609 we have @xmath806 _ to bound @xmath809 _ : let @xmath810 , where @xmath811 .
further , let @xmath812 then , as in , we have @xmath813 we now proceed to approximate @xmath814 by @xmath815 and @xmath816 by @xmath817 .
this is rather similar to the corresponding approximation in the bounds on @xmath673 , so we only present the main differences .
first , let @xmath818 we also define @xmath819 and set @xmath820 .
our aim now is to provide a bound on @xmath821 . by the smoothness of the function @xmath822 and * ( a1)*(2 )
we have that @xmath823 for @xmath824 . from this and similar bounds to , we find that @xmath825
. we therefore have @xmath826 which is as in the @xmath827 case except with the factor of @xmath828 missing .
note now that @xmath829 using , similar bounds to and the same arguments as leading up to , @xmath830 where @xmath831 and @xmath832 now let @xmath833 and @xmath834 . similarly to , we have @xmath835 similarly to the arguments leading up to , it follows that @xmath836 where the powers on the log factors are smaller because of the absence of the factor of @xmath837 in .
the remainder of the work required to bound @xmath809 is very similar to the work done from to , using also .
we then have @xmath838 as required . | many statistical procedures , including goodness - of - fit tests and methods for independent component analysis , rely critically on the estimation of the entropy of a distribution . in this paper ,
we seek entropy estimators that are efficient in the sense of achieving the local asymptotic minimax lower bound . to this end
, we initially study a generalisation of the estimator originally proposed by @xcite , based on the @xmath0-nearest neighbour distances of a sample of @xmath1 independent and identically distributed random vectors in @xmath2 .
when @xmath3 and provided @xmath4 ( as well as other regularity conditions ) , we show that the estimator is efficient ; on the other hand , when @xmath5 , a non - trivial bias precludes its efficiency regardless of the choice of @xmath0 .
this motivates us to consider a new entropy estimator , formed as a weighted average of kozachenko leonenko estimators for different values of @xmath0 .
a careful choice of weights enables us to obtain an efficient estimator in arbitrary dimensions , given sufficient smoothness .
in addition to the new estimator proposed and theoretical understanding provided , our results also have other methodological implications ; in particular , they motivate the prewhitening of the data before applying the estimator and facilitate the construction of asymptotically valid confidence intervals of asymptotically minimal width . |
the traffic of vesicles , organelles , protein complexes , messenger rna , and even viruses within the cells of living beings is driven by the molecular motors of the cytoskeleton which move along cytoskeletal filaments in a directed fashion @xcite .
there are three large classes of cytoskeletal motors , kinesins and dyneins which move along microtubules , and myosins which move along actin filaments .
these motors use the free energy released from the hydrolysis of adenosinetriphosphate ( atp ) , which represents their chemical fuel , for active movement and to perform mechanical work .
they move in discrete steps in such a way that one molecule of atp is used per step .
typical step sizes are @xmath0 nm , typical motor velocities are in the range of @xmath1m / sec . since the interior of cells is quite crowded and motors are strongly attracted by the filaments , which leads to relatively large motor densities along the filaments , it is interesting to study the collective traffic phenomena which arise from motor motor interactions , in particular the formation of traffic jams due to the mutual exclusion of motors from filament sites . to study these cooperative phenomena
theoretically we have introduced new variants of driven lattice gas models @xcite which have been studied extensively during the last years both by our group @xcite and by several other groups @xcite and which will be described below .
these models are related to lattice gas models for driven diffusive systems and exclusion processes as studied in the context of non - equilibrium phase transitions @xcite and highway traffic @xcite . since molecular motors can be studied in a systematic way using biomimetic systems which consist of a small number of components ( such as motors , filaments , and atp ) , they can also serve as model systems for the experimental investigation of driven diffusive systems .
although the traffic of cargo particles pulled by molecular motors within cells is remarkably similar to the macroscopic traffic on streets or rails , there is an important difference which is a direct consequence of the nanoscale size of molecular motors : the motor filament binding energy can be overcome by thermal fluctuations which are ubiquitous on this scale , and molecular motors therefore have a finite walking distance or run length after which they unbind from the filament along which they move .
this walking distance is typically of the order of 1 @xmath1 m for a single motor molecule . likewise ,
unbound motors which diffuse freely in the surrounding aqueous solution , can bind to a filament and start active movement .
in contrast to highway traffic , where additional cars enter only at on - ramps , i.e. at specific locations , binding of molecular motors occurs along the full length of the filaments .
in addition to stepping along a one - dimensional track and mutual exclusion , lattice models for the traffic of molecular motors must therefore also describe the dynamics of motor filament binding and unbinding as well as the diffusive movement of the unbound motors .
in contrast to the transport properties of single motor molecules which have been studied extensively during the last 15 years @xcite , the traffic phenomena in many - motor systems have only recently attracted the interest of experimentalists and are still largely unexplored from the experimental point of view .
the quantity of main interest has so far been the profile of the bound motor density along a filament .
density profiles with a traffic jam - like accumulation of motors at the end of filaments have been observed in vivo for a kinesin - like motor which was overexpressed in fungal hyphae @xcite . recently
, motor traffic jams have also been observed in biomimetic in vitro systems using both conventional kinesin ( kinesin 1 ) @xcite and the monomeric kinesin kif1a ( kinesin 3 ) @xcite . in the following
, we will give a short overview over the lattice models for molecular motors and discuss the motor traffic in various systems which differ mainly in the compartment geometry and the arrangement of filaments .
in section [ jams ] , we address the length of motor jams on filaments and argue that , in the presence of a large motor reservoir this jam length is typically of the order of the walking distance .
longer jams are found in confined geometries as discussed in section [ closed ] . in the last section of the paper , we briefly review our results for systems with two motor species .
to describe the interplay of the movements of bound and unbound motors , we have introduced a class of lattice models which incorporate the active movement of bound motors , the passive diffusion of unbound motors , and the motor filament binding and unbinding dynamics @xcite .
these models can also account for motor motor interactions such as their mutual exclusion from binding sites of the filament .
we describe the motor movements as random walks on a ( in general , three - dimensional ) cubic lattice as shown in fig .
[ fig : latticemodel](a ) . certain lines on this lattice represent the filaments .
the lattice constant is taken to be the motor step size @xmath2 which for many motors is equal to the filament periodicity .
when a motor is localized at a filament site , it performs a biased random walk . per unit time @xmath3
, it makes forward and backward steps with probabilities @xmath4 and @xmath5 , respectively . with probability @xmath6
, the motor makes no step and remains at the same site .
the latter parameter is needed to account for the fact that if the lattice constant is given by the motor step size , unbound diffusion over this scale is much faster than an active step of a bound motor .
finally , the motor hops to each of the adjacent non - filament sites with probability @xmath7 and unbinds from the filament .
the total unbinding probability per unit time is @xmath8 with the number @xmath9 of adjacent non - filament sites which is given by @xmath10 and @xmath11 for filaments in solution and filaments immobilized to a surface , respectively . at non - filament sites ,
the motor performs a symmetric random walk and hops to all neighboring sites with probability @xmath12 per time @xmath3 .
this choice of the hopping rate for unbound motor movements implies that the time scale @xmath3 is given by the diffusion coefficient @xmath13 of unbound motors via @xmath14 .
if it reaches a filament site , it binds to the filament with the sticking probability @xmath15 .
, a motor unbinds from the filament by stepping to an adjacent non - filament site .
unbound motors perform symmetric random walks and , when reaching a filament site , rebind to it with probability @xmath15 .
mutual exclusion implies that motors can not step to lattice sites that are already occupied by another motor .
( b ) in some situations , the unbound motor density @xmath16 ( or the corresponding concentration @xmath17 ) can be considered as constant . in that case , bound motors move along the filament as in ( a ) and unbind from it with probability @xmath8 .
binding of a motor to an empty filament site occurs with probability @xmath18 which can also be expressed as @xmath19 as discussed in section [ jams ] . ]
the behavior at the filament end has to be specified separately .
we consider two possibilities : ( i ) active unbinding of motors at the filament end where motors at the last filament site make a forward step with rate @xmath4 as at the other filament sites , but the latter step brings them to the forward non - filament neighbor site , so that their total unbinding probability is @xmath20 , and ( ii ) thermal unbinding where a motor at the last filament site does not make a forward step , but has an increased waiting probability @xmath21 . in that case , unbinding occurs with probability @xmath22 .
the hopping rates can be chosen in such a way that one incorporates the measured transport properties of single motors such as velocity , diffusion coefficient , and average walking distance before unbinding from the filament @xcite .
finally , motor - motor interactions can easily be incorporated into these models . in the following
we mainly consider the mutual exclusion of motors from lattice sites .
exclusion is most important at filament sites ( since the motors are strongly attracted to these sites ) , but in principle also applies to unbound motors . in the last section of this article
, we consider cooperative binding of motors to the filament . in that case , the binding and unbinding probabilities depend on the occupation of the nearest neighbor filament sites .
when considering many - motor systems , one is often interested in the motor densities and currents profiles rather than the single - motor trajectories .
the quantities of main interest are then the bound and unbound motor densities @xmath23 and @xmath16 . if gradients of the unbound motor density along the direction parallel to a filament can be neglected either because unbound diffusion is very fast or because the space available for unbound diffusion is large , so that motors remain unbound for a long time before rebinding to the filament the unbound density can be treated as constant . in that case , one obtains a one - dimensional model for the filament which is coupled to a reservoir of unbound motors with constant motor density as studied in refs . @xcite . per unit time @xmath3
, a motor on the filament unbinds with probability @xmath22 and binding of a motor from the reservoir to an empty lattice site occurs with probability @xmath24 , as shown in fig .
[ fig : latticemodel](b ) .
this situation will be discussed in section [ jams ] .
the motor traffic through tube - like compartments in which one or several filaments are aligned parallel to the cylinder axis represents a simple system which mimics the transport in axons .
we have studied tube - like systems with various kinds of boundary condition : closed systems @xcite , periodic boundary conditions @xcite , open boundaries coupled to motor reservoirs @xcite , and half - open systems @xcite .
the simplest case is given by periodic boundary conditions which can be solved exactly @xcite .
in this case , the stationary probability distribution is given by a product measure ; the bound and unbound motor densities are constant and satisfy the radial equilibrium condition @xmath25 where the last approximation usually holds under experimentally accessible conditions , where the unbound density is small , but the bound motor density can be of the order of one motor per binding site .
the bound motor current is given by @xmath26 with the bound motor velocity @xmath27 .
as a function of the bound motor density or of the total number @xmath28 of motors within the tube , it exhibits a maximum and decreases for high motor densities due to motor jamming as shown in fig . [
fig : j_phasendiag](a ) . for a filament in a tube with periodic boundary conditions as a function of the number @xmath29 of motor particles per length within the tube .
( b ) phase diagram for an open tube coupled to motor reservoirs with densities @xmath30 and @xmath31 at the left and right end of the tube . ]
if the tube is coupled to motor reservoirs at its orifices , the motor traffic exhibits boundary - induced phase transitions related to those of the one - dimensional asymmetric simple exclusion process ( asep ) @xcite . as for the asep , a low - density , high - density and maximal - current phase are present , and correspond to situations where the bottleneck which limits the transport is given by the left boundary , the right boundary or the interior of the tube , respectively . in all three phases ,
the motor densities are approximately constant and satisfy radial equilibrium sufficiently far from the boundaries .
the location of the transition lines within the phase diagram is quite sensitive to the precise choice of the boundary conditions and can be shifted by tuning the model parameters .
a particularly simple case is obtained if we impose radial equilibrium at the boundaries . in this case , the phase diagram , which is shown in fig .
[ fig : j_phasendiag](b ) , is independent of the motor transport properties and the geometric parameters of the tube , and the phase diagram corresponds exactly to the asep phase diagram .
from an experimental point of view , the simplest system , for which one can study molecular motor traffic , is given by one or several immobilized filaments which are in contact with a solution with a certain motor concentration .
for typical in vitro systems , unbound motor diffusion is very fast and the space available for unbound diffusion is large , so that we can describe the unbound motors by a constant density @xmath16 . in the following , we will use dimensionful quantities with units typically used by the experimentalist , and therefore characterize the unbound motors by the concentration @xmath17 , which is typically in the nano - molar range , rather than by the local volume fraction @xmath16 . in these units ,
the rate for the binding of an unbound motor to an empty filament site is given by @xmath19 where @xmath32 is the second - order binding rate .
it is related to the binding rate in density units via @xmath33 with @xmath34 and is most conveniently expressed in terms of the dissociation constant @xmath35 which has the dimension of concentration and is typically of the order of @xmath36 nm .
if the filament is long compared to the motor walking distance , the bound motor density is constant except for the regions close to the filament end and given by the equilibrium of the binding / unbinding dynamics , @xmath37 as shown in fig .
[ fig : fil_in_res](a ) and ( b ) for thermal and active unbinding at the filament end , respectively , using parameters for conventional kinesin .
likewise , the current along this part of the filament is given by @xmath38 . if the motors unbind thermally at the filament end , a ( rather short ) traffic jam forms at the filament end , where the motors accumulate .
note that no jam is obtained if the motors unbind actively as shown in fig .
[ fig : fil_in_res](b ) . the length of the jam region can be defined as @xmath39 with the filament length @xmath40 and @xmath41 .
an estimate of @xmath42 can be obtained from the balance of currents @xmath43,\ ] ] where @xmath44 is the forward current at the last filament site .
( [ eq : currbal ] ) leads to the jam length @xmath45^{-1},\ ] ] where @xmath46 is the average bound density in the jam region and @xmath47 is the walking distance of the motors as given by @xmath48 . for thermal unbinding of motors at the filament end ,
@xmath49 . if we estimate the density within the jam by the maximal value , @xmath50 , eq .
( [ eq : l * ] ) leads to @xmath51 for the jam length in agreement with simulations which show that @xmath52 with a prefactor close to one .
this estimate shows that , for filaments in contact with a solution with constant motor concentration , the jam length @xmath42 is of the order of the walking distance @xmath47 .
longer jam lengths can arise ( i ) if the unbinding rate @xmath53 decreases with increasing density or ( ii ) if a gradient in the concentration of unbound motors is build up @xcite which increases binding to the filament in the jam region and thus also increases the last term of eq .
( [ eq : l * ] ) . on a filament in contact with a solution with constant unbound motor concentration @xmath17 as a function of the coordinate @xmath54 parallel to the filament for ( a ) thermal and ( b ) active unbinding at the filament end .
note that a traffic jam only occurs for thermal unbinding at the filament end and that this jam is rather short , of the order of the walking distance .
the parameters are as appropriate for kinesin , @xmath55s , @xmath56 nm , @xmath57 m/s , and for a microtubule of length @xmath58 m . ]
if filaments are embedded into closed compartments , the motor current along these filaments leads to the build - up of density gradients within these compartments @xcite .
these gradients are particularly pronounced in tube - like compartments where all filaments are aligned in parallel and with the same orientation along the tube axis . for low motor densities , the motors are essentially localized at that end of the tube towards which their active movements are directed .
if exclusion can be neglected , the bound and unbound motor densities decrease exponentially if one moves away from this tube end .
the length scale @xmath59 of the exponential decrease is given by @xmath60 , the ratio of the walking distance of bound motors and the distance unbound motor diffuse before rebinding .
in this expression , @xmath27 and @xmath13 are the bound velocity and the unbound diffusion coefficient of the motors , respectively , and @xmath61 is the cross - section of the tube .
a constant unbound motor density is a good approximation if @xmath62 , i.e. , for large unbound diffusion coefficients @xmath13 and for large tube radii .
is balanced by a diffusive current of unbound motors which diffuse back with the diffusion coefficient @xmath13 .
( b ) corresponding profiles of the bound motor density @xmath23 as a function of the coordinate @xmath54 along the filament .
a traffic jam domain at the right end of the tube builds up both for thermal and active unbinding of motors from the last filament site ( solid and dashed lines , respectively ) . with increasing overall motor concentration
, the crowded domain spreads to the left . ]
if the overall motor density is increased in these systems , the region in which the motors are localized develops into an extended crowded domain , see fig .
[ fig : endab ] .
the length @xmath42 of this domain defines the jam length for these systems .
in contrast to the systems discussed in the previous section , the jam length can be larger than the walking distance and increases with increasing overall motor concentration until the crowded domain spreads over the full tube length . in this crowded domain ,
the density profiles can approximately be described by local radial equilibrium .
for the case of a half - open tube , which is very similar to the closed tube , but more easily accessible to analytical methods , the latter approximation shows that the jam length scales essentially as @xmath63 @xcite rather than @xmath64 as for a filament in contact with a constant unbound motor density .
the jam length is given by @xmath65 where @xmath66 is a function of the ratio of the binding and unbinding probabilities and of the bound motor density @xmath30 in the reservoir to which the tube is coupled at its open end @xcite .
if the boundary density is sufficiently close to one , @xmath66 behaves as @xmath67 and the jam length diverges logarithmically with @xmath68 . for the closed tube , @xmath66 is determined by an integral constraint which fixes the total number of motors within the tube .
in addition , the traffic jam is present for both thermal and active unbinding at the filament end @xcite as shown in fig .
[ fig : endab](b ) .
this means that the crowded domains are due to a combination of the motor behavior at the last filament site and the motor accumulation in the region close to the filament end .
the latter accumulation is strongly geometry - dependent .
we have also studied centered or aster - like filament systems @xcite . in this case
, the accumulation of motors in the center of an aster is much weaker than in tube - like systems and , in fact , determined by a power law rather than by an exponential . as for filaments in contact with a reservoir with constant unbound motor density ,
traffic jams are obtained only for thermal unbinding at the filament end , but not for active unbinding .
in addition , when the overall motor density is increased , the traffic jams remain short in this case .
the main effect of an increase in the overall motor density is a flattening of the density profile .
each molecular motor moves either towards the plus- or towards the minus - end of the corresponding filament , but different types of motors move into opposite directions along the same filament .
in this situation , cooperative binding of the motors to the filament in such a fashion that a motor is more likely to bind and less likely to unbind next to a bound motor moving in the same direction , while it is less likely to bind and more likely to unbind next to a motor with opposite directionality leads to spontaneous symmetry breaking @xcite : if the motor motor interactions , which we characterize by a single interaction parameter @xmath69 , are stronger than a certain critical interaction strength @xmath70 , one motor species occupies the filament , while the other one is largely excluded from it .
this symmetry breaking has been found both for tube - like compartments with periodic boundary conditions and for systems with a constant unbound motor density . in the latter case ,
symmetry breaking occurs , independent of the choice of the boundary conditions provided that the system size or the filament length is large compared to the motors walking distance .
note that , in contrast to the previously reported example for symmetry breaking in a driven diffusive system , the bridge model @xcite , the symmetry breaking here is not boundary - induced .
symmetry breaking has two interesting consequences .
first , it implies that for @xmath71 there is a discontinuous phase transition if the relative concentrations of the two motor species are varied .
this transition is accompanied by hysteresis , which is again not boundary - induced , in contrast to the hysteresis which was reported recently for another driven diffusive system @xcite .
second , if several filaments are aligned in parallel and with the same orientation , this symmetry breaking leads to the spontaneous formation of traffic lanes for motor traffic with opposite directionality @xcite .
the traffic phenomena in systems with many molecular motors can be described by stochastic lattice gas models which are similar to asymmetric exclusion processes , but have the additional property that the motors can unbind from the filamentous track and diffuse in the surrounding fluid .
these systems exhibit a variety of cooperative phenomena and , in addition to their importance for our understanding of the traffic within cells and for prospective applications in nanotechnology , provide promising model systems for the experimental study of driven diffusive systems . | molecular motors perform active movements along cytoskeletal filaments and drive the traffic of organelles and other cargo particles in cells .
in contrast to the macroscopic traffic of cars , however , the traffic of molecular motors is characterized by a finite walking distance ( or run length ) after which a motor unbinds from the filament along which it moves .
unbound motors perform brownian motion in the surrounding aqueous solution until they rebind to a filament .
we use variants of driven lattice gas models to describe the interplay of their active movements , the unbound diffusion , and the binding / unbinding dynamics .
if the motor concentration is large , motor - motor interactions become important and lead to a variety of cooperative traffic phenomena such as traffic jams on the filaments , boundary - induced phase transitions , and spontaneous symmetry breaking in systems with two species of motors . if the filament is surrounded by a large reservoir of motors , the jam length , i.e.
, the extension of the traffic jams is of the order of the walking distance .
much longer jams can be found in confined geometries such as tube - like compartments . |
the calculus on time scales is a recent field , introduced by bernd aulbach and stefan hilger in 1988 @xcite , that unifies and extends difference and differential equations into a single theory @xcite .
a time scale is a model of time , and the new theory has found important applications in several fields that require simultaneous modeling of discrete and continuous data , in particular in the calculus of variations @xcite , control theory @xcite , and optimal control @xcite .
other areas of application include engineering , biology , economics , finance , and physics @xcite . the present work is dedicated to the study of problems of calculus of variations on an arbitrary time scale @xmath0 . as particular cases , one gets the standard calculus of variations @xcite by choosing @xmath1 ; the discrete - time calculus of variations @xcite by choosing @xmath2 ; and the @xmath3-calculus of variations @xcite by choosing @xmath4 , @xmath5 . in section [ sec : prel : ts ] we briefly present the necessary notions and results of time scales , delta derivatives , and delta integrals .
let @xmath0 be a given time scale with at least three points , @xmath6 , and @xmath7 be of class @xmath8 .
suppose that @xmath9 and @xmath10 .
we consider the following optimization problem on @xmath0 : @xmath11=\int_a^b l(t , q^\sigma(t),q^\delta(t ) ) \delta t \longrightarrow \min_{q\in \mathcal{d}},\ ] ] where @xmath12\cap \mathbb{t } \rightarrow \mathbb{r}^n,\ q\in \mathrm{c}^1_{rd},\ q(a)=q_a,\ q(b)=q_b\}\ ] ] for some @xmath13 , and where @xmath14 is the forward jump operator and @xmath15 is the delta - derivative of @xmath3 with respect to @xmath0 .
for @xmath1 we get the classical fundamental problem of the calculus of variations , which concerns the minimization of an integral @xmath16=\int_a^b l(t , q(t),\dot{q}(t ) ) dt\ ] ] over all trajectories @xmath17 satisfying given boundary conditions @xmath18 and @xmath19 .
several classical results on the calculus of variations are now available to the more general context of time scales : ( first ) euler - lagrange equations @xcite ; necessary optimality conditions for isoperimetric problems @xcite and for problems with higher - order derivatives @xcite ; the weierstrass necessary condition @xcite ; and noether s symmetry theorem @xcite . in this paper
we prove a new result for the problem of the calculus of variations on time scales : we obtain in section [ sec : mr ] a time scale version of the classical _ second euler - lagrange equation _
@xcite , also known in the literature as the dubois - reymond necessary optimality condition @xcite .
the classical second euler - lagrange equation asserts that if @xmath3 is a minimizer of , then @xmath20 \\ = - \partial_1 l ( t , q(t ) , \dot{q}(t))\ , , \end{gathered}\ ] ] where @xmath21 , @xmath22 , denotes the partial derivative of @xmath23 with respect to its @xmath24-th argument . in the autonomous case , when the lagrangian @xmath25 does not depend on the time variable @xmath26 , the second euler - lagrange condition is nothing more than the second erdmann necessary optimality condition : @xmath27 along all the extremals of the problem , which in mechanics corresponds to the most famous conservation law conservation of energy . for a survey of the classical optimality conditions we refer the reader to (
* ch . 2 ) .
here we just recall that is one of the cornerstone results of the calculus of variations and optimal control @xcite : it has been used , for example , to prove existence , regularity of minimizers , conservation laws , and to explain the lavrentiev phenomena . main result of the paper gives an extension of to an arbitrary time scale ( theorem [ secondel ] ) : if @xmath3 is a solution of problem , then @xmath28 = - \partial_1 l(t , q^\sigma(t),q^\delta(t ) ) \ , .\end{gathered}\ ] ] as an application , we show in section [ sec : appl : nt ] how one can use the new second euler - lagrange equation to prove the noether s principle on time scales : the invariance of functional with respect to a one - parameter family of transformations implies the existence of a conserved quantity along the time scale euler - lagrange extremals ( theorem [ noether ] ) . when problem is autonomous one has invariance with respect to time translations and the corresponding noether s conservation law gives an extension of the second erdmann equation to time scales : @xmath29 in classical mechanics gives conservation of energy .
the conservation law tells us that an analogous result remains valid on an arbitrary time scale .
however , the role of the classical hamiltonian @xmath30 in is substituted by a time - scale hamiltonian @xmath31 in , , @xmath32 , with the new term @xmath33 on the left - hand side of depending on the graininess of the time scale .
for a general introduction to the calculus on time scales we refer the reader to the book @xcite . here
we only give those notions and results needed in the sequel . as usual , @xmath34 , @xmath35 , and @xmath36 denote , respectively , the set of real , integer , and natural numbers .
a _ time scale _ @xmath0 is an arbitrary nonempty closed subset of @xmath37 . besides standard cases of @xmath34 ( continuous time ) and @xmath35 ( discrete time ) ,
many different models of time are used . for each time scale
@xmath0 the following operators are used : * the _ forward jump operator _ @xmath38 , @xmath39 for @xmath40 and @xmath41 if @xmath42 ; * the _ backward jump operator _ @xmath43 , @xmath44 for @xmath45 and @xmath46 if @xmath47 ; * the _ forward graininess function _
@xmath48 , @xmath49 . if @xmath50 , then for any @xmath51 , @xmath52 and @xmath53 . if @xmath54 , then for every @xmath55 , @xmath56 , @xmath57 and @xmath58 .
a point @xmath59 is called _ right - dense _ , _ right - scattered _ , _ left - dense _ or _ left - scattered _ if @xmath60 , @xmath61 , @xmath62 , or @xmath63 , respectively .
we say that @xmath26 is _ isolated _ if @xmath64 , that @xmath26 is _ dense _ if @xmath65 . if @xmath66 is finite and left - scattered , we define @xmath67 otherwise , @xmath68 .
let @xmath69 and @xmath70 .
the _ delta derivative _ of @xmath71 at @xmath26 is the real number @xmath72 with the property that given any @xmath73 there is a neighborhood @xmath74 of @xmath26 such that @xmath75 for all @xmath76 .
we say that @xmath71 is _ delta differentiable _ on @xmath77 provided @xmath72 exists for all @xmath70 .
we shall often denote @xmath78 by @xmath79 if @xmath71 is a composition of other functions .
the delta derivative of a function @xmath80 ( @xmath6 ) is a vector whose components are delta derivatives of the components of @xmath71 . for @xmath81 , where @xmath82 is an arbitrary set , we define @xmath83 . for delta
differentiable @xmath71 and @xmath84 , the next formulas hold : @xmath85 if @xmath86 , then @xmath87 is delta differentiable at @xmath88 if and only if @xmath71 is differentiable in the ordinary sense at @xmath26 .
then , @xmath89 . if @xmath54 , then @xmath90 is always delta differentiable at every @xmath55 with @xmath91 .
let @xmath92 , @xmath10 .
we define the interval @xmath93_{\mathbb{t}}$ ] in @xmath0 by @xmath94_{\mathbb{t}}:=\ { t \in \mathbb{t } : a\leq t\leq b\}.\ ] ] open intervals and half - open intervals in @xmath0 are defined accordingly .
[ monotonia delta ] let @xmath95_{\mathbb{t}}\rightarrow\mathbb{r}$ ] be a continuous function that has a delta derivative at each point of @xmath93_{\mathbb{t}}^\kappa$ ]
. then @xmath71 is increasing , decreasing , non - decreasing , and non - increasing on @xmath93_{\mathbb{t}}$ ] if @xmath96 , @xmath97 , @xmath98 and @xmath99 for all @xmath100_{\mathbb{t}}^\kappa$ ] , respectively .
a function @xmath101 is called a _
delta antiderivative _ of @xmath102 provided @xmath103 in this case we define the _ delta integral _ of @xmath71 from @xmath104 to @xmath105 ( @xmath92 ) by @xmath106 if @xmath50 , then @xmath107 where the integral on the right hand side is the usual riemann integral .
if @xmath108 , where @xmath109 , then @xmath110 for @xmath10 .
in order to present a class of functions that possess a delta antiderivative , the following definition is introduced : a function @xmath111 is called _ rd - continuous _ if it is continuous at the right - dense points in @xmath0 and its left - sided limits exist ( finite ) at all left - dense points in @xmath0 .
a function @xmath112 is _ rd - continuous _ if all its components are rd - continuous .
we remark that a rd - continuous function defined on a compact interval , with real values , is bounded .
the set of all rd - continuous functions @xmath112 is denoted by @xmath113 , or simply by @xmath114 .
similarly , @xmath115 and @xmath116 will denote the set of functions from @xmath114 whose delta derivative belongs to @xmath114 .
[ antiderivada ] every rd - continuous function has a delta antiderivative .
in particular , if @xmath117 , then the function @xmath118 defined by @xmath119 is a delta antiderivative of @xmath71
. the following results will be very useful in the proof of our main result ( theorem [ secondel ] ) .
[ theorems ] assume that @xmath120 is strictly increasing and @xmath121 is a time scale . 1 . ( chain rule )
let @xmath122 . if @xmath123 and @xmath124 exist for all @xmath125 , then @xmath126 2 .
( derivative of the inverse ) the relation @xmath127 holds at points @xmath128 where @xmath129 3 .
( substitution in the integral ) if @xmath130 is a @xmath114 function and @xmath131 is a @xmath116 function , then for @xmath132 , @xmath133 we say that @xmath134_{\mathbb{t } } , \mathbb{r}^n)$ ] is a local minimizer for problem ( [ problem ] ) if there exists @xmath135 such that @xmath136\leq i[y]\ ] ] for all @xmath137_{\mathbb{t } } , \mathbb{r}^n)$ ] satisfying the boundary conditions @xmath18 , @xmath19 , and @xmath138_{\mathbb{t}}^\kappa}\mid y^{\sigma}(t)-y_{\ast}^{\sigma}(t)\mid + \sup_{t \in [ a , b]_{\mathbb{t}}^\kappa}\mid y^{\delta}(t)-y_{\ast}^{\delta}(t)\mid < \delta \ , , \ ] ] where @xmath139 denotes a norm in @xmath140 .
we recall now the ( first ) euler - lagrange equation as presented in @xcite . as in the introduction
, we use @xmath21 to denote the partial derivative of @xmath25 with respect to the @xmath24-th variable ( or group of variables ) .
[ thm : elts ] if @xmath3 is a local minimizer of , then @xmath3 satisfies the following euler - lagrange equation : @xmath141_{\mathbb{t}}^{\kappa}.\ ] ]
the following theorem presents a generalization to time scales of the second euler - lagrange equation @xcite ( also known as the dubois - reymond equation @xcite ) . _
( the second euler - lagrange equation on time scales ) _ : [ secondel ] if @xmath142 is a local minimizer of problem , then @xmath3 satisfies the equation @xmath143 for all @xmath144_{\mathbb{t}}^\kappa$ ] , where @xmath145 @xmath146 and @xmath147 .
let @xmath148 be a local minimizer of functional @xmath149 in .
we will prove that there exists @xmath150 , @xmath151 , that satisfies the condition @xmath152_{\mathbb{t}}^\kappa.\ ] ] if @xmath153 , then any @xmath150 satisfies condition .
suppose now that @xmath154 .
then there exists some @xmath155 such that @xmath156 where we suppose that @xmath157 . since @xmath158 is bounded on @xmath93_{\mathbb{t}}^\kappa$ ] , then there exist @xmath159 such that @xmath160_{\mathbb{t}}^\kappa.\ ] ] let @xmath161 where @xmath162 if @xmath163 . if @xmath164 we can choose @xmath165 such that @xmath166 . if @xmath167 we can choose @xmath165 such that @xmath168 .
the map @xmath169_{\mathbb{t}}\rightarrow \mathbb{r}$ ] defined by @xmath170 is delta differentiable with @xmath171 and , by theorem [ monotonia delta ] , @xmath172 is strictly increasing on @xmath93_{\mathbb{t}}$ ] .
note that @xmath173_{\mathbb{t}})$ ] is a new time scale ( because @xmath172 is continuous and @xmath93_{\mathbb{t}}$ ] is closed ) . by @xmath174
we denote the forward jump operator and by @xmath175 we denote the delta derivative on @xmath176 .
let @xmath177 and define @xmath178 for @xmath179 .
note that @xmath180 and @xmath181 by the chain rule and from , @xmath182 which gives @xmath183 by the derivative of the inverse applied to @xmath172 we can conclude that @xmath184 note that , since @xmath185 , then @xmath186 and @xmath187 from ( [ eq5 ] ) , ( [ eq3 ] ) , and the substitution in the integral , @xmath188 & = \int_a^b l(t , q_0^\sigma(t),q_0^\delta(t ) ) \delta t\\ & = \int_\alpha^\beta \tilde{l}(\tau,\eta_0^{\tilde{\sigma}}(\tau ) , \eta_0^{\tilde{\delta}}(\tau ) ) \tilde{\delta } \tau = : \tilde{i}[\eta_0 ] , \end{split}\ ] ] where @xmath189 for @xmath190 , @xmath191 , @xmath192 , @xmath193 and @xmath194 .
let @xmath195 we remark that @xmath196 was chosen so small that the constraint @xmath197 is always satisfied for any function @xmath198 in the `` nearby '' of @xmath199 .
since @xmath200 is by assumption a local minimizer of @xmath149 in @xmath201 , it follows that @xmath199 is a local minimizer of @xmath202 in @xmath203 , so it satisfies the euler - lagrange equation ( in integral form ) @xmath204 where @xmath205 is a constant vector .
differentiating we obtain @xmath206 and @xmath207 using , , and ( [ eq3 ] ) we obtain @xmath208 note that @xmath209 hence , by the euler - lagrange equation ( [ eq9 ] ) , we may conclude that @xmath210 the last equality may be rewritten as @xmath211\\ = & - \left[\partial_3 l\left(t , q_0^\sigma(t),q_0^\delta(t)\right ) - \int_a^t \partial_2 l\left(s , q_0^\sigma(s),q_0^\delta(s)\right ) \delta s - c_1\right ] .
\end{split}\ ] ] using the euler - lagrange equation for @xmath200 we arrive at the intended statement .
if @xmath50 , then the equation ( [ 2equationel ] ) simplifies due to the fact that @xmath212 , and we obtain the classical second euler - lagrange equation ( , , @xcite ) : if @xmath3 is a local minimizer of the classical functional of the calculus of variations , then @xmath213\\ = -\partial_1 l ( t , q(t ) , \dot{q}(t))\end{gathered}\ ] ] holds for all @xmath100 $ ] . in the autonomous case , theorem [ secondel ]
gives an extension of the classical second erdmann condition : if @xmath142 is a local minimizer of the problem @xmath214=\int_a^b l(q^\sigma(t),q^\delta(t ) ) \delta t \longrightarrow \min_{q\in \mathcal{d}},\ ] ] then @xmath3 satisfies equation for all @xmath144_{\mathbb{t}}^\kappa$ ] .
[ example1 ] let @xmath77 be a time scale with @xmath215 , @xmath216 .
consider problem ( [ problem ] ) with @xmath217 and a lagrangian @xmath25 given by @xmath218 .
the second euler - lagrange equation for this problem is @xmath219 and the extremal is @xmath220 with @xmath221 let @xmath222 and consider the following problem on @xmath0 : @xmath223 = \int_0 ^ 1 \left[(q^\delta(t))^2 - 1\right]^2 \delta t \longrightarrow \min \ , , \\
q(0 ) = 0 \ , , \quad q(1 ) = 0 \ , , \\ q
\in c_{rd}^1(\mathbb{t } ; \mathbb{r } ) \ , .\end{gathered}\ ] ] the euler - lagrange equation takes the form @xmath224 = \text{const}\ ] ] while the second euler - lagrange equation asserts that @xmath225 \left[1 + 3 ( q^\delta(t))^2\right ] = \text{const } \ , .\ ] ] let @xmath226 for all @xmath227 , and @xmath228 .
one has @xmath229 , @xmath230 , and @xmath231 , @xmath232 .
we see that @xmath233 is an extremal , , it satisfies the euler - lagrange equation .
however @xmath233 can not be a solution to the problem since it does not satisfy the second euler - lagrange equation .
in fact , any function @xmath3 satisfying @xmath234 , @xmath125 , is an euler - lagrange extremal . among them , only @xmath235 for all @xmath125 and those with @xmath236 satisfy our condition .
this example shows a problem for which the euler - lagrange equation gives several candidates which are not the solution to the problem , while our second euler - lagrange equation gives a smaller set of candidates .
moreover , the candidates obtained from our condition lead us directly to the explicit solution of the problem .
indeed , the null function and any function @xmath3 with @xmath237 and @xmath236 , @xmath125 , gives @xmath238 = 0 $ ] .
they are minimizers because @xmath238 \ge 0 $ ] for any function @xmath239 .
let @xmath240_{\mathbb{t}}\rightarrow \mathbb{r}^n$ ] , @xmath241 , and consider a one - parameter family of infinitesimal transformations @xmath242 where @xmath243 , @xmath244_{\mathbb{t}}\times \mathbb{r}^n\rightarrow\mathbb{r}$ ] , and @xmath245_{\mathbb{t}}\times \mathbb{r}^n\rightarrow\mathbb{r}$ ] are delta differentiable functions .
we assume that for every @xmath246 and every @xmath247 , the map @xmath93\ni t \mapsto \alpha(t):= t(t , q(t ) , \epsilon)\in\mathbb{r}$ ] is a strictly increasing @xmath116 function and its image is again a time scale with the forward shift operator @xmath248 and the delta derivative @xmath249 .
we recall that the following holds : @xmath250 [ def .
invariance ] functional @xmath149 in is said to be invariant on @xmath74 under the family of transformations if @xmath251 where , for simplicity of notation , we omit the arguments of functions @xmath252 and @xmath253 : @xmath254 , @xmath255 .
note that the invariance notion presented in ( * ? ? ?
* definition 5 ) implies definition [ def .
invariance ] . indeed , for any subinterval @xmath256_{\mathbb{t}}\subseteq [ a , b]_{\mathbb{t}}$ ] , any @xmath257 , and any @xmath247 , one has @xmath258 from the arbitrariness of @xmath259 and @xmath260 it follows that @xmath261 and this implies @xmath262 [ thm : invariance ] functional @xmath149 in is invariant on @xmath74 under the family of transformations if and only if @xmath263 for all @xmath100_{\mathbb{t}}^\kappa$ ] and all @xmath257 , where @xmath264 since @xmath265 the definition of invariance is equivalent to @xmath266 which proves the desired result .
[ example2 ] for example [ example1 ] one has invariance under the family of transformations with @xmath267 and @xmath268 , where @xmath269 and @xmath270 are arbitrary constants . in order to simplify expressions , we write @xmath271 instead of @xmath272 . similarly for the partial derivatives of @xmath25 .
we recall that @xmath3 is an _
extremal _ to problem if it satisfies the euler - lagrange equation .
[ noether ] if functional @xmath149 in is invariant on @xmath74 in the sense of definition [ def .
invariance ] ( lemma [ thm : invariance ] ) , then @xmath273\cdot\tau(t , q)\end{gathered}\ ] ] is constant along all the extremals of problem ( [ problem ] ) .
we must prove that @xmath274\end{gathered}\ ] ] is equal to zero along all the extremals of problem ( [ problem ] ) .
we begin noting that @xmath275 \cdot\tau^{\sigma}(t , q)\\ & \quad + \bigl[l(t , q^{\sigma},q^{\delta})-\partial_{3}l(t , q^{\sigma},q^{\delta})\cdot q^{\delta}\\ & \qquad\quad -\partial_{1}l(t , q^{\sigma},q^{\delta})\cdot\mu(t)\bigr ] \cdot\tau^{\delta}(t , q ) \ , .
\end{split}\ ] ] using the first and second euler - lagrange equations and , respectively , we conclude that @xmath276 since @xmath277 , then @xmath278 hence , @xmath279 using lemma [ thm : invariance ] we arrive at the intended conclusion . if @xmath50 , then @xmath212 and theorem [ noether ] reduces to the classical noether s theorem ( , , @xcite ) : if the classical fundamental functional of the calculus of variations is invariant , then @xmath280\cdot\tau(t , q)\ ] ] is constant along all the extremals of the problem .
[ example3 ] for the problem of example [ example2 ] one has from theorem [ noether ] that @xmath281 along the extremals @xmath3 of the problem .
this is indeed true : from example [ example1 ] we know that the extremals have the form @xmath220 for some constants @xmath282 ; thus , the conservation law takes the form @xmath283 .
in this paper we obtain a second euler - lagrange equation and a second erdmann condition for the problem of the calculus of variations on time scales .
since both necessary optimality conditions are important and extremely useful results in the calculus of variations and optimal control when @xmath284 , we claim that the present results are also useful for the development of the recent theory of the calculus of variations on time scales @xcite .
as pointed out to us by richard vinter , our second euler - lagrange equation in the time scales setting seems to be useful in a framework for studying the asymptotics of time discretization . as an example of application of our main results , we give a simpler and more elegant proof to the noether symmetry theorem on time scales obtained in 2008 @xcite , which allows to obtain conserved quantities along the extremals of the problems .
standard noetherian constants of motion are violated due to the presence of a new term that depends on the graininess @xmath285 of the time scale , while in the classical context @xmath53 .
the importance of noether s conservation laws in the calculus of variations , optimal control theory , and its applications in engineering , are well recognized @xcite .
their role on the general context of optimal control on time scales is an entirely open area of research . in particular
, it would be interesting to investigate the techniques of @xcite with the recent higher - order euler - lagrange equations on time scales @xcite for a possible extension of theorem [ noether ] to variational problems on time scales with higher - order delta or nabla derivatives .
the question of obtaining conserved quantities along the extremals of higher - order problems of the calculus of variations on time scales remains an interesting open question .
zbigniew bartosiewicz was supported by biaystok university of technology grant s / wi/1/08 ; natlia martins and delfim f. m. torres by the r&d unit `` centre for research in optimization and control '' ( ceoc ) of the university of aveiro , cofinanced by the european community fund feder / poci 2010 .
z. bartosiewicz , u. kotta , e. pawuszewicz and m. wyrwas , algebraic formalism of differential one - forms for nonlinear control systems on time scales , proc .
estonian acad .
phys . math . *
56 * ( 2007 ) , no . 3 , 264282 .
z. bartosiewicz , e. piotrowska and m. wyrwas , stability , stabilization and observers of linear control systems on time scales , proceedings of the 46th ieee conference on decision and control new orleans , la , usa , dec . 12 - 14 , 2007 , 28032808 .
p. d. f. gouveia and d. f. m. torres , automatic computation of conservation laws in the calculus of variations and optimal control , comput .
methods appl . math . * 5 * ( 2005 ) , no . 4 , 387409 .
arxiv : math/0509140 j. seiffertt , s. sanyal and d. c. wunsch , hamilton - jacobi - bellmam equations and approximate dynamic programming on time scales , ieee trans .
. , part b : cybern .
* 38 * ( 2008 ) , no . 4 , 918923 .
d. f. m. torres , carathodory equivalence , noether theorems , and tonelli full - regularity in the calculus of variations and optimal control , j. math .
( n. y. ) * 120 * ( 2004 ) , no . 1 , 10321050 .
arxiv : math.oc/0206230 | the fundamental problem of the calculus of variations on time scales concerns the minimization of a delta - integral over all trajectories satisfying given boundary conditions . in this paper
we prove the second euler - lagrange necessary optimality condition for optimal trajectories of variational problems on time scales . as an example of application of the main result
, we give an alternative and simpler proof to the noether theorem on time scales recently obtained in [ j. math .
anal .
appl . 342
( 2008 ) , no . 2 , 12201226 ] .
* mathematics subject classification 2000 : * 49k05 , 39a12 .
* keywords : * calculus of variations ; optimal control ; euler - lagrange , dubois - reymond , and second erdmann necessary optimality conditions ; noether s theorem ; time scales . |
topological superconductor is a novel phase of matter that has been theoretically predicted but yet to be experimentally verified . among topological materials ,
topological superconductivity is especially interesting because it is a platform to realize majorana particles - an elusive particle that is its own antiparticle .
furthermore , topological superconductors have been proposed as a platform for topological quantum computation .
@xcite the robustness of the topological surface states makes this avenue an attractive alternative to traditional methods for realizing quantum computation .
@xcite a topological superconductor must have a full superconducting gap in the bulk with odd parity pairing , and the fermi surface must enclose an odd number of time reversal invariant momenta in the brillouin zone , i.e. the fermi surface must contain an odd number of high symmetry points such as @xmath3 , z , x , etc .
it also has a topologically protected gapless surface state with majorana fermions .
@xcite cu@xmath0bi@xmath1se@xmath2 has been proposed as a leading candidate for topological superconductivity @xcite and has sparked a lot of interest .
experiments have shown that by intercalating cu between se layers in known topological insulator bi@xmath1se@xmath2 the compound becomes superconducting at 3.8 k. @xcite cu@xmath0bi@xmath1se@xmath2 has been confirmed to be a bulk superconductor with a full pairing gap by specific heat measurement .
@xcite there are some reports of surface andreev bound states through the observation of zero bias conductance peak ( zbcp ) @xcite , but other reports that the zbcp can be removed with gating .
@xcite recent works using scanning tunneling spectroscopy also did not observed the zbcp .
@xcite arpes measurements have argued against the topological superconducting mechanism in cu@xmath0bi@xmath1se@xmath2 by reporting an even number of time reversal invariant momenta in the brillouin zone .
@xcite both arpes and quantum oscillation experiments show a dirac dispersion in cu@xmath0bi@xmath1se@xmath2 - a characteristic feature of topological systems .
@xcite the continual interest in cu@xmath0bi@xmath1se@xmath2 and surmounting controversy of its exotic phase motivates this study for a more complete look at quantum oscillations in magnetization .
this work is a continuation and expansion on our previous study of the de haas - van alphen effect in cu@xmath4bi@xmath1se@xmath2 @xcite and now includes several samples at a variety of doping levels and complete angular dependence . from mapping out the fermi surface ,
we reveal a closed ellipsoidal fermi surface that becomes increasingly elongated with increased carrier density . at high carrier concentration
, the fermi surface crosses the brillouin zone boundary and becomes open and quasi - cylindrical .
amplitude damping analysis reveals a strongly anisotropic effective mass .
the slope of the energy - momentum dispersion is unchanged with increased fermi momentum confirming a linear , dirac - like band structure in cu@xmath0bi@xmath1se@xmath2 .
the manuscript will first introduce torque magnetometry which we use to resolve quantum oscillations , second it will discuss the results of the angular dependence of the quantum oscillation frequencies , then it will cover the various damping mechanisms of the quantum oscillation amplitude and the parameters extracted from that analysis .
single crystals of cu@xmath0bi@xmath1se@xmath2 were grown by melting stoichiometric mixtures of high purity elements bi ( 99.999% ) , cu ( 99.99% ) , and se ( 99.999% ) in a sealed evacuated quartz tube then slowly cooling the mixture from 850@xmath5c down to 620@xmath5c at which point the crystal was quenched in cold water .
the doping level was determined according to the mole ratio of the reactants used in the crystal growth , but the nominal doping did not end up corresponding with the measured carrier concentration leaving the precise number under suspicion .
therefore , in this study , we look at how parameters evolve with increased carrier concentration rather than the unreliable nominal doping .
the samples used in the study were cut out of large boule of crystals .
they are generally black , and the typical size is about 5 mm @xmath6 2 mm @xmath6 0.5 mm .
quantum oscillations are used to resolve fermi surface geometry and to discover electronic properties of topological materials .
oscillations in magnetization , the de haas - van alphen effect ( dhva effect ) , arise from the quantization of the fermi surface into landau levels . to measure quantum oscillations in magnetization , @xmath7 , we employed a highly sensitive torque magnetometry method .
torque mangetometry measures the magnetic susceptibility anisotropy of the sample by putting the sample in a tilted magnetic field , @xmath8 , where both @xmath8 and @xmath7 are confined to the x - z plane .
the torque is then given by @xmath9 and @xmath10 , where @xmath11 is the tilt angle of the magnetic field @xmath12 away from the crystalline @xmath13 axis and @xmath14 .
we glue the sample to the head of a thin film cantilever .
both brass cantilevers and kapton cantilevers with a metalized surface were used . the thinner 0.001 inch
brass cantilevers with a higher young s modulus and the thicker 0.003 inch katpon thin films with a lower young s modulus offer different spring constants that can provide a balance between strength for heavier samples and sensitivity .
the magnetic torque was tracked by measuring the capacitance between the metal surface of the cantilever and a thin gold film underneath .
an example of oscillations in the torque data after background subtraction is shown in fig .
[ figtorque ] with a schematic of the experimental setup in the upper right corner .
oscillations arise from landau level quantization .
the frequency of this oscillation is proportional to the cross section of the fermi surface , @xmath15 , by the onsager relation : @xmath16 to further analyze the oscillation torque pattern , a polynomial background is subtracted from the @xmath17 curve to get the oscillatory torque @xmath18 . a fast fourier transform ( fft ) of the oscillatory @xmath18 vs. @xmath19
is given in the lower left inset of fig .
[ figtorque ] revealing a single fermi pocket .
the dhva effect was observed in all of our cu@xmath0bi@xmath1se@xmath2 crystals .
a typical example of our torque data as a function of @xmath20 with the polynomial background subtracted is shown in fig .
[ figtorque ] . a single frequency of oscillations in magnetization reveal a single fermi pocket .
at 300 mk the oscillation frequency was measured as a function of angle up to 90@xmath5 .
we further measured the temperature dependence of the oscillation amplitude at two different angles up to 25 k. a detailed analysis of the temperature and angular dependence of the quantum oscillations revealed the following features : 1 ) resolved quantum oscillations up to 90@xmath5 at low carrier concentration shows a closed ellipsoidal fermi surface , 2 ) the fermi surface gets progressively elongated in the z - direction as carrier concentration increases and becomes open at high carrier density , 3 ) there is a strong effective mass anisotropy , and 4 ) the fermi velocity is unchanged with increased carrier concentration supporting previous reports of a linear , dirac - like dispersion in cu@xmath0bi@xmath1se@xmath2 .
figure [ figangtorq ] shows the torque signal from sample 4 . in panel
a , oscillations are clearly seen up to 90@xmath5 in raw data indicating a closed fermi surface .
panel b shows the fft of the raw signal from panel a. clear angular dependence can be tracked up to 90@xmath5 , where @xmath8 is parallel to the plane .
previous studies @xciteof the dhva effect measured quantum oscillations up to 35@xmath5 .
the observation of quantum oscillations up to 90@xmath5 is important confirmation of the previous result that the fermi surface is an ellipsoid @xcite .
we note the sample with this 3d fermi surface is superconducting , and the meissner effect of sample 4 is shown in fig .
[ figsupercon ] and discussed in detail there .
figure [ figang ] shows the angular dependence of the oscillation frequency for the various samples .
the dashed lines are ellipsoidal fits given by @xmath21+(\frac{k_{f}^{x}}{k_{f}^{z}})^{2}\sin^{2}[\phi])^{-\frac{1}{2}}$ ] where @xmath22 is the frequency of the quantum oscillations at a particular @xmath11 , and the fitted parameters are @xmath23 ( the quantum oscillation frequency at @xmath11 = 0@xmath5 ) and @xmath24 ( a measure of the eccentricity of the fermi surface ) .
most of the samples are fit well by a closed , ellipsoidal fermi surface ; however , for the highest carrier concentration sample , a closed fermi surface fitting yields @xmath25 = 4.69 nm@xmath26 , which is longer than the brillouin zone height @xcite of 3.28 nm@xmath26 .
thus , it is clear that the fermi surface becomes open at high carrier concentration - which was not seen in previous dhva studies where there was only one sample of lower doping .
@xcite the angular dependence of the quantum oscillation frequency provides the size of the fermi pocket . from the onsager relation ,
the frequency of the quantum oscillation is proportional to the cross - section area given by @xmath27@xmath28@xmath29 , with @xmath30 and @xmath31 two semi - axes of the elliptical fermi surface . thus @xmath23 yields @xmath28 = @xmath32 and the eccentricity gives @xmath25 . for sample 4a ,
@xmath28 = @xmath32 = 0.95nm@xmath26 and @xmath33 = 2.06 . for a closed fermi pocket , the bulk carrier concentration , @xmath34 ,
is given by @xmath35 .
for the sample with the open fermi surface , we calculated the bulk carrier concentration from finding the volume of the fermi surface .
this volume is arrived by integrating the ellipsoidal fit up to the brillouin zone boundary .
we assume that the deviation from the ellipsoidal fit around the brillouin zone boundary due to bending is small . in this case
the carrier concentration is given by @xmath36 where @xmath37 is the @xmath38 distance .
this yields a carrier concentration for sample 5 .
the inferred carrier densities @xmath34 are listed with other electronic parameters in table [ density ] .
.[density ] summary of results in order of increasing carrier concentration . *
the value of @xmath39 for sample 5 is ill - defined since @xmath40 is taller than the brillouin zone .
this is the value extracted from the ellipsoidal fit . [
cols="^,^,^,^,^",options="header " , ] [ parameter ] in addition to determining the electronic state , we further measured the the superconducting fraction of cu doped bi@xmath1se@xmath2 .
the magnetic susceptibility was measured in an quantum design magnetic property measurement system 2 weeks after the high field torque experiments .
the sample with the lowest carrier concentration shows a superconducting transition with a 16% superconducting volume as seen in fig .
[ figsupercon ] .
two of the higher carrier concentration samples showed no superconducting transition suggesting either that the sample quality deteriorates with time and exposure or the superconducting phase is killed in the over - doped regime . at high carrier concentration ,
the fermi surface becomes quasi - cylindrical and contains both the @xmath3 and z points .
this indicates that a topological superconducting state does not exist at high carrier concentration since the fermi surface must enclose an odd number of time reversal invariant momenta in the brillouin zone for a topological superconductor .
however , since superconductivity coexists with the closed fermi surface ( which can only contain the @xmath3 point ) in the low carrier density sample , a topological superconducting state can still exist in the lower carrier density samples of cu@xmath0bi@xmath1se@xmath2 .
( color online ) volume susceptibility measurements of 3 different samples .
sample 4 , the sample with lowest carrier concentration , shows a superconducting transition at 3 k and a 16% superconducting volume .
samples 3 and 5 do not show any superconducting property most likely due to sample quality degradation over time .
, width=3 ] we note that sample 4 , though having the highest level of disorder , is the only sample to show a superconducting transition .
the sample from the previous dhva study had a dingle temperature of 23.5 k much like the other samples and it also exhibited superconductivity .
@xcite therefore , it is not the case that superconductivity only occurs in the extreme case of highly disordered samples , rather we suggest that sample 4 had undergone the least amount of degradation and just happened to be the sample with the highest disorder .
quantum oscillations in magnetization were resolved using highly sensitive torque magnetometry up to 31 t. a single fermi pocket was observed to be increasingly elongated with added carriers .
the effective mass has strong anisotropy , and the fermi velocity remains unchanged with increasing fermi momentum suggesting a linear , dirac - like dispersion .
the nature of the transition of the fermi surface topology is an interesting question . at higher concentration
, the elongated 3d ellipsoidal fermi surface touches the fermi surface in the neighboring brillouin zone , mandating the transition from the 3d fermi surface to a 2d quasi - cylindrical one .
such an dramatic change of the fermi surface topology suggests a lifshitz transition as the cu brings it extra carriers .
two experimental consequences are essential to confirm the dimensionality change and probe the nature of the transition . first , at higher @xmath34
, the quasi-2d fermi surface shall have two quantum oscillation frequencies , a large one from the belly , and a small one from the neck .
the large frequency is what we observed in our dhva measurement , @xcite and confirmed by the sdh measurements .
@xcite in contrast , the small neck frequency was not observed either in our dhva measurements , nor in the sdh results .
@xcite this point is in particular puzzling , though ref .
@xcite reports that the signal arising from the small neck may be too small do to large effective mass effects .
therefore , the quantum oscillation measurements at dilution refrigerator temperature range is called to resolve the second oscillation frequency to confirm the 2d to 3d transition
. another interesting experiment would be to the enhancement of thermopower near the 3d to 2d transition .
a topology change in the electronic state usually leads to a large thermopower , a typical signature of lifshitz transition . @xcite
further thermoelectric measurements are essential to confirm this nature .
if the dimensionality changes indeed occur and enhance the thermopower greatly , the cu doping might lead to another interesting application of topological materials in thermoelectrics . * acknowledgement * we are grateful to discussion with liang fu , kai sun , and a. kanigel , .
the work is supported by the national science foundation under award number eccs-1307744 ( low field torque magnetometry ) , the department of energy under award number de - sc0008110 ( high field torque magnetometry ) , by the start up fund and the mcubed project at the university of michigan ( low field magnetic susceptibility characterization ) , and by the national science foundation under award number dmr-1255607 ( sample growth ) .
the high - field experiments were performed at the national high magnetic field laboratory , which is supported by nsf cooperative agreement no .
dmr-084173 , by the state of florida , and by the doe .
we thank the assistance of tim murphy and ju - hyun park of nhmfl .
b. j. lawson acknowledges the support by the national science foundation graduate research fellowship under grant no .
f031543 . t.
asaba thanks the support from the nakajima foundation . | cu@xmath0bi@xmath1se@xmath2 has drawn much attention as the leading candidate to be the first topological superconductor and the realization of coveted majorana particles in a condensed matter system .
however , there has been increasing controversy about the nature of its superconducting phase .
this study sheds light on present ambiguity in the normal state electronic state , by providing a complete look at the quantum oscillations in magnetization in cu@xmath0bi@xmath1se@xmath2 at intense high fields up to 31 t .
our study focuses on the angular dependence of the quantum oscillation pattern in a low carrier concentration .
as magnetic field tilts from along the crystalline c - axis to ab - plane , the change of the oscillation period follows the prediction of the ellipsoidal fermi surface . as the doping level changes , the 3d fermi surface is found to transform into quasi - cylindrical at high carrier density .
such a transition is potentially a lifshitz transition of the electronic state in cu@xmath0bi@xmath1se@xmath2 . |
observations of lyman-@xmath2 ( ly@xmath2 ) sources are a powerful probe of the high - redshift universe ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in particular , the prominence of the ly@xmath2 line at @xmath8 allows for spectroscopic confirmation of redshift measurements of individual distant galaxies .
ly@xmath2 sources are also a compelling probe of the cosmic dark ages leading up to reionization
see @xcite for a perspective on high-@xmath6 observations .
historically , @xcite determined that galaxies from the first billion years after the big bang would be powerful emitters of ly@xmath2 photons , though observations of these sources eluded us for longer than expected .
however , robust detections are becoming more regular , especially if the stellar mass is comparable to the milky way or the star formation rate ( sfr ) is elevated ( e.g. sfr @xmath9 ; * ? ? ?
* ) . within the earliest galaxies hard uv radiation from massive stars
is reprocessed into ly@xmath2 photons ; however , because neutral hydrogen ( hi ) is opaque to the ly@xmath2 line , many of these photons may be resonantly trapped , and consequently suffer significant dust absorption . despite these effects , observations have determined that the ly@xmath2 escape fraction , @xmath10 , actually increases at higher redshifts @xcite . at some point , although the photons are no longer destroyed by dust , they are scattered out of the line of sight and some fraction of the ly@xmath2 emission is lost to the background as their sources become spatially extended ly@xmath2 haloes @xcite .
various mechanisms have been explored to explain the unusually high @xmath10 of high-@xmath6 galaxies . in all likelihood
this is a result of the complicated resonant line transfer , galactic structure , and peculiar dust properties .
for example , multiple scatterings that facilitate excursions to the wings of the frequency profile ; large - scale flows that induce doppler shifts ; and the geometry of dense , dusty clouds within a clumpy interstellar medium that provide pathways for escape ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in this work ,
we push these questions to the very first galaxies ( for a review see * ? ? ?
assessing the observability of such early ly@xmath2 sources is nontrivial . indeed , going to higher redshifts introduces physical effects that compete in either strengthening or attenuating the ly@xmath2 signal @xcite . on one hand ,
the intergalactic medium ( igm ) becomes increasingly neutral at higher @xmath6 , resulting in a more difficult escape for ly@xmath2 photons @xcite . on the other hand
, the same igm also becomes increasingly devoid of dust @xcite .
furthermore , population iii ( so - called pop iii ) stellar sources are predicted to have been more efficient ionizers , boosting the ly@xmath2 luminosity @xcite .
the fact remains that high - redshift ly@xmath2 sources are being observed out to @xmath11 .
still , many details regarding the epoch of reionization ( eor ) , or the inhomogeneous phase transition around @xmath12 , are uncertain and may greatly affect interpretations of ly@xmath2 transfer through the igm @xcite .
some of the most effective methods for identifying high - redshift objects involve the ly@xmath2 line .
in particular , lyman - break galaxies ( lbgs ) are generally massive galaxies for which neutral hydrogen produces a sharp drop in the spectra due to absorption @xcite .
lyman-@xmath2 emitters ( laes ) are young , less - massive galaxies with active star formation and strong ly@xmath2 emission @xcite .
it is an important frontier to push ly@xmath2 selection methods towards the highest possible redshifts .
for galaxies at @xmath13 the neutral fraction of the intervening igm increases enough for their spectra to yield complete absorption of photons blueward of the ly@xmath2 line .
this is the well - known `` gunn - peterson trough '' @xcite which is characteristic of lbgs .
however , these massive , evolved galaxies become increasingly rare at high redshifts .
the lae luminosity function also declines as redshift increases , and the observed trend is robustly established for @xmath14 ( e.g. * ? ? ?
* ; * ? ? ?
* ) and expected to continue beyond @xmath15 @xcite . because a strong detection of the highly - redshifted ly@xmath2 line requires the emitter to be young and relatively dust free conditions which are naturally expected for the first galaxies laes are likely their typical manifestation .
high - redshift ly@xmath2 candidates must be followed up by spectroscopy in order to guard against false positives from foreground contaminants .
fortunately , moderate- to high-@xmath6 surveys are underway that will dramatically increase the sample size of ly@xmath2 galaxies and better characterize their statistical properties . for example , the hobby - eberly telescope dark energy experiment ( hetdex ) is a large integral - field spectroscopic survey expected to detect a million laes @xcite .
currently , there are several candidates at @xmath16 ( e.g. * ? ? ?
* ) , with the highest spectroscopically - confirmed source announced at @xmath17 by @xcite .
other records have been found using gamma - ray bursts ( grbs ) , active galactic nuclei ( agn ) , or ( sub-)mm observations of redshifted thermal dust emission @xcite .
however , it is unclear how these other selection methods relate to ly@xmath2 predictions .
such connections may complement ly@xmath2 observations , even if the phenomena originate from the luminous deaths of individual massive stars ( i.e. grbs ) or are not associated with a ` normal ' activity of the first galaxies ( e.g. agn or high amounts of dust ) .
ly@xmath2 radiative transfer within the first galaxies is a timely problem because next - generation facilities will provide high resolution data by the end of the decade .
the _ james webb space telescope _ ( _ jwst _ ; * ? ? ?
* ) and large - aperture ground - based observatories ) , thirty meter telescope ( tmt ; http://www.tmt.org/[www.tmt.org ] ) , and the european extremely large telescope ( e - elt ; http://www.eso.org/sci/facilities/eelt/[www.eso.org/sci/facilities/eelt ] ) . ] offer the prime avenue for observing ly@xmath2 emission at the high-@xmath6 frontier and will significantly contribute to our understanding of the ionization history at the end of the dark ages @xcite .
however , significant progress has also been made on a number of complementary probes of the high-@xmath6 universe .
several 21-cm array experiments ) , the murchison wide - field array ( mwa ; http://www.mwatelescope.org/[www.mwatelescope.org ] ) , the precision array to probe the epoch of reionization ( paper ; http://eor.berkeley.edu/[eor.berkeley.edu ] ) , and ultimately the square kilometer array ( ska ; http://www.skatelescope.org/[www.skatelescope.org ] ) . ] are coming online to map the distribution of hi over the course of reionization .
the ly@xmath2 and 21-cm lines are related through the wouthuysen - field mechanism for which ly@xmath2 scatterings pump electrons into the excited hyperfine state , thereby coupling the spin and kinetic temperatures @xcite . finally , an ideal complementary ly@xmath2 probe is encoded in the cosmic infrared background ( cib ) because the integrated radiation from all background stars and galaxies has been redshifted to ir wavelengths @xcite .
the ly@xmath2 contribution is seen through the correlation of sources across characteristic length scales ( for a review see * ? ? ?
a number of authors have studied ly@xmath2 radiative transfer within different contexts .
we have greatly benefited from and hope to add to the body of work in this area .
a partial list of references include : @xcite .
the state of the art is to apply post - processing radiative transfer to realistic hydrodynamical simulations , which is justified for many large scale systems .
we use this method in conjunction with semi - analytic models to ascertain the feedback of ly@xmath2 radiation on the galactic assembly process .
our focus on the very first galaxies , in their proper cosmological context , is different from previous research that has targeted more massive systems at redshifts close to , or after , reionization ( e.g. * ? ? ?
such systems require a statistical description of ly@xmath2 transmission through the igm as described by @xcite and @xcite .
photons that scatter out of the line of sight due to the neutral fraction of the igm are effectively lost to the background .
therefore , the observability of ly@xmath2 emitters will provide independent constraints on reionization @xcite .
measurements of the ly@xmath2 flux from first galaxies depend heavily on the observed line of sight in addition to the properties of the host system and igm .
therefore , rather than solving a potentially intractable transfer equation with complex angular dependence we take advantage of monte - carlo radiative transfer techniques to accurately build emergent spectral energy distributions ( seds ) . in order to perform ly@xmath2 simulations
we have developed a new massively parallel code called @xmath4 the @xmath0smic @xmath1yman-@xmath2 @xmath3ransfer code . in section [ subsec :
rs ] , the basic physics of ly@xmath2 transport is presented . in section [ sec : methodology ] , we discuss the general methodology behind @xmath4 and provide further algorithmic details in section [ sec : schemes ] . in section [ sec : tests ] , the code is tested against both static and dynamic setups . in section [ sec :
toy ] , we construct idealized analytical models to explore how fundamental parameters , including halo mass , virialization redshift , bulk velocity , and ionization structure , affect ly@xmath2 transport in the first galaxies .
these well - motivated models help test our methods and sensitivity .
section [ sec : sim ] describes our implementation of the _ ab initio _ cosmological simulation of @xcite as post - processing conditions for @xmath4 . in section [ sec : results ]
we analyze and discuss the emergent line of sight flux distributions and surface brightness profiles for both the idealized analytic cases and the cosmological simulation .
finally , in section [ sec : conc ] we reflect on the implications of this study with regard to future ly@xmath2 observations with the _ jwst_.
photons with frequencies close to the ly@xmath2 resonance line , corresponding to the transition from the first excited state ( 2p ) to the ground state ( 1s ) , may be absorbed and quickly re - emitted by neutral hydrogen .
thus , in optically thick environments ( @xmath19 ) the main mechanism of spatial diffusion is by the rare excursion to the lorentz wing of the spectral line .
the cross section @xmath20 and number density @xmath21 describe the optical depth @xmath22 along a photon s path : @xmath23 where @xmath24 specifies the frequency dependence .
it is often convenient to express frequency in a dimensionless manner as the number of doppler widths from line centre @xmath25 where @xmath26 hz is the frequency of ly@xmath2 and the thermal doppler width of the profile is @xmath27 .
the thermal velocity in terms of @xmath28 is @xmath29 furthermore , if the natural ly@xmath2 line width is @xmath30 hz then the ` damping parameter ' represents the relative broadening of the natural line width : @xmath31 therefore , the final cross section is @xmath32 where @xmath33 is the oscillator strength of the ly@xmath2 transition and the hjerting - voigt function @xmath34 is the dimensionless convolution of lorentzian and maxwellian distributions , @xmath35 for reference , we define the cross section at line centre as @xmath36 @xmath37 . for a region of constant density @xmath21 such as a cell in a computational domain the integral in equation ( [ eq : tau ] ) simplifies to @xmath38 typically , the parameter @xmath39 is much less than unity so the hjerting - voigt function is dominated in the centre by a resonant scattering doppler core , @xmath40 , and the wings are dominated by the lorentzian component , @xmath41 .
if the approximate frequency marking the crossover from core to wing is denoted by @xmath42 , i.e. where @xmath43 , then the hjerting - voigt function is roughly @xmath44 see sections [ subsub : h_approx ] and [ subsub : x_cw_approx ] for a more rigorous discussion as well as numerical approximations for @xmath34 and @xmath42 . for most conditions with significant neutral hydrogen density
the gas is optically thick to core photons .
however , in the wing , @xmath34 can be quite small allowing a photon to escape with greater ease .
the approximate optical depth for a wing photon is then @xmath45 which is independent of temperature as is immediately apparent when written in terms of the doppler velocity @xmath46 .
note the relation between doppler frequency and velocity : @xmath47 .
regions of high column density , @xmath48 pc @xmath49 @xmath50 , may produce very high opacities .
the trapped ly@xmath2 photons are therefore exposed to greater extinction from dust .
this may be best measured by the optical depth at line centre , which has moderate temperature dependence and can be read from equation ( [ eq : tauh ] ) as @xmath51~t_4^{-1/2}$ ] .
the @xmath4 code is based on previous monte - carlo radiative transfer ( mcrt ) algorithms ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the code reads initial conditions of velocity , density , and temperature for each cell of a three - dimensional grid employing adaptive mesh refinement ( amr ) . sampling the ly@xmath2 emission profile of high opacity systems is possible because acceleration schemes avoid unnecessary computations , e.g. frequent core scatterings .
additionally , @xmath4 is massively parallel allowing a greater number of photon packets and therefore less statistical error .
section [ sec : methodology ] outlines the general methodology for a ly@xmath2 transport code while section [ sec : schemes ] describes the specific implementations used in @xmath4 .
while the monte - carlo method for ly@xmath2 radiative transfer is fairly standard , each portion of section [ sec : schemes ] contains significant discussion or numerical schemes unique to this work . in order to sample ly@xmath2 observables
we apply the method described below to individual photon packets . depending on the desired resolution and physical setup a large number of packets
may be necessary to obtain statistical convergence .
unless otherwise specified the number of photon packets in each simulation is @xmath52 .
the initial spatial distribution of ly@xmath2 photons is based on the physical setup , i.e. initial conditions .
the first mechanism for producing ly@xmath2 line emission is interstellar recombination as a result of ionizing radiation from hot stars .
the second mechanism is collisional excitation of neutral hydrogen , usually resulting from shocks caused by accretion or supernovae .
an initialization criterion accounts for photoionizing sources and diffuse emission , both of which may be significant for atomic cooling haloes .
however , throughout this work we choose to initialize photons from the central location @xmath53 .
the initial direction @xmath54 of each photon is drawn from an isotropic distribution in the rest frame of the embedded source . for convenience ,
velocities are expressed in terms of the thermal velocity : @xmath55 the photon is emitted at the natural frequency of the ly@xmath2 photon @xmath56 in the rest frame of the atom . to obtain the initial frequency @xmath57 in the moving frame of ambient gas
we apply a doppler shift appropriate to the velocity of the atom @xmath58 @xcite @xmath59 to be explicit , @xmath56 is drawn from a lorentzian distribution and the components of @xmath60 are each taken from a maxwellian distribution describing the thermal motion of the ambient gas .
although we use the expression in equation ( [ eq : init_freq ] ) , the memory of @xmath57 is quickly lost by multiple scattering events , so for optically thick environments one may simply inject the photon at line centre , or @xmath61 .
the propagation distance of any photon is determined by the optical depth @xmath22 drawn from an exponential distribution .
this is because the mean optical depth @xmath62 defines the mean free path @xmath63 , or the average distance a photon can travel without being absorbed ( or scattered ) by the intervening medium @xcite .
therefore , travel beyond each mean free path becomes less and less probable .
formally , the mean free path is then @xmath64 . from equation ( [ eq : tau ] )
the respective optical depths of infinitesimal paths are @xmath65 so the probability of _ no _ interaction is @xmath66 . with @xmath67 partitions of @xmath68 , the numerical representation of discrete intervals ,
the probability distribution function over the total path is the product @xmath69 which has been normalized so that @xmath70 .
the cumulative distribution function is the integrated distribution defined by @xmath71 , which can be inverted to give the optical depth at which an interaction event occurs , @xmath72 where @xmath73 is drawn from a univariate distribution . to perform cartesian - like ray tracing
, the monte - carlo method selects a photon with an optical depth @xmath74 according to equation ( [ eq : lnr ] ) . because individual cells are regions of uniform density we may equate this with the calculated optical depth @xmath75 from equation ( [ eq : tauh ] ) to find the propagation distance , i.e. @xmath76 provides @xmath77
. however , this often leads to a scattering which is outside the original cell .
therefore , we first calculate the optical depth required to travel through the cell : @xmath78 , where @xmath79 is the distance from the current position to the point where the photon exits the cell .
if @xmath80 the ` spent ' optical depth is subtracted from the current optical depth , i.e. @xmath81 .
likewise , the current position is updated to @xmath82 and ray tracing continues through the next cell .
as photons traverse from cell to cell the temperature @xmath83 and ambient bulk velocity @xmath84 may change .
therefore , doppler shifting induces terms of @xmath85 corresponding to a nonrelativistic lorentz transformation , i.e. one where @xmath86 .
additionally , the frequency differs between cells due to the @xmath87 scaling relation .
the frequency in the new ( primed ) cell is then @xmath88 if @xmath89 the optical depth is ` exhausted ' in the current cell and the photon undergoes a scattering event .
the propagation distance is recalculated according to the remaining optical depth , i.e. @xmath90 $ ] .
the position is advanced by @xmath91 in the @xmath54 direction and the algorithm proceeds to update the frequency and direction as described in section [ sec : scatter ] .
the photon ray traces along a new direction with each subsequent scattering until it ultimately escapes the computational domain .
a scattering event changes the photon s frequency according to the atom s velocity @xmath60 , the final @xmath92 and initial @xmath54 directions , and a small recoil effect to satisfy conservation of momentum @xmath93 the recoil parameter @xmath94 is defined as ( see e.g. * ? ? ?
* ) @xmath95 which is included but negligible for the applications of this paper .
the velocity of the scattering atom is most conveniently expressed in terms of its parallel @xmath96 and perpendicular @xmath97 components .
the perpendicular magnitudes are unbiased by the photon s frequency and are therefore drawn from a gaussian , i.e. @xmath98 . however , the magnitude of the parallel velocity @xmath99 is affected by the presence of the resonance line .
the distribution function for @xmath100 depends on frequency as the convolution of a gaussian with a doppler - shifted lorentzian peak : @xmath101 which highly favors velocities with @xmath102 for core photons . for wing photons
( @xmath103 ) , however , the probability that an atom has a high enough velocity to doppler shift into resonance becomes vanishingly small , so @xmath104 becomes increasingly gaussian .
the angle @xmath105 between the incident and outgoing scattering directions is governed by the phase ( probability ) function @xmath106 where @xmath107 is the degree of polarization for @xmath108 scattering . due to the physical symmetry
the phase function is independent of the azimuthal angle @xmath109 . for core photons with @xmath110 the electron transition to the @xmath111 state results in isotropic scattering , i.e. @xmath112 , while the @xmath113 transition results in polarization with @xmath114 @xcite .
since a quantum state with angular momentum @xmath115 has a spin multiplicity of @xmath116 , ly@xmath2 photons in the core are excited to the @xmath111 state with @xmath117 probability and the @xmath113 state with @xmath118 probability .
nonresonant wing photons on the other hand are dominated by rayleigh scattering because their wavelength is much larger than the bohr radius , @xmath119 , therefore the resultant polarization is maximal , i.e. @xmath120 @xcite .
anisotropic scattering is used throughout @xmath4 .
the previous section was written with little regard to computational efficiency , something we now seek to remedy .
the details are presented in order of introduction , rather than importance . and
our approximation for different values of temperature .
the crossover from core to wing @xmath42 as given by equ .
( [ eq : xcw ] ) is also shown .
bottom panel : a demonstration of the relative per cent error @xmath121 $ ] where @xmath122 is given by equ .
( [ eq : ab ] ) .
the yellow , blue , green , and red curves correspond to temperatures of @xmath123 k , @xmath124 k , @xmath125 k and @xmath126 k , respectively .
the overlying dashed lines represent the best case scenario when only keeping first order terms in @xmath39 , i.e. using the exact dawson integral.,title="fig : " ] and our approximation for different values of temperature .
the crossover from core to wing @xmath42 as given by equ .
( [ eq : xcw ] ) is also shown .
bottom panel : a demonstration of the relative per cent error @xmath121 $ ] where @xmath122 is given by equ .
( [ eq : ab ] ) .
the yellow , blue , green , and red curves correspond to temperatures of @xmath123 k , @xmath124 k , @xmath125 k and @xmath126 k , respectively . the overlying dashed lines represent the best case scenario when only keeping first order terms in @xmath39 , i.e. using the exact dawson integral.,title="fig : " ] an approximation for the voigt profile is important because @xmath34 is evaluated after every scattering . a substantial effort has gone into studying this profile and implementing efficient algorithms with double precision accuracy
see e.g. @xcite and references therein .
however , the ly@xmath2 resonance line is a unique application with a specific parameter range , so the approximations used in this work are entirely our own .
we require our algorithm to provide better than one per cent accuracy for all frequencies and all realistic temperatures . to do this we first evaluate the integral @xmath34 with special functions and expand to second order in @xmath39 : @xmath127 where the ( complex ) complementary error function is related to the area under a gaussian by @xmath128 and the associated
dawson integral is @xmath129 .
@xmath4 takes advantage of the fact that @xmath34 is symmetric by first evaluating @xmath130 .
then because the behavior of the profile differs for small and large @xmath6 the domain is decomposed into three regions .
the ` core ' region is derived by expanding around @xmath131 , while the ` wing ' region is an asymptotic expansion .
an intermediate region acts as a smooth transition between the two .
the approximations utilize continued fractions in order to maximize the efficiency of a small number of operations , e.g. @xmath132 additions and @xmath133 divisions .
see appendix [ appendix : h_approx ] for the implementation of @xmath134 .
second and higher order terms in @xmath39 are required to achieve one per cent accuracy for sub - kelvin temperatures . however , since the cmb temperature floor prevents gas from reaching such low temperatures only first order terms are used in @xmath4 .
figure [ fig : herror ] demonstrates the relative error , i.e. @xmath121 $ ] , for temperatures ranging from @xmath135 k. as can be seen , the error is well controlled , improving even further for higher temperatures .
$ ] , is shown in green .
the yellow curve in the insert is @xmath136 as given by equation ( [ eq : xcw ] ) .
all curves cover a temperature range of @xmath137 $ ] k. ] the crossover from core to wing @xmath42 determines when @xmath34 changes from a gaussian to a lorentzian .
this is important because @xmath42 marks the point where difficulties arise in generating @xmath100 and also identify whether core skipping is necessary ( see sections [ sec : upar ] and [ sec : coreskip ] ) .
our calculation assumes that the core and wing limits of @xmath34 compete in their contribution to the profile : @xmath138 equation ( [ eq : xcwa ] ) provides a conservative approximation for @xmath42 , providing a sharp boundary with the core ( see fig .
[ fig : herror ] ) . the exact solution can be written in terms of the lower branch of the lambert @xmath139 function , which in turn may be approximated by a low - order rational function with a relative accuracy of less than @xmath140 per cent for the same parameter range as @xmath122 . in summary ,
our optimal calculation of the core - to - wing crossover frequency is @xmath141 where @xmath142 and @xmath143 $ ] .
the relative per cent error is shown in fig .
[ fig : xcwerror ] .
as described in section [ sec : scatter ] ( cf .
equation [ eq : uparf ] ) the distribution for the parallel velocity @xmath100 with respect to the incoming photon is @xmath144 to good approximation this profile resembles a gaussian with a sharp peak around the point @xmath102 ( see fig .
[ fig : upar ] ) . unfortunately , @xmath104 is not integrable so we instead use the inverted cumulative distribution function method on a related distribution and employ the rejection method to accept each draw . in this case
the comparison function is chosen to be @xmath145 and draws are accepted with a probability of @xmath146 . however , we only employ this algorithm for small frequencies , i.e. @xmath147 , because as @xmath148 increases the method becomes quite inefficient .
we learn more about @xmath104 by examining its behavior when @xmath149 . here
the peak at @xmath102 is pushed so far into the wing that there are essentially no atoms with speeds fast enough to absorb at the doppler shifted resonance line , i.e. @xmath150 .
therefore , we may simply draw from a proper gaussian with a slight modification the peak is shifted by @xmath151 .
we demonstrate this by finding the local extrema in the core : @xmath152 here we have assumed @xmath153 .
it is straightforward to show @xmath154 so this is a local maximum as expected .
therefore , for @xmath155 we draw from a gaussian with mean @xmath151 .
as given by equ .
( [ eq : uparf ] ) for different values of incoming frequency @xmath148 .
the profile resembles a gaussian with a sharp peak around the point @xmath102 . for large @xmath148
it becomes too improbable for atoms to have velocities high enough to doppler shift into resonance so the peak at @xmath102 disappears and a shifted gaussian is a good approximation .
the frequencies sampled are @xmath156 . ]
the intermediate region is problematic for either of these methods .
therefore , we follow @xcite and use the piecewise comparison function : @xmath157 & \ ; u_\| \leq u_0 \\
g_2 = e^{-u_0 ^ 2 } / \left [ a^2 + ( x - u_\|)^2 \right ] & \ ; u_\| > u_0 \end{cases}\ ] ] where @xmath158 is a separation parameter and the corresponding acceptance fraction is @xmath159 for @xmath160 and @xmath161 for @xmath162 .
we now restrict the discussion to positive @xmath148 , which is possible because @xmath163 allows us to recover velocities drawn from negative @xmath148 . the probability that a velocity is less than @xmath158 is @xmath164 where @xmath165
if @xmath166 , a univariate , then @xmath105 is drawn uniformly from the interval @xmath167 $ ] , otherwise @xmath168 $ ] . finally , a velocity candidate , @xmath169 is accepted if another univariate , @xmath170 , is less than the acceptance fraction , i.e. @xmath159 or @xmath161 for each respective region . the algorithm from @xcite works
well as long as the probability of the two regions are balanced , i.e. in @xmath4 we attempt to maintain @xmath171 .
the reason for this is that @xmath158 controls the acceptance fraction , so if @xmath158 is too small then we do not gain much for larger @xmath148 and conversely if @xmath158 is too large then we defeat the purpose for smaller @xmath148 .
the exact value of @xmath172 is very sensitive to both @xmath148 and @xmath39 , so we can only hope for average acceptance rates to be reasonable given the variance in @xmath173 . if we assume @xmath174 then to first order in @xmath39 we have : @xmath175 the behavior of this function is different for core and wing photons . in the core
it is reasonable to assume a perturbation from the natural peak location and apply the transformation @xmath176 where @xmath177 admits an analytic solution to the approximate balance of @xmath171 .
again , to first order in @xmath39 equation ( [ eq : papprox ] ) becomes @xmath178 the solution in terms of @xmath179 is @xmath180 where the final equality allows the approximation to extend to larger @xmath148 and is used in @xmath4 for @xmath181 . for the region in which the wing dominates ,
i.e. @xmath182 , the algorithm suffers an identity crisis .
as discussed earlier , the high-@xmath148 behavior of @xmath183 approaches a gaussian distribution though each of the candidate @xmath100 samples are from a lorentzian distribution .
therefore , the exact value of @xmath158 is less important as long as it is greater than @xmath42 .
we test various prescriptions for @xmath184 , also varying initial frequency @xmath148 and temperature @xmath83 , in order to minimize the average number of draws in a simulation .
a linear function , matched to the previous region provides sufficient acceptance of candidate random numbers .
we use the following separation constant for @xmath185 , denoted the ` wing ' region : @xmath186 figure [ fig : u0p ] demonstrates the fractional probability @xmath187 for drawing @xmath188 given our piecewise prescription of @xmath189 for @xmath181 and @xmath190 for @xmath191 .
the efficiency of the algorithm is especially sensitive to frequency .
that @xmath188 .
this separation captures the behavior of @xmath94 for frequencies above and below the transitional frequency @xmath42 .
the axes are frequency @xmath192 and ( log@xmath193 ) temperature @xmath194 k. ] in optically thick regimes photons spend much of their time undergoing core scatterings with negligible diffusion in physical or frequency space . these scatterings can be avoided by only selecting atoms with perpendicular velocity components greater than a critical frequency , i.e. photons have zero mean free path if @xmath195 .
following @xcite we employ the box - muller method to generate two independent draws according to : @xmath196 where @xmath197 and @xmath198 are univariates .
the speedup time achieved here is significant since the probability of drawing a wing photon is roughly @xmath199 corresponding to skipping roughly @xmath200 core scatterings . the crucial problem is to find an appropriate value for the critical frequency @xmath201 .
@xmath4 introduces an algorithm with core - skipping based on both local and nonlocal criteria .
this is especially important for high resolution , adaptively structured grids where the range of densities may cover several orders of magnitude .
we desire a local determination of @xmath201 that accelerates the code but does not artificially push photons too far into the wings .
as noted by @xcite the important parameter is the product @xmath202 so we seek a relation of the form @xmath203 .
the model we consider for the near zone environment is that of an optically thick static uniform sphere .
this is motivated by the idealized geometry and an analytic solution for the angular averaged intensity @xmath204 at the surface first given by @xcite , @xmath205 which has been normalized to @xmath206 , reflecting an integration over solid angle .
the peaks are located at @xmath207 , which is derived by solving the equation @xmath208 , or equivalently @xmath209 with @xmath210 .
therefore , the peak heights correspond to @xmath211 .
we next expand equation ( [ eq : sphere ] ) around @xmath212 and define @xmath201 for large @xmath202 as the frequency where @xmath213 reaches a small fraction of @xmath214 , giving @xmath215 .
thorough tests demonstrate this expression for @xmath201 is valid for all @xmath216 ( see appendix [ appendix : x_crit ] ) .
furthermore , it has a negligible effect on the emergent spectrum but greatly reduces the computation time . in summary
, @xmath4 uses the following approximation : @xmath217 if the photon is already in the wing we do not use a cutoff because there are no core scatterings to skip .
the final ingredient in equation ( [ eq : xcrit ] ) to be explained is how to calculate the product @xmath202 .
@xmath4 employs a combination of local and nonlocal estimates of how aggressive to be with core - skipping . as we can not possibly predetermine the escape path of a given photon we instead place a conservative limit on @xmath202 for different lines of sight
the local criterion is computed on - the - fly , using the minimum optical depth to the edge of the current cell @xmath218 . however , in highly refined regions ( perhaps with partial ionization ) and for scattering events near cell edges this can be quite small , i.e. if @xmath79 is the minimum distance to the cell boundary equation ( [ eq : tauh ] ) gives @xmath219 it is apparent that if @xmath21 is roughly constant then the only nonlocal quantity needed is the physical size of the system .
therefore , the minimum integrated column density along rays emanating from the scattering event sets up an effective sphere with the intensity of equation ( [ eq : sphere ] ) . rather than calculate this at every scattering we combine the cell - based determination with a nonlocal ( nl ) estimate : @xmath220 where the path is followed
as long as the relative change in neutral hydrogen density remains less than a prescribed threshold , i.e. @xmath221 . a value of @xmath222 ignores the nonlocal scheme completely and setting @xmath223 is too aggressive , failing to even detect sharp ionization fronts . to avoid double counting the local contribution ,
the paths originate at the edge of each cell and proceed outward . in practice
the paths include at least the six directions of the coordinate axes and possibly more to achieve greater angular coverage .
the nonlocal estimate can be computed once for each cell as the initial conditions are read in .
@xmath4 currently uses the sum of the local determination @xmath218 and the nonlocal estimate @xmath224 as the value of @xmath202 used in equation ( [ eq : xcrit ] ) .
we note that other criteria could be used for estimating @xmath202 .
for example , the nonlocal integration might stop if the velocity gradient exceeds the threshold for sobolev escape ; however , in most cases this is a secondary factor with little affect on core - skipping .
finally , additional local distances may be used in certain cases .
of special interest is to use the jeans length to estimate core - skipping in idealized galactic setups .
the jeans length @xmath225 acts as a physical upper limit for the size of a system with uniform density @xmath226 .
a primordial gas with mean molecular weight @xmath227 , mass fraction of hydrogen @xmath228 , and density @xmath229 corresponds to a jeans length of @xmath230 and a local @xmath202 of @xmath231 so that @xmath232 for @xmath233 k and @xmath234 .
monte carlo codes benefit greatly from parallel computation because each photon packet is an independent event .
therefore , the scalability is nearly linear .
@xmath4 uses the message passing interface ( mpi ) libraries to implement dynamic load balancing between different processors .
this is done because photons that undergo many scatterings take longer to escape .
therefore , the master process allocates a reasonable amount of work to each slave , e.g. ( @xmath235% of the photons)/(number of processors ) , until the required number of photons are assigned .
the nonlocal determination of @xmath236 for each cell , see equation ( [ eq : ataunl ] ) , is also performed with an efficient parallel computation .
at this stage it is important to verify the code against known solutions .
we choose tests that are complementary to each other in order to isolate certain key aspects of ly@xmath2 transport .
the static test is the well - known neufeld analytical solution for an optically thick homogeneous slab @xcite .
the dynamic test is that of an isotropically expanding sphere with different maximal velocities as described by @xcite . of a static , homogeneous slab at @xmath233 k. the central optical depth
@xmath237 covers a range of moderate to extreme values , where the yellow , blue , green , and red curves represent @xmath4 simulations using @xmath238 photon packets for @xmath239 , @xmath240 , @xmath241 , and @xmath242 , respectively .
the agreement with the approximate analytic solutions ( thin black lines ) is quite good , especially for higher values of @xmath237 . at lower optical depths the underlying assumptions used to derive equation ( [ eq : neufeld ] ) break down , therefore the true measure of this test
is the limiting behavior as @xmath237 tends to infinity .
bottom panel : face - on line of sight radial surface brightness profiles for the same slabs as calculated by equation ( [ eq : sb ] ) of section [ sub : surface_brightness_construction ] .
the scaling is such that the highest optical depth system is unity at the centre , i.e. @xmath243 . the dashed line is a rough analytic fit to guide the eye .
, title="fig : " ] of a static , homogeneous slab at @xmath233 k. the central optical depth @xmath237 covers a range of moderate to extreme values , where the yellow , blue , green , and red curves represent @xmath4 simulations using @xmath238 photon packets for @xmath239 , @xmath240 , @xmath241 , and @xmath242 , respectively .
the agreement with the approximate analytic solutions ( thin black lines ) is quite good , especially for higher values of @xmath237 . at lower optical depths the underlying assumptions used to derive equation ( [ eq : neufeld ] ) break down , therefore the true measure of this test
is the limiting behavior as @xmath237 tends to infinity .
bottom panel : face - on line of sight radial surface brightness profiles for the same slabs as calculated by equation ( [ eq : sb ] ) of section [ sub : surface_brightness_construction ] .
the scaling is such that the highest optical depth system is unity at the centre , i.e. @xmath243 . the dashed line is a rough analytic fit to guide the eye .
, title="fig : " ] the angular averaged intensity @xmath244 for a static one - dimensional uniform slab was derived by @xcite and @xcite .
photons are injected at line centre at the origin and continuously scatter until escape occurs at a ` centre - to - edge ' optical depth of @xmath237 .
when the bulk motion of the gas is set to zero , @xmath245 , and the recoil effect is ignored , @xmath246 , the emergent spectra are double - peaked and symmetric around the line - central frequency @xmath247 .
the intensity at the surface is given by @xmath248 which has been normalized to @xmath206 , reflecting an integration over solid angle .
the peaks are located at @xmath249 , which is derived by solving the equation @xmath208 , or equivalently @xmath250 with @xmath210 .
therefore , the peak heights correspond to @xmath251 . as can be seen in fig .
[ fig : neufeld ] the agreement between numerical and analytic solutions is quite good for high optical depths .
for example , at an optical depth of @xmath252 when comparing the simulated profile ( red line ) with the approximate analytic solution ( thin black line ) there is a maximum difference of @xmath253 per cent .
the neufeld approximation of equation ( [ eq : neufeld ] ) fails to capture the correct spectra at lower optical depths because core scatterings also contribute to the spatial diffusion of ly@xmath2 photons .
therefore , the true measure of this test is the limiting behavior as @xmath237 tends to infinity .
the bottom panel of fig .
[ fig : neufeld ] demonstrates the face - on radial surface brightness profiles for the same slabs as calculated by equation ( [ eq : sb ] ) of section [ sub : surface_brightness_construction ] .
the scaling is such that the highest optical depth system is unity at the centre , i.e. @xmath243 . the dashed line is a rough analytic fit to guide the eye and is given in the figure .
the shaded regions represent one standard deviation of uncertainty on @xmath254 due to radial binning .
finally , this test has also been performed at other temperatures with similar results , cf .
appendix [ appendix : temperature_tests ] .
such simulations of ly@xmath2 scatterings through optically thick slabs help measure the effectiveness of the acceleration schemes , cf .
appendix [ appendix : x_crit ] .
to test the code for the case of a nonzero gas bulk velocity we compare our results to previous simulations with isotropic expansion of a uniform sphere
see @xcite .
in general , velocity gradients affect ly@xmath2 escape because photons are doppler shifted out of line centre thereby reducing the effective optical depth . in general , even velocity fields on the order of the thermal velocity in photoionized gas ( @xmath255 ) can change the emergent spectrum of ly@xmath2 photons .
following @xcite we consider the intensity of an isothermal ( @xmath233 k ) , homogeneous sphere of column density @xmath256 experiencing isotropic expansion .
the hubble relation gives the velocity @xmath257 of the gas at the position @xmath258 from the centre of a sphere with radius @xmath73 : @xmath259 where the hubble - like parameter @xmath260 sets the maximal velocity @xmath261 at the edge of the sphere , i.e. @xmath262 .
figure [ fig : velocity ] shows the result of this test which demonstrates excellent agreement with @xcite .
the static case ( @xmath263 ) also agrees with the analytical solution of @xcite for a static , optically thick spherical `` slab '' .
this test is also similar to that of @xcite who calculated the zero - temperature spectrum of a ly@xmath2 source embedded in a neutral , homogeneous igm undergoing hubble expansion .
whereas a static solution produces two distinct peaks for blue and red modes of escape , the blue photons are continuously redshifted back to the core and the red mode becomes the only means of escape .
this inevitably leads to free streaming on cosmological ( @xmath264mpc ) scales .
unfortunately , there is no analytical solution for a medium possessing both thermal and bulk motions .
figure [ fig : velocity ] illustrates the effect of increasing @xmath261 , which acts to suppress the blue peak until it disappears entirely by @xmath265 . at first
the red peak is pushed further from line centre , however , past a critical @xmath261 the peak approaches the centre again because the velocity gradient facilitates escape .
the bottom panel of fig .
[ fig : velocity ] shows the ( averaged ) line of sight radial surface brightness profile for each sphere .
the scaling is such that the static sphere is unity at the centre , i.e. @xmath243 . the dashed line is a reference for a flat profile to guide the eye .
only the largest velocity gradients substantially alter the profile so that the apparent image is much more concentrated toward the centre .
galaxy formation is a highly complex process ; however , studying the formation and radiative transport in the first , comparatively simple , systems provides an ideal laboratory for the physics involved @xcite .
we here employ two complementary methodologies to represent the structure and dynamics of a first galaxy for @xmath4 first , in section [ sec : toy ] we construct idealized analytic models of these galaxies , and second , in section [ sec : sim ] we extract a virialized halo from an _ ab initio _ cosmological simulation for post - processing .
these first galaxy models will then be the input for the radiative transfer calculations discussed in section [ sec : results ] .
our idealized models allow us to explore the basic physics by directly adjusting parameters in order - of - magnitude fashion .
our cutout from the cosmological simulation , on the other hand , provides us with one representative example of a realistic first galaxy .
we can thus gauge the validity of our exploratory toy models .
we now explore idealized models for the first galaxies at redshift @xmath266 in preparation for extractions from a cosmological simulation .
the number density of hydrogen nuclei @xmath267 is built up from the following assumptions : first , we require spherical symmetry so that @xmath268 .
second , we adopt a power law density profile within the galaxy , i.e. @xmath269 out to the edge of the galaxy @xmath270 defined as the point where @xmath267 equals the background igm , @xmath271 .
an isothermal law with @xmath272 provides a good description of a virialized system ( e.g. * ? ? ?
* ) . finally , we prefer a non - cuspy ` core ' in the centre of the galaxy , meaning the density profile flattens off to a constant density @xmath273 within a core radius of @xmath274 pc .
this is inspired by observations of low surface brightness galaxies that suggest a softening in the centre @xcite . in summary ,
the model is piecewise in the radial coordinate @xmath275 according to : @xmath276 here @xmath271 is the background atomic hydrogen number density derived from a @xmath277cdm model with cosmological parameters taken from the @xcite , and incorporating constraints from the _ wilkinson microwave anisotropy probe _ ( wmap ; * ? ? ?
* ) , the _ south pole telescope _
( spt ; * ? ? ?
* ) , and the _ atacama cosmology telescope _ ( act ; * ? ? ?
specifically , the hubble constant is taken to be @xmath278 km s@xmath279 mpc@xmath279 while the fractional energy contributions of baryons , matter , and dark energy are @xmath280 , @xmath281 , and @xmath282 , respectively . therefore , @xmath271 is @xmath283 , or @xmath284 where @xmath228 is the mass fraction of hydrogen and @xmath285 is the critical energy density at present .
furthermore , the edge of the galaxy is given by solving the equation @xmath286 , which yields a radius of @xmath287 .
the density parameter @xmath273 is then found by normalizing the overall mass of hydrogen in the galaxy @xmath288 to some value , e.g. @xmath289 for an atomic cooling halo .
the total mass is given by the integral @xmath290 equation ( [ eq : mhtot ] ) is a cubic polynomial in @xmath273 whose solution is not particularly insightful .
however , we may expand about large masses @xmath288 to consolidate the leading order terms .
these terms provide a relative accuracy in @xmath273 of better than 1 part per billion for masses larger than @xmath291 . for the parameters chosen above this
implies a central density of @xmath292 where we have introduced the dimensionless parameter : @xmath293 this term may also be found by dropping the @xmath294 term in equation ( [ eq : mhtot ] ) .
the @xmath295 parameter relates the ( leading order ) ratio of core and background densities via @xmath296 and the ratio of core and edge radii via @xmath297 . with this central density
the optical depth to ly@xmath2 scattering of the core region is at least @xmath298 , a value that would increase with a larger core radius , a lower temperature , or a more massive system .
we are primarily interested in the mass and density of hydrogen , therefore we decided to use @xmath288 in the comparison .
the total or virial mass of the galaxy is larger for two main reasons : @xmath299 the mass fraction of hydrogen to baryons is less than unity , i.e. @xmath228 . @xmath300 the contribution of baryonic mass is considerably less than the contribution of dark matter , which is usually more extreme in smaller galaxies .
in fact , if a substantial amount of gas is lost through ram pressure stripping or supernova blowout , for example , the baryonic mass may be significantly below the cosmological baryon fraction of @xmath301 per cent @xcite . at any rate we can reasonably relate the virial and total hydrogen masses by @xmath302
. the model above did not include an ionized region around a central star cluster .
this is important because ly@xmath2 photons can easily escape such regions .
furthermore , a strmgren analysis suggests that the ionized region may be on the order of @xmath303 for a reasonable set of parameters : @xmath304 where we have defined normalized values for ionizing photon rate @xmath305 and density @xmath306 .
the normalization used for @xmath307 is a plausible guess at the rate expected from the starbursts residing in the first galaxies ( see section [ sec : central_starburst ] for further discussion ) .
the expression was simplified by approximating the total case b recombination rate by @xmath308 cm@xmath5 s@xmath279 @xcite . in order to understand how the ionized region affects ly@xmath2
transfer we need to consider how the full three - dimensional ionization structure evolves in time .
the strmgren radius is a good approximation early on when the radiation is bottled up and the ionization rate balances the recombination rate , i.e. @xmath310 .
however , as the ionization front moves out , the hii regions blow pockets through which ly@xmath2 photons may escape .
therefore , we distinguish between _ early _ ionization scenarios when the hii region is ultra - compact as in equation ( [ eq : stromgren ] ) and _ late _ scenarios when the hii starburst regions have overtaken the entire halo .
these cases bracket the entire evolution .
the late stages of the ionization structure must be anisotropic because the cosmological filaments guide the ionized bubbles into a butterfly - shaped morphology around the centre . for simplicity we model this as a bipolar cavity .
specifically , in addition to the strmgren sphere , our _ late _ scenario also has a biconical region out to @xmath270 .
the size of the hii region is determined by the total number of ionizing photons , i.e. @xmath311 . if the photons are produced at a constant rate and the volume of the cone is described by the opening angle @xmath312 from the axis of symmetry then the time to evacuate the cavity is at least of order @xmath313 therefore , the _ late _
scenario corresponds to a time when the biconic region out to @xmath270 is fully ionized
see fig .
[ fig : idealized_density ] for an edge - on view of the column density through one such halo . for a `` late '' model at a source redshift of @xmath314 with @xmath315 .
the spatial scale , in physical kiloparsecs and arcseconds , are provided .
the edge - on view of the halo demonstrates the butterfly - shaped ionization structure of the models . for comparison with eq .
( [ eq : t_late ] ) the corresponding ionizing photon rate is @xmath316 , giving a late time lower limit of @xmath317 yr . ] the central starburst provides a feedback mechanism on the host galaxy . in order to motivate a spherically symmetric velocity law we consider the relative strength of the gravitational force @xmath318 to the radiation force @xmath319 . as a first estimate , if we attribute the radiation pressure to thompson scattering in the single - scattering limit , then because of the @xmath320 scaling of both forces , their ratio is of order @xmath321 where @xmath322 and @xmath323 are respectively the total luminosity and mass of the entire cluster .
the normalization is chosen to correspond to the mass to light ratio for pop iii stars , i.e. @xmath324 ( see section [ sec : central_starburst ] ) .
for simplicity equation ( [ eq : rad_grav ] ) assumes a point source . in reality , the ratio decreases radially from the centre of the gravitational potential as the dark matter halo mass dominates the gravitational force .
furthermore , equation ( [ eq : rad_grav ] ) is only applicable to ionized gas , whereas ly@xmath2 scattering occurs in neutral gas so other sources of pressure must also be present for our idealized law .
we consider a force multiplier in analogy to that of ( cak ; * ? ? ?
* ) originating from many resonant and optically thin lines . the multiplier can amplify the scattering efficiency by many orders of magnitude .
thus , the luminosities considered here ( @xmath325 ) may be capable of sustaining radiation - driven winds @xcite .
indeed , the ly@xmath2 line alone may be responsible for much of the opacity these winds require @xcite .
the cak theory uses mass conservation and momentum balance to arrive at a ` beta law ' with an exponent of @xmath326 for the radial velocity profile @xmath327 where @xmath328 is the size of the source .
the terminal velocity @xmath329 is typically larger than the escape velocity @xmath330 by a factor of a couple .
plausible values for our models are obtained by requiring that the mass loss rate @xmath331 be less than the maximally efficient mass loss for single scattering @xmath332 for which the radiation momentum is imparted entirely to the gas . in the halo s isothermal region ,
i.e. @xmath333 , the velocity approaches @xmath329 and @xmath334 so @xmath335 in principle , multiple scattering in the optically thick environment could boost @xmath329 by a factor of the square root of the optical depth , or at least an order of magnitude .
however , for simplicity our wind models all assume the velocity profile of equation ( [ eq : wind ] ) with @xmath336 pc , the approximate size of the central star cluster , and @xmath337 , a value comparable to the thermal velocity @xmath338 .
we note that our value of @xmath329 is also close to the escape velocity of the system , @xmath339 @xmath340^{1/3}$ ] @xmath341^{1/2}$ ] . for completeness
, the original cak force multiplier @xmath342 would need to be reevaluated for h and he lines of gas with primordial composition .
it is beyond the scope of this work to do so here but for reference the model of @xcite derives values of @xmath343 and @xmath344 .
the parameter @xmath345 is either the electron - scattering optical depth @xmath346 or the sobolev optical depth @xmath347 in static or expanding media , respectively .
the density and velocity profiles considered here are not self - consistent in the dynamical sense .
however , our simplified @xmath348-model , calibrated in the above fashion , should give us a rough window into the otherwise extremely complex physics of galactic winds ( e.g. * ? ? ?
this also avoids overconstructing models that are already quite idealized .
finally , the radiation pressure driven winds considered above may require some dust and metals to be present in the ism .
however , we have not included dust in these primordial environments .
further study on how dust affects ly@xmath2 transfer in anisotropic models should be carried out and is left for future work .
still , various mechanisms such as scattering in clumpy media have been proposed to explain why the ly@xmath2 escape fraction is so significant in environments where core depletion seems unavoidable .
in fact , @xcite suggest that dust attenuation is unlikely to affect ly@xmath2 escape at @xmath349 .
for our purposes , we expect dust to lower the bolometric line flux by at most a factor of a few .
for an example of coupling large scale cosmological simulations with dust and radiative transfer models designed to constrain igm ionization and the dust distribution in the ism of early galaxies see @xcite and @xcite .
the expression @xmath350 is based on comparing an expanding of the analytical solution for a static uniform sphere to the height of its peak see equation ( [ eq : sphere ] ) .
however , the constant of proportionality must be found by empirical tests .
the tests were run on many different values of @xmath202 , however , we only show two to demonstrate the validity across the parameter space .
the first is for @xmath351 , for which equation ( [ eq : xcrit ] ) gives @xmath352 , while the second is for @xmath353 , where @xmath354 . as can be seen from fig . [
fig : xcrit_test ] , both values produce excellent results . designed to compare different values of @xmath201 .
the converged solution ( yellow histogram ) is given by @xmath355 while a low value of @xmath356 provides excellent agreement and is shown in black .
a sample of values which are too high and affect the emergent spectrum are @xmath357 and are respectively given by orange , red , and purple dotted histograms .
bottom panel : same as the top panel except for @xmath353 .
a value of @xmath358 is sufficiently converged for our purposes .
an acceptable value of @xmath359 is given by a black line while non - converged values of @xmath360 are again given by the orange , red , and purple dotted histograms . here @xmath202 is large enough that the analytical solution of equ .
( [ eq : sphere ] ) is accurate , so it is included as the green dashed line in the background .
both tests used @xmath361 photon packets.,title="fig : " ] designed to compare different values of @xmath201 .
the converged solution ( yellow histogram ) is given by @xmath355 while a low value of @xmath356 provides excellent agreement and is shown in black .
a sample of values which are too high and affect the emergent spectrum are @xmath357 and are respectively given by orange , red , and purple dotted histograms .
bottom panel : same as the top panel except for @xmath353 .
a value of @xmath358 is sufficiently converged for our purposes .
an acceptable value of @xmath359 is given by a black line while non - converged values of @xmath360 are again given by the orange , red , and purple dotted histograms . here @xmath202 is large enough that the analytical solution of equ .
( [ eq : sphere ] ) is accurate , so it is included as the green dashed line in the background . both tests used @xmath361 photon packets.,title="fig : " ]
in order to test the grid structure of the cosmological simulation for edge effects and sensitivity to the extraction size we examine the emergent spectra for cubes with a centre to edge distance of @xmath125 pc , @xmath362 kpc , @xmath132 kpc , and @xmath363 kpc .
the largest size represents the radiative transfer through the entire @xmath364 mpc@xmath365 comoving volume .
figure [ fig : box_size ] demonstrates the convergence of the flux density @xmath366 toward that of the largest extraction , although there is still a significant variance across different lines of sight due to the inhomogeneous nature of the cosmic structure . in order to demonstrate the `` thermal effects '' on ly@xmath2 radiative transfer , e.g. via the doppler width @xmath372 and core - wing transition @xmath42 ,
we include two test suites of uniform slabs with varying temperature .
the first maintains a constant optical depth at line centre of @xmath373 while the second ensures a fixed column density of @xmath374 @xmath50 , which corresponds to @xmath373 at @xmath375 k. the angular averaged intensity for each model is plotted in fig .
[ fig : temperature_tests ]
. even though the optical depth or column density is fixed , the product @xmath376 depends on temperature and the emergent spectra is affected .
furthermore , we note that although the lorentz wing optical depth is always independent of temperature , @xmath377 and @xmath378 are proportional to @xmath379 and @xmath380 , respectively .
( [ eq : x_f ] ) and ( [ eq : recoil_parameter ] ) demonstrate that for recoil shifting to appreciably affect the ly@xmath2 profile it must be comparable to the typical redistribution at each scattering event .
for example , if @xmath381 then under thermally - dominated redistribution we typically have @xmath382 .
however , lorentzian - dominated redistribution may become a reasonable approximation at low temperatures . in this case , the scattering atom s parallel velocity component is roughly @xmath383 and the ratio to recoil is significantly reduced for core photons , i.e. @xmath384 . the total relative contribution over all scattering events induces a runaway reddening effect in cold , optically thick environments .
radiative transfer .
both cases represent the angular averaged intensity as a function of doppler velocity @xmath385 for uniform slabs of varying temperature . for direct comparison between the models the normalization is arbitrary but consistent in all cases .
the top panel maintains a constant optical depth at line centre of @xmath373 while the bottom panel ensures a fixed column density of @xmath374 @xmath50 , corresponding to @xmath373 at @xmath233 k. even though the optical depth or column density are fixed , the product @xmath376 depends on temperature and the emergent spectra is affected .
the number of photon packets used in these simulations is @xmath386 . ] | we present the @xmath0smic @xmath1yman-@xmath2 @xmath3ransfer code ( @xmath4 ) , a massively parallel monte - carlo radiative transfer code , to simulate lyman-@xmath2 ( ly@xmath2 ) resonant scattering through neutral hydrogen as a probe of the first galaxies .
we explore the interaction of centrally produced ly@xmath2 radiation with the host galactic environment .
ly@xmath2 photons emitted from the luminous starburst region escape with characteristic features in the line profile depending on the density distribution , ionization structure , and bulk velocity fields .
for example , anisotropic ionization exhibits a tall peak close to line centre with a skewed tail that drops off gradually .
idealized models of first galaxies explore the effect of mass , anisotropic hii regions , and radiation pressure driven winds on ly@xmath2 observables .
we employ mesh refinement to resolve critical structures .
we also post - process an _ ab initio _ cosmological simulation and examine images captured at various distances within the 1 mpc@xmath5 comoving volume .
finally , we discuss the emergent spectra and surface brightness profiles of these objects in the context of high-@xmath6 observations .
the first galaxies will likely be observed through the red damping wing of the ly@xmath2 line .
observations will be biased toward galaxies with an intrinsic red peak located far from line centre that reside in extensive hii super bubbles , which allows hubble flow to sufficiently redshift photons away from line centre and facilitate transmission through the intergalactic medium . even with gravitational lensing to boost the luminosity this preliminary work indicates that ly@xmath2 emission from stellar clusters within haloes of @xmath7 is generally too faint to be detected by the _ james webb space telescope _ ( _ jwst _ ) .
[ firstpage ] lyman-@xmath2 emission
radiative transfer resonant scattering line : profiles cosmology : theory galaxies : formation galaxies : high - redshift |
toys can be helpful in increasing students motivation in the classroom . in presentations for popularizing and communicating science to more general audiences
, they can also help increasing the appreciation and interest in the physical science , sometimes in such a way that everyone ( especially non - scientists ) will probably grasp some fundamental concepts .
however , they should be used with care : the physical description of some toys is not so easy @xcite , even in the framework of simplified models , and their usefulness is sometimes limited .
but , at least for motivation purposes , they are always valuable @xcite . in this paper
we describe the motion of a toy that , due to an internal source of energy , jumps while rotates .
the toy is a kangaroo but , as far as the physics description is concerned , being an object with the form of a kangaroo is just a detail , it could be something else ( even a living being ) . among the numerous possible objects suitable for illustration and demonstration purposes , a toy , performing on top of the instructor s table during the classroom , is definitely more likely to attract the student s attention .
the accurate description of all steps of the toy s motion is intricate but some simplified assumptions are possible and meaningful .
this allows us to transform the real complicated problem into a feasible one , which is useful , in this particular case , for establishing a correspondence between the descriptions of translations and rotations , on the one hand , and , on the other hand , to bridge mechanics and thermodynamics .
the jump of the kangaroo is funny , possibly even a bit mysterious , and our aim is to apply the pertinent physical laws to describe and understand the various phases of the motion . though the mechanical description of rotations and translations is the result of the very same newton s second law , students have a clear preference for translations . since our toy performs a movement that is a combination of a translation and a rotation , it can be useful for underlining the parallelism between the mechanical treatment of each type of motion _ per se_.
we shall assume constant forces and constant torques , therefore the real problem reduces to an almost trivial one .
nevertheless , there are some subtle points that are easy to emphasize with a simple example . in previous papers @xcite we analyzed , from the mechanical and thermodynamical point of view , quite a few systems , essentially either in translation or in rotation . here
we combine both types of motion and , again , we stress the thermodynamical aspects in each phase of the motion , their similarities and asymmetries .
the design and the manufacturing of the toy yields the kangaroo to perform a full rotation ( 360@xmath4 ) in the air while it jumps .
this is because of its mass , of its shape ( therefore of its moment of inertia ) , of the articulations between the legs and the body and also because of the power provided by the internal source of energy .
the manufacturer should define and include an internal energy source suitable for the toy to perform a full turn around the centre of mass while its centre of mass raises sufficiently high and drops down in the air . if a rotation angle of @xmath5 is not met , the toy does nt work .
our demonstration kangaroo is a plastic three euros toy , bought in a street vendor , whose source of internal energy is a spiral spring ( so , it is a low cost and very ecological item no batteries inside ) . the real toy performing a back somersault . from a movie , we extracted the pictures that represent the different phases of the motion : the preparation of the jump ( a)-(b ) ; the jump when the toy has no contact with the ground ( c)-(h ) ; and the final phase ( i)-(j ) when the toy stops after an initial contact with the ground .
, width=642 ] after providing the necessary energy to the spiral spring , we put the toy on top of a table and release it .
a back somersault by the toy immediately starts , as shown in figure [ fig : cang1 ] , and it comprises three phases : ( 1 ) the kangaroo , initially with the flexed legs , suddenly stretches them while rises its center of mass , increases its speed and starts rotating ( a)-(b ) ; ( 2 ) in this phase , ( c)-(h ) the toy has no contact with the ground , and rotates while its center of mass describes a parabola ; ( 3 ) this is the landing " phase that starts when the feet first come in contact with the ground , and it lasts until the toy completely stops ( i)-(j ) .
mechanics and thermodynamics are two different branches of physics with many interrelations .
however , interestingly enough , in most university physics curricula , as well as in the high school , thermodynamics and mechanics almost do not intercept .
this is not the case in everyday life where both are strongly connected : a most common example is the automobile @xcite , but there are many other examples @xcite .
we shall see that our funny kangaroo also helps in illustrating this kind of bridging .
there is no physics surprise " in the interpretation of the motion , we just have to apply , in combination , basic laws of mechanics and thermodynamics . with reasonable simplifying assumptions , that do not spoil the essence of the physical description , we are able to reduce the real problem to a classroom example that certainly students enjoy while they learn how basic physics principles work .
in section 2 we briefly introduce the general formalism that will be applied in the analysis of the motion of the toy .
the discussion of the dynamics is presented in section 3 and it is essentially known .
however , the subtle energetic issues related to the motion in phases 1 and 3 , described in section 4 , are probably less known or undervalued by instructors . in section 5
we present the conclusions .
let us briefly review the basic theoretical framework needed for the mechanical and thermodynamical analyzes .
a more detailed presentation can be found in @xcite . for a system of constant mass @xmath6 , newton s
second law can be expressed by p_cm= _
t_0^t f_ext t , [ eq1 ] where @xmath7 is the resultant external force acting upon the system and @xmath8 is the variation of the system center of mass linear momentum in the time interval @xmath9 .
equation ( [ eq1 ] ) already incorporates the third law of mechanics which implies that the resultant of the internal forces vanishes , @xmath10 .
the above equation , expressed in a vector form , can also be equivalently given in a scalar form by means of k_cm= _
ext_cm [ eq2 ] the so - called center of mass equation . on the left hand side , one has the variation of the center of mass kinetic energy ( @xmath11 ) and , on the right hand side
, the pseudo work @xcite performed by the resultant external force . for the pseudo - work , the resultant force and the center of mass displacement should be considered [ see the integral in ( [ eq2 ] ) ] , whereas for the _ work _ it is each force and its own displacement that matters . equation ( [ eq1 ] ) , which is an integral form of newton s equation , and the center of mass equation ( [ eq2 ] ) are physically equivalent , though they use different physical magnitudes they express the same fundamental law of mechanics , so they are general and do apply to all systems undergoing whatever process . when the mechanical systems perform rotations , other forms of newton s fundamental law
are better suited such as @xcite l = i= _ t_0^t _ ext t k_rot= 1 2 i ^2 = _ ext [ eq3 ] for the rotation of a system of constant moment of inertia @xmath12 around a principal axis of inertia containing the center of mass .
the system is acted upon by an external torque , of magnitude @xmath13 , whose direction is along the rotation axis .
these equations are simplified versions of the most general ones , and in ( [ eq3 ] ) one does not need to consider the vector character of the angular momentum , @xmath14 , of the angular velocity , @xmath15 , or of the torque , @xmath16 . the two equations ( [ eq3 ] ) for the rotation , together with ( [ eq1 ] ) and ( [ eq2 ] ) , for the translation , are the pertinent ones for the mechanical description of the motion presented in the next section .
a different physical law , also applicable to any system and to any process , is the first law of thermodynamics which , incidentally , also involves typical mechanical quantities .
that principle is a statement on energy conservation and it can be expressed by the equation k_cm+
w_ext + q . [ totale ] on the left hand side , in addition to the variation of the center of mass kinetic energy , one has the variation of the internal energy and , on the right hand side , the energy fluxes to / from the system are expressed . in other words , the left hand side of equation ( [ totale ] ) refers to the total energy variation of the system , whereas the right hand side expresses the energy that crosses the system s boundary i.e. the energy that enters or leaves the system through its boundary .
this energy transfer is the sum of two contributions : the _ external work _ i.e. the sum of the works performed by each external force , @xmath17 which is given by @xmath18 ; and @xmath19 , that is the heat flow to / from the surroundings .
if either @xmath20 or @xmath19 are positive , that means an energy transfer to the system , leading to an increase of the left hand side of ( [ totale ] ) ; if any of them is negative , that means an energy flow to the surroundings with a corresponding decrease of the total energy of the system .
it is worth noticing that , whereas the pseudo - work of the resultant external force leads to the variation of the centre of mass kinetic energy , as stated by equation ( [ eq2 ] ) , the real work , together with the heat , may change both that kinetic energy and/or the internal energy of the system , as stated by equation ( [ totale ] ) . by expressing the first law of thermodynamics in terms of equation ( [ totale ] )
one implicitly assumes that @xmath21 includes _ all _ energy variations that may contribute to the total internal energy variation of the system .
those include the variations of rotational [ such as @xmath22 as given by equation ( [ eq3 ] ) ] and translational kinetic energies with respect to the center of mass , in addition to the variations due to temperature changes , variations of internal chemical energy ( associated with chemical reactions ) , work performed by internal forces , etc .
of course , any process should also coply with the second law , besides the first law of thermodynamics .
we stress that equations ( [ eq2 ] ) and ( [ totale ] ) provide , in general , complementary information , since they correspond to two distinct fundamental laws of nature .
the study of the movement of the toy described in the next sections illustrates that complementarity .
diagrammatic description of the three phases of the motion . in the first phase ( 1 )
there is contact with the ground while the centre of mass moves horizontally and vertically ; in phase ( 2 ) the centre of mass describes a parabola while the toy performs a 360@xmath4 turn around ; the final phase ( 3 ) is the deceleration phase.,width=604 ] we have shown , in figure [ fig : cang1 ] , the real toy performing the back somersault that comprises three phases . in figure
[ fig : cang2 ] we illustrate pictorially these three phases of the kangaroo s motion .
the forces exerted by the ground in phases 1 and 3 are surely time dependent @xcite . as a simplifying assumption , we replace these variable forces by constant forces that produce , in each phase , exactly the same impulse as the real force @xcite . in phase 1 , for instance , the force @xmath23 exerted by the ground is replaced by the _
constant _ force @xmath24 such that @xmath25 , where @xmath26 is the duration of that phase , and similarly for phase 3 . at the end of the initial phase ,
the center of mass velocity is @xmath27 and , at the beginning of phase 3 , the center of mass velocity is @xmath28 .
regarding the rotation , the torque with respect to the center of mass , in phase 1 , is also a time dependent function .
the torque acts during the time interval @xmath26 , producing a certain variation of the angular momentum of the system ( note that now the moment of inertia of the toy , @xmath12 , also slightly varies because of the legs articulations ; however , this variation is rather small the legs are very light in comparison with the rest of the body and we can assertively adopt a constant @xmath12 ) . here
, our approximation consists in assuming a constant torque such that @xmath29 , where @xmath30 is a constant vector .
the torque provides a clockwise angular acceleration , being @xmath31 the angular velocity at the end of phase 1 . in figure
[ fig : cang3 ] we represent the _ constant _ forces and _ torques _ in phases 1 and 3 .
the constant torque @xmath32 in phase 3 leads to an counterclockwise angular acceleration that reduces to zero the initial angular velocity ( @xmath31 ) .
of course , in all phases , the weight , @xmath33 , is always acting , but its torque always vanishes with respect to the center of mass .
constant forces acting on the kangaroo during the initial ( 1 ) and the final ( 3 ) phases .
note that the torques produced by the contact forces are opposite ; in phase ( 1 ) the torque leads to the rotation of the toy , whereas the torque at phase ( 3 ) produces an angular deceleration and eventually the rotation ceases . , width=415 ] the fact that we are considering constant forces and torques considerably simplifies the integrals in the general expressions presented in section 2 .
these forces and torques are time dependent and they also depend on the position of the centre of mass ( the forces ) and of the rotation angle , @xmath34 ( the torques ) . the real forces and torques produce certain impulses and angular impulses ( right hand sides of equation ( [ eq1 ] ) and of the first equation ( [ eq3 ] ) ) .
the simplifying assumption consists in replacing the real forces and torques by constant vectors that produce exactly the same impulses . in this way we are simplifying , but not oversimplifying , the problem , making it handy , even practicable in the classroom context .
the constant forces and torques can be regarded as average ones , producing the same momentum ( linear and angular ) variations as the real ones .
of course we could use a probably more realistic force such as @xmath35 , where @xmath36 is a parameter .
this would complicate the approach because ( 1 ) would no longer be a trivial integration
. with a sophisticated force sensor ( the toy is very light ) it would be possible , in principle , to figure out @xmath37 and @xmath38 and adjust them with analytic functions . with known time dependent analytic functions the integral in ( [ eq1 ] ) would be straightforward but the integral in ( [ eq2 ] ) still would require the knowledge of @xmath39 ( the same for @xmath40 ) and the integral in ( [ eq3 ] ) would require the knowledge of @xmath41 ( the same for @xmath42 ) .
a more quantitative analysis of the motion is out of the scope of the present study and this is why we are using constant forces ( and torques ) to keep the problem within manageable limits .
phase 2 consists of a parabolic motion of the center of mass ( neglecting air resistance , of course ) combined with an uniform rotation . in figure
[ fig : cang4 ] we show the trajectory of the center of mass of the toy .
assuming constant forces , the trajectory in phase 1 is exactly a straight line .
in fact , for a constant force along @xmath43 and @xmath44 , and for a body that starts from rest , the accelerations @xmath45 and @xmath46 are constants .
therefore , @xmath47 and @xmath48 , hence , by eliminating @xmath49 in both equations , this leads to @xmath50 ( @xmath51 is a constant ) , whose graph is a straight line . in phase 3
the function @xmath52 is more complicated because there are initial velocities along @xmath43 and @xmath44 and @xmath53 is , in general , not a linear function .
however , since the center of mass only moves very little and for a very short time , the trajectory can be approximated by a straight line ( the difference between the straight line phase 3 in figure [ fig : cang4 ] and the actual trajectory is tiny , even indistinguishable within the precision of the drawing ) .
kangaroo s center of mass trajectory . in the first phase it is exactly a straight line , whereas in the second phase it is a parabola . in the third phase it is almost indistinct from a straight line.,width=340 ] in summary , regarding the center of mass motion ,
it is uniformly accelerated along both @xmath43 and @xmath44 in phase 1 , it is a projectile motion in phase 2 and , finally , it is a uniformly retarded motion in phase 3 along @xmath43 as well as along @xmath44 .
regarding the rotational motion , the angular acceleration is constant in phase 1 , in phase 2 the angular velocity is constant , and in phase 3 the angular acceleration is again constant , producing an uniformly retarded angular motion .
angular displacement around the axis that passes through the kangaroo s center of mass versus time .
the relative durations of the initial and final phases are exaggerated .
, width=283 ] concerning the angular displacement , the toy performs a complete turn around its centre of mass .
most of the time , the toy is in the air , hence the 360@xmath4 turn is almost executed in phase 2 . in phase 1
, the supposedly constant torque produces a quadratic dependence with time of the angular displacement : @xmath54 , where @xmath55 is the angular acceleration ; the function @xmath56 varies linearly with time during phase 2 : @xmath57 , where @xmath31 is the angular velocity at the end of phase 1 and during phase 2 ; and it again varies quadratically in phase 3 : @xmath58 , where @xmath59 is the magnitude of the angular acceleration in the third phase .
the time , @xmath49 , in the previous expressions starts at the beginning of each phase . in figure
[ fig : cang5 ] the function @xmath60 is shown , assuming that the turn corresponds to exactly @xmath61 ( in practice , this is only an approximate value ) and exaggerating , to make the figure more clear , the durations of phases 1 and 3 .
in addition to our previous assumptions , we consider that the magnitude of the velocities @xmath62 and @xmath63 is the same , i.e. the kangaroo touches the ground with its center of mass exactly at the level it occupies when the parabolic motion starts ( such an assumption is convenient to simplify the analysis but it could be relaxed ) . for the sake of a general discussion we let the displacement @xmath64 be different from @xmath65 ( the same for the initial and final @xmath44 displacements ) . in such a case ,
the magnitudes of the forces @xmath24 and @xmath40 are different and different are the time intervals during which they act .
indeed , we can stay more general , since such generality has no drastic consequence on the formalism or on its clarity .
so , we assume that the contact constant forces , @xmath24 and @xmath40 , do not necessarily have the same magnitude . in figure
[ fig : cang6 ] we represent the vertical component of the resultant force , @xmath66 , acting on the toy ( part [ a ] ) and the horizontal component of the same resultant force , @xmath67 , ( part [ b ] ) .
again the time intervals in the initial and final phases are exaggerated .
the important point to notice is that , for each graph , the algebraic sum of the represented areas , i.e. of the impulses along @xmath44 and @xmath43 should add up to zero : the toy stars from rest and it comes to rest at the end of the jump , therefore the total variation of the linear momentum is zero in both the @xmath43 and @xmath44 directions .
the same is true for the angular impulses : those in phases ( 1 ) and ( 3 ) cancel out and , in phase 2 , the angular impulse is zero
. resultant force acting on the toy : vertical component [ a ] , and horizontal component , [ b ] .
the represented areas are the impulses .
the impulses , in each graph , add up to zero .
the relative durations of the initial and final phases are exaggerated .
there is a picture similar to [ b ] for the angular impulse ( positive in phase 1 , negative in phase 3 and null during phase 2 ) .
, width=604 ] each of the contact forces has a vertical component ( normal reaction ) and a horizontal component ( static friction force ) .
we may decompose the contact forces according to @xmath68 and @xmath69 . to fix the notation we summarize in table [ tab1 ] the displacements , velocity variations , accelerations and other magnitudes for each phase of the motion .
0.5 cm .[tab1 ] magnitudes at each phase of the kangaroo s motion .
all quantities are positive , so the negative entries are explicitly written with a minus sign .
the torque is along the @xmath70 axis , perpendicular to the @xmath71 plane . [
cols="<,^,^,^ " , ] phases ( 1 ) and ( 3 ) are the most interesting ones , since phase ( 2 ) corresponds to the well known projectile motion . equation ( [ eq1 ] ) and the first equation ( [ eq3 ] ) , on the one hand , and equation ( [ eq2 ] ) and the second equation in ( [ eq3 ] ) , on the other hand , lead to \ { rl m v_0x & = f t_0 + m v_0y & = ( n -g ) t_0 + i _ 0 & = t_0 . and \
{ rl 1 2 m v_0 ^ 2 & = f x_0 + ( n - g)y_0 + & + 1 2 i _ 0 ^ 2 & = _ 0 , .
[ phase1a ] where @xmath6 is the mass of the toy and @xmath12 its moment of inertia ( that may slightly vary with time but we assume it is constant ) .
moreover , equation([totale ] ) , leads to m v_0 ^ 2 + 1 2 i _ 0 ^ 2 + u _ = - g y_0 , [ phase1b ] where @xmath72 is the part of the internal energy variation due to the spiral spring ( this is the potential elastic energy delivered by the spring ) .
the other part of the variation of the internal energy of the system is the kinetic rotational energy in ( [ phase1b ] ) .
the corresponding equations for phase 3 are \
{ rl m v_0x & = f t_0 + m v_0y & = ( n -g ) t_0 + i _ 0 & =
t_0 . and \ { rl 1 2 m v_0 ^ 2 & = f x_0 + ( n-g)y_0 + & + 1 2 m _ 0 ^ 2 & = _0
. [ phase3a ] and -1 2 m v_0 ^ 2 - 1 2 i _ 0 ^ 2 = g y_0 + q , [ phase3b ] where @xmath19 is the heat transfer to the surroundings during the process ( in this phase @xmath73 ) .
we stress that neither @xmath24 nor @xmath40 perform any work because ( ideally ) their application points do not move .
as already mentioned , phase 2 corresponds to a projectile motion combined with a uniform rotation . in this part
only a conservative force is acting , hence the sum of the translational kinetic energy , of the rotational kinetic energy and of the potential gravitational energy remains constant ( the rotational kinetic energy , @xmath74 , is , itself , constant ) . for this phase , the center of mass equation ( [ eq2 ] ) and the first law of thermodynamics ( [ totale ] ) provide exactly the same information , namely m ( v^2 - v^2_0 ) = - g ( y - y_0 ) ( note that now , in equation ( [ totale ] ) , @xmath75 and @xmath76 , the process is a purely mechanical one , because we have neglected the air friction ) .
regarding kinematical aspects , the maximum height reached by the center of mass , the horizontal distance traveled by the center of mass and the time of flight are given by h_max = y_0+v_0y^2 2 g , d= 2 v_0x v_0y g , t = 2 v_oy g .
[ rts ] for this phase , we may write @xmath77 .
if we take @xmath78 , for the time of flight given in ( [ rts ] ) , one finds a relation between the vertical component of the velocity at the end of phase 1 and the angular velocity at that very same moment : @xmath79 .
in this section we explicitly show that there are asymmetric energetic issues related to the motion in phases 1 and 3 , though the mechanical description of those phases is symmetric in the sense that in the first and in the third phases the impulses of the resultant forces are equal in magnitude with opposite directions .
let us go back to phase 1 .
the required energy for the kangaroo to perform the back somersault is obviously provided by the spiral spring and ultimately by the person that winds it . when the spring is released , we assume the elastic energy decrease to be given by @xmath80 , with @xmath81 the initial stored energy and @xmath82 the elastic constant characterizing the spring ( the angle @xmath83 , with initial value @xmath84 , is a decreasing function of time ) . as this energy decreases ,
the articulated toy starts the jump . combining equations ( [ phase1a ] ) and ( [ phase1b ] )
, we may express the internal energy variation by u _ & = & - ( 1 2 m v_0 ^ 2 + 1 2 i _ 0 ^ 2 + gy_0 ) + & = & - < 0 .
the first line explicitly shows that the internal ( elastic ) energy is converted into another form of mechanical energy : translational and rotational kinetic energies plus potential gravitational energy .
this phase is reversible it evolves with no variation of the entropy of the universe @xcite . in the second line , the kinetic energy terms are expressed by the pseudo works related to the contact force and to its torque . regarding phase 3 , equations ( [ phase3a ] ) and ( [ phase3b ] )
now lead to q & = & - ( 1 2 m v_0 ^ 2 + 1 2 i _ 0 ^ 2 + gy_0 ) + & = & - < 0 . this is the heat released in phase 3 of the motion .
this energy is simply lost in the sense that it flows from the system to the surroundings , where it is dissipated , and the entropy of the universe increases @xcite .
the relation between the original internal energy variation and this heat can be found combining the previous equations , or simply by applying directly equation ( [ totale ] ) to the whole process : on the left hand side of that equation the variation of the center of mass kinetic energy is zero and the variation of the internal energy is solely @xmath72 ( to simplify the discussion we are assuming no temperature variations in the toy ) ; on the right hand side of equation ( [ totale ] ) , the work is only due to the weight ( again we stress that neither the normal forces , nor the static friction forces perform any work ) ; finally , there is the heat flow to the surroundings .
altogether , equation ( [ totale ] ) yields for the overall process q = u _ + g ( y_0-y_0 ) . [ caloor ] for @xmath85
the energy initially stored in the spring totally dissipates in the surroundings , which behaves as a heat reservoir at temperature @xmath86 .
if the final position of the center of mass lies at an higher level than the initial one , part of the initial stored energy is used to raise the center of mass of the kangaroo , as expressed in equation ( [ caloor ] ) . in that case
, the magnitude of the heat is smaller than the magnitude of the elastic potential energy initially stored in the toy .
the overall process is clearly irreversible and , for @xmath85 , the entropy variation of the universe is @xmath87 ( the heat should now be considered from the point of view of the reservoir , and this is the reason for the minus sign ) . when @xmath88 and @xmath85 , even though the mechanical analysis becomes quite symmetric for the initial and final phases , there is a clear thermodynamical asymmetry : in phase 1 , there is mechanical energy production at the expend of the internal energy of the system ; in phase 3 , the mechanical energy is lost " in the sense that it dissipates as heat in the surroundings and it can not be recovered in an useful way .
we may say that , at the end of phase 1 , the system still keeps the energy in an organized form " as mechanical energy ( hence , no entropy increase ) , whereas in the final phase there is no mechanism to recover the decrease of mechanical energy : in particular that energy can not be stored back in the spiral spring .
if that could be possible , the kangaroo would perform back somersaults continuously ... we all know that this is not the case : after one jump one has to rewind the spiral spring for the next jump . in a recent publication
we discuss the same type of asymmetries in the context of human movements , like jumping and walking @xcite . as in the example discussed in this paper , in one part of the process there is a direct conversion of internal energy into mechanical energy but , in the other part , the mechanical energy dissipates as heat in the surroundings .
the motion of the kangaroo studied in this paper can be used in the classroom to explicitly show the correspondence between translations and rotations .
the example also serves to demonstrate that , for certain mechanical systems performing movements in which one part is symmetric with respect to the other , the thermodynamical behavior of both parts may be different . more precisely , in one part
there may be production of mechanical energy , due to an internal energy source , a process that does not increase the entropy , whereas in the other part of the motion there occurs dissipation of mechanical energy , which is transferred to the surroundings as heat , and there is an unescapable entropy increase .
we should emphasize that the first phase of the problem studied in this paper is somehow similar to the springboard diver discussed in ref .
@xcite the main difference is that the internal ( elastic ) energy variation should then be replaced by the variation of the gibbs free energy in the person s muscles @xcite . there are of course many other examples but the point we would like to emphasize here is the fact that energetic aspects are usually a bit underrated in textbooks @xcite .
the discussion presented in this paper comes in the sequence of a series of papers @xcite devoted to the link between mechanics and thermodynamics , a tendency that , fortunately , is already present in modern textbooks @xcite .
our aim is to contribute to fill the gap one encounters in most treatments of classical mechanics , and we do hope that our discussion , motivated by a kangaroo " , is relevant for physics education .
j. gmez , c. fiolhais , m. fiolhais _ the cartesian diver and the fold catastrophe _ , am . j. phys . * 70 * 710 - 714 ( 2002 ) ; j. gmez , r. valiente , c. fiolhais , m. fiolhais , _ experiments with the drinking bird _ , am . j. phys . * 71 * 1257 - 1263 ( 2003 ) j. gmez , c. fiolhais , m. fiolhais , _ toys in physics lectures and demonstrations a brief review _ , phys
. edu . * 44 * 53 - 64 ( 2009 ) ; j. gmez , c. fiolhais , m. fiolhais , _ juguetes en clases y demostraciones de fsica _ , revista iberoamericana de fsica * 6 * 45 - 56 ( 2010 ) j. gmez , m. fiolhais , _ from mechanics to thermodynamics : analysis of selected examples _ , eur
. j. phys . * 34 * 345 - 357 ( 2013 ) j. gmez , m. fiolhais , _ thermodynamics in rotating systems analysis of selected examples _ ,
* 35 * ( 2014 ) 015013 ( 14pp ) j. gmez , m. fiolhais , _ forces on wheels and fuel consumption in cars _
j. phys . * 34 * 1005 - 1013 ( 2013 ) j. gmez , m. fiolhais , _ the physics of a walking robot _ , phys .
educ . * 48 * 455 - 458 ( 2013 ) c m penchina , _ pseudowork - energy principle _ , am . j. phys .
* 46 * 295 - 296 ( 1978 ) b a sherwood , _ pseudowork and real work _ , am .
* 51 * 597 - 602 ( 1983 ) a j mallinckrodt , h. s. leff , _ all about work _ , am .
* 60 * 356 - 365 ( 1992 ) c e mungan , _ a primer on work - energy relashionships for introductory physics _ , phys . teach .
* 43 * 10 - 16 ( 2005 ) j w jewett , jr .
_ energy and the confused student v : the energy / momentum approach to problems involving rotating and deformable systems _
46 * 269 - 274 ( 2008 ) m alonso , e j finn _ university physics vol .
i , mechanics _
addison wesley publishing company ( 1967 ) ; d kleppner , r kolenkow , _ an introduction to mechanics _ , mc graw - hill publishing company ( 1973 ) p a tipler , g mosca , _
physics for scientists and engineers _ 5th ed .
w h freeman and co. new york 2004 , sec .
9 - 4 o. a. haugland , _ walking through the impulse - momentum theorem _ , phys .
teach . * 51 * 78 - 79 ( 2013 ) h. s. leff , _ removing the mystery of entropy and thermodynamics - part iv _ , phys .
teach . * 50 * 215 - 217 ( 2012 ) h. s. leff , _ removing the mystery of entropy and thermodynamics - part iii _ , phys .
50 * 170 - 172 ( 2012 ) j. gmez , m. fiolhais , _ thermodynamics asymmetries in whirling , jumping and walking _ , eur
. j. phys .
* 35 * ( 2014 ) in print j walker , _ halliday and resnick principles of physics
. international student version _ john wiley and sons , hoboken 2011 .
p. atkins , _ the laws of thermodynamics . a very short introduction _
, oxford u. press , oxford uk 2010 , pp . 65 - 72 r w chabay and b a sherwood , _ matter and interactions _
wiley and sons , hoboken 2011 . | we describe the physics of an articulated toy with an internal source of energy provided by a spiral spring .
the toy is a funny low cost kangaroo which jumps and rotates .
the study consists of a mechanical and a thermodynamical analysis which makes use of the newton and center of mass equations , the rotational equations and the first law of thermodynamics .
this amazing toy provides a nice demonstrative example how new physics insights can be brought about when links with thermodynamics are established in the study of mechanical systems . 0.6 cm plus 1pt minus 1pt = 1500 * the physics of articulated toys a + jumping and rotating kangaroo * 0.4 cm j. gmez@xmath0 , m. fiolhais@xmath1 0.1 cm _
@xmath2 departamento de fsica aplicada _
+ _ universidad de cantabria _ + _ e-39005 santander , spain _ + 0.1 cm _
@xmath3 departamento de fsica and centro de fsica computacional
_ + _ universidade de coimbra _ + _ p-3004 - 516 coimbra , portugal _ |
network coding technique has significantly improved the performance of communication networks , and has been studied extensively in the past two decades .
index coding problem ( icp ) can be considered as a special case of network coding problem @xcite .
icp has emerged as an important topic of recent research due to its applications in many of the practically relevant problems including that in satellite networks , topological interference management , wireless caching and cache enabled cloud radio access networks for 5 g cellular systems .
the noiseless index coding problem with side information was first studied in @xcite as an informed - source coding - on - demand ( iscod ) problem , in which a central server ( sender ) wants to broadcast data blocks to a set of clients ( receivers ) which already has a proper subset of the data blocks .
the problem is to minimize the data that must be broadcast , so that each receiver can derive its required data blocks .
consider the case of a sender with @xmath2 messages denoted by the set @xmath3 , @xmath4 , @xmath5 is a field with @xmath6 elements , which it broadcasts as coded messages , to a set of @xmath7 receivers , @xmath8 .
each receiver @xmath9 wants a subset @xmath10 of the messages , knows a priori a proper subset @xmath11 of the messages , where @xmath12 , and is identified by the pair @xmath13 .
the noiseless index coding problem is to find the smallest number of transmissions required and is specified by @xmath14 .
the set @xmath11 is referred to as the side information available to the receiver @xmath15 . an index code ( ic ) for a given icp @xmath14 is defined by an encoding function , @xmath16 and a set of @xmath7 decoding functions @xmath17 @xmath18 corresponding to the @xmath7 receivers , such that , @xmath19 in this paper , we consider icp over binary field ( @xmath20 ) .
the integer @xmath21 , as defined above is called the length of the index code . for noiseless broadcast channels , an index code of minimum length
is called an optimal index code @xcite , @xcite . even though this is interesting
theoretically , index coding over noisy channels is more practical .
noisy index coding over a binary symmetric channel was considered in @xcite , @xcite .
binary transmission of index coded bits were assumed .
for this set up the problem of identifying the number of optimal index codes possible for a given icp is important and that was studied in @xcite , @xcite . a special case of icp over gaussian broadcast channel , based on multidimensional qam constellation with @xmath22 points , where every receiver demands all messages ( which it does not have ) from the source , was considered in @xcite .
the case of noisy index coding over awgn broadcast channel , along with minimum euclidean distance decoding , was studied in @xcite , where the receivers demand a subset of messages as defined in @xcite . an algorithm to map the broadcast vectors to psk signal points so that the receiver with maximum side information gets maximum psk side information coding gain , was also proposed .
the algorithm assumes that an index code is given and is applicable only for one specific order of priority ( in the non increasing order of amount of side information ) among the receivers .
minimum euclidean distance of the effective broadcast signal set seen by a receiver , was considered as the basic parameter which decides the message error probability of the receiver and the proposed algorithm tries to maximize the minimum euclidean distance . in this paper
, we discuss the maximum likelihood ( ml ) decoder for index coded psk modulation .
further , we study the case in which the length of the index code is specified for the icp but not necessarily the index code .
the receivers can have any a priori defined arbitrary order of priority among themselves .
for a chosen priority order , we consider all possible index codes , to obtain the mappings to appropriate psk constellation which will result in the best message error performance in terms of psk index coding gain ( psk - icg , defined in section [ isd_section ] ) of the receivers , respecting the defined order of priority .
consider a noisy index coding problem with @xmath2 messages , over @xmath23 which uses an awgn broadcast channel for transmission .
for the icp @xmath14 , consider index codes of length @xmath1 , @xmath24 , which will generate @xmath0 broadcast vectors ( elements of @xmath25 ) .
the broadcast vectors are mapped to @xmath0-psk signal points , so that @xmath0-psk modulation can be used , to minimize the bandwidth requirement .
note that , transmitting one @xmath0-psk signal point instead of @xmath1 binary bits ( as in noiseless index coding ) , results in @xmath26 fold saving in bandwidth .
our contributions are summarized below : * we derive a decision rule for maximum likelihood decoding which gives the best message error performance , for any receiver @xmath15 , for a given index code and mapping . * we show that , at very high snr , the message error performance of the receiver employing ml decoder , depends on the minimum inter - set distance ( defined in section [ isd_section ] ) .
the mapping which maximized the minimum inter - set distance is optimal for the best message error performance at high snr .
* for the icp @xmath14 , when the receivers are arranged in the decreasing order of priority , we propose an algorithm to find ( index code , mapping ) pairs , each of which gives the best message error performance for the receivers , for the given order of priority .
using any one of the above ( index code , mapping ) pairs , the highest priority receiver achieves the maximum possible gain ( psk - icg ) that it can get using any ic and any mapping for @xmath0-psk constellation , at very high snr .
given that the highest priority receiver achieves its best performance , the next highest priority receiver achieves its maximum gain possible and so on in the specified order of priority .
let @xmath27 \triangleq \{1,2, ... ,n\}$ ] .
for a vector @xmath28 and a subset @xmath29 of @xmath27 $ ] ( for any integer @xmath30 ) , where @xmath31 , @xmath32 denote the vector @xmath33 .
we consider the noisy index coding problem over @xmath23 with a single sender having a set of messages @xmath3 , @xmath34 , and a set of @xmath7 receivers , @xmath8 , where each receiver @xmath15 is identified by @xmath13 , the want set and the known set .
let , @xmath35 be the set of indices corresponding to the known set .
it is sufficient to consider the case where each receiver demands only one message .
if there is a receiver which demands more than one message , it can be considered as @xmath36 equivalent receivers each demanding one message and having the same side information .
each @xmath15 , @xmath37 $ ] wants the message @xmath38 , where @xmath39 \rightarrow [ n]$ ] and @xmath40 , @xmath41 $ ] .
for the given icp , we consider scalar linear index codes of length @xmath1 ( not necessarily the minimum or optimum length ) , such that the set of all broadcast vectors gives @xmath42 let @xmath43 be an @xmath44 encoding matrix for one such index code , @xmath45 .
let @xmath46 and @xmath47 denote the message vector and the broadcast vector respectively , where @xmath48 [ exmp : eg1 ] consider the following icp with @xmath49 and @xmath50 .
the side information available with the receivers is as follows : @xmath51 . for this icp we can choose a scalar linear index code of length @xmath52 , as given by the following encoding matrix @xmath43 .
+ the index coded bits are given by + @xmath54 as @xmath55 . instead of using @xmath1 bpsk transmissions , the @xmath1 index coded bits of @xmath56
are sent as a signal point from a @xmath0-psk signal set , over an awgn channel , to save bandwidth @xcite . in this paper
we consider index coded @xmath0-psk modulation for a chosen @xmath1 and so when we refer to index codes of length @xmath1 , we consider only those index codes for which the set of all broadcast vectors is @xmath57 .
let the chosen @xmath0-psk signal set be denoted as @xmath58 .
assume that for the index code @xmath45 a mapping scheme specifies the mapping of @xmath57 to the signal set @xmath59 .
all receivers are assumed to know the encoding matrix , @xmath43 for the index code @xmath45 .
let @xmath60 be a realization of @xmath61 .
as each receiver @xmath15 knows some messages ( from its side information ) , @xmath15 needs to consider only a subset of @xmath57 for decoding and this subset is called the _ effective broadcast vector set_. for a chosen index code based on the encoding matrix @xmath43 , the _ effective broadcast vector set
_ seen by @xmath15 for @xmath62 is defined by , @xmath63 \setminus \mathcal{i}_{i}\}. \end{aligned}\ ] ] the corresponding set of signal points in @xmath0-psk constellation is referred to as the _ effective broadcast signal set _ seen by @xmath15 for @xmath62 and is denoted by @xmath64 . for a chosen index code , all effective broadcast signal sets and effective broadcast vector sets seen by @xmath15 are of the same size ( @xmath65 where @xmath66 ) .
half the number of broadcast vectors in an effective broadcast vector set corresponds to @xmath67 and the remaining half corresponds to @xmath68 .
so , we can partition an effective broadcast vector set into two subsets as defined below .
the _ 0-effective broadcast vector set
_ seen by @xmath15 for @xmath62 is defined by , @xmath69 \setminus ( \mathcal{i}_{i } \cup \{f(i)\})\}. \end{aligned}\ ] ] the corresponding set of signal points in @xmath0-psk constellation is referred to as the _ 0-effective broadcast signal set
_ seen by @xmath15 for @xmath62 and is denoted as @xmath70 .
the _ 1-effective broadcast vector set _ seen by @xmath15 for @xmath62 is defined by , @xmath71 \setminus ( \mathcal{i}_{i } \cup \{f(i)\})\}. \end{aligned}\ ] ] the corresponding set of signal points in @xmath0-psk constellation is referred to as the _ 1-effective broadcast signal set
_ seen by @xmath15 for @xmath62 and is denoted as @xmath72 .
the effective broadcast vector sets , 0-effective broadcast vector sets and 1-effective broadcast vector sets seen by @xmath73 for the ic in example [ exmp : eg1 ] is given in table [ table : tbl1 ] .
it is clear that , two different realizations of @xmath61 may have the same effective broadcast vector set .
however , 1-effective broadcast vector set for a particular realization of @xmath61 may become the 0-effective broadcast vector set of another realization of @xmath61 and vice versa .
but the way in which the effective broadcast vector set gets partitioned will be the same .
for example consider the case of @xmath74 and @xmath75 in table [ table : tbl1 ] .
.effective broadcast vector sets and its partitions ( seen by @xmath73 ) for the ic in example [ exmp : eg1 ] .
[ cols="<,<,<,<",options="header " , ] [ table : tbl2 ] there are @xmath76 optimal mappings for @xmath77 .
the set @xmath78 has @xmath76 ( index code , mapping ) pairs which are optimal for @xmath77 , with the index code being the same for all the pairs .
consider @xmath73 .
after all pairs in @xmath78 are considered , the maximum value possible for @xmath79 and there are 128 pairs which are optimal for @xmath73 .
now consider @xmath80 .
there are 24 pairs which are optimal with @xmath81 . for @xmath82
there are 16 optimal pairs with minimum inter - set distance @xmath83 . for @xmath84 and @xmath85
all these pairs give the same minimum inter - set distance .
these 16 pairs are the optimal mappings for the ic considered .
one of these mappings is given in fig .
[ figure : fig2](a ) . .
we have considered the icp given in example [ exmp : eg1 ] and used algorithm [ alg : a1 ] to obtain all optimal ( index code , mapping ) pairs .
one such pair , @xmath86 has the index code as given in example [ exmp : eg1 ] and mapping as given in fig .
[ figure : fig1](b ) .
we compared this optimal mapping with another mapping @xmath87 ( shown in fig .
[ figure : fig1](c ) ) which is not optimal for the same index code , @xmath88 .
the pair @xmath89 , the output set obtained from the execution of the algorithm .
we obtained the message error probability of the receivers for the two different mappings , by simulation .
the first mapping , ( @xmath90 ) used algorithm [ alg : a1 ] and the second mapping ( @xmath87 ) used an algorithm based on maximizing the minimum euclidean distances @xcite .
simulation results are given in fig .
[ figure : sim1 ] . ) . ]
the performance of receivers @xmath77 , @xmath73 and @xmath80 is significantly better with @xmath90 than with @xmath87 at high snr .
the minimum inter - set distances are more for @xmath90 ( fig.[figure : fig1](b ) ) than for @xmath87 ( fig.[figure : fig1](c ) ) . for receivers
@xmath82 and @xmath84 , the minimum inter - set distances are same for both the mappings .
we have carried out simulation based studies to compare the performance of the receivers for the icp and the ic given in example [ exmp : eg2 ] for two different mappings as given in fig .
[ figure : fig2 ] .
the mapping ( @xmath90 ) given in fig .
[ figure : fig2](a ) used algorithm [ alg : a1 ] and the mapping ( @xmath87 ) given in fig [ figure : fig2](b ) used the algorithm based on maximizing the minimum euclidean distances @xcite .
simulation results are given in fig .
[ figure : sim2 ] . ) . ] for @xmath77 , @xmath80 , @xmath82 , @xmath84 and @xmath85 , the minimum inter - set distances and hence the performances are the same for both the mappings . but
the performance of receiver @xmath73 is significantly better with @xmath90 than with @xmath87 at high snr .
the simulation results indicate the effectiveness of the algorithm based on minimum inter - set distances ( algorithm [ alg : a1 ] ) for mapping the broadcast vectors to psk signal points . it should be noted that algorithm [ alg : a1 ] does not guarantee that all the receivers will perform better or as good as that with any other algorithm .
it is possible that , a mapping based on some algorithm ( say , algorithm 2 ) gives a better performance to a receiver @xmath91 than that with algorithm [ alg : a1 ] .
but then there will be a receiver @xmath15 which performs better with algorithm [ alg : a1 ] than with algorithm 2 , where @xmath15 is a higher priority receiver than @xmath91 .
in other words , algorithm [ alg : a1 ] attempts to maximize the gain achieved by the receivers by considering the receivers in the given order of priority .
this is further illustrated in example [ exmp : eg3 ] .
[ exmp : eg3 ] consider the following icp with @xmath49 and @xmath50 .
the side information available with the receivers is as follows : @xmath92 . for this icp a scalar linear index code of length @xmath93 ( not optimal ) ,
is specified as @xmath94 .
assume that the decreasing order of priority is given as @xmath95 .
using algorithm [ alg : a1 ] , optimal mappings for the specified ic is obtained , of which one mapping ( @xmath90 ) is given in fig .
[ figure : fig3](a ) .
another mapping @xmath87 , is found by using the algorithm based on maximizing the minimum euclidean distances @xcite and is given in fig .
[ figure : fig3](b ) .
simulation results comparing the performance of the receivers for these two mappings are given in fig .
[ figure : sim3 ] . .
] ) . ]
it is clear from fig .
[ figure : sim3 ] that , @xmath73 performs better with @xmath90 than with @xmath87 . but
@xmath80 , which is of lower priority than @xmath73 , has better performance with @xmath87 .
in this paper we have considered index coded psk modulation and have derived a decision rule for the ml decoder which minimizes the message error probability of the receivers , for a given icp .
we have introduced the concept of inter - set distances and illustrated its importance in noisy index coding problems .
it was also shown that at high snr the dominant factor which decides the message error is the minimum inter - set distance and so an optimal mapping must maximize the minimum inter - set distance .
subsequently , we have considered the problem of finding optimal ( index code , mapping ) pairs across all possible mappings for all possible index codes of length @xmath1 , for a chosen @xmath0-psk modulation .
this problem was not addressed so far in literature .
the algorithm which is proposed for a given icp , can find ( index code , mapping ) pairs , each of which gives the best psk - icg for the receivers , for any given order of priority . finding all index codes of a chosen length ( greater than or equal to the optimal length ) for a given icp
is in general np hard . if it is too complex to find all the index codes , the algorithm can be executed by considering a chosen set of index codes .
but the complexity of the proposed algorithm increases exponentially with the length of the index code .
this work was supported partly by the science and engineering research board ( serb ) of department of science and technology ( dst ) , government of india , through j.c .
bose national fellowship to b. sundar rajan .
y. birk and t. kol , coding on demand by an informed source ( iscod ) for efficient broadcast of different supplemental data to caching clients , " _
ieee trans.inf.theory_ , 52(6 ) , june 2006 , pp .
2825 - 2830 .
anoop thomas , kavitha radhakumar , attada chandramouli and b. sundar rajan , optimal index coding with min - max probability of error over fading channels , " in _ proc .
ieee pimrc . , _
hong kong , 2015 , pp .
889 - 894 anoop thomas , kavitha radhakumar , attada chandramouli and b. sundar rajan , single uniprior index coding with min - max probability of error over fading channels , " accepted for publication in ieee transactions on vehicular technology .
kavitha radhakumar and b. sundar rajan , on the number of optimal index codes , " proceedings of ieee international symposium on information theory , ( isit 2015 ) , hong kong , 14 - 19 june 2015 , pp .
1044 - 1048 .
kavitha radhakumar , niranjana ambadi and b. sundar rajan , on the number of optimal linear index codes for unicast index coding problems , " in _ proc .
47th annu .
ieee wireless communications and networking conference _ , doha , qatar , april 2016 , pp .
1897 - 1903 . | index coded psk modulation over an awgn broadcast channel , for a given index coding problem ( icp ) is studied . for a chosen index code and an arbitrary mapping ( of broadcast vectors to psk signal points ) , we have derived a decision rule for the maximum likelihood ( ml ) decoder .
the message error performance of a receiver at high snr is characterized by a parameter called _ * psk index coding gain ( psk - icg)*_. the psk - icg of a receiver is determined by a metric called _ * minimum inter - set distance*_. for a given icp with an order of priority among the receivers , and a chosen @xmath0-psk constellation we propose an algorithm to find _ * ( index code , mapping ) * _ pairs , each of which gives the best performance in terms of psk - icg of the receivers .
no other pair of index code ( of length @xmath1 with @xmath0 broadcast vectors ) and mapping can give a better psk - icg for the highest priority receiver .
also , given that the highest priority receiver achieves its best performance , the next highest priority receiver achieves its maximum gain possible and so on in the specified order of priority . |
the stage leading up to dynamic collapse of a magnetically subcritical cloud core to a protostar or a group of protostars is believed to be largely quasi - static , if the responsible process is ambipolar diffusion ( e.g. , mestel & spitzer 1956 , @xcite , @xcite , @xcite , @xcite ) . to describe the transition between quasi - static evolution by ambipolar diffusion and dynamical evolution by gravitational collapse
, @xcite introduced the idea of a pivotal state , with the scale - free , magnetostatic , density distribution approaching @xmath8 for an isothermal equation of state ( eos ) when the mass - to - flux ratio has a spatially constant value , a condition that @xcite and @xcite termed `` isopedic '' .
numerical simulations of the contraction of magnetized clouds justify the assumption of a nearly constant mass - to - flux ratio in the pivotal core .. outside the starred point , the mass - to - flux value exhibits greater variation , but this occurs only because @xcite impose starting values for the mass - to - flux in the envelope that are @xmath9 times the critical value ( see also figs . 4a and 8b in @xcite ) . such small ratios for the bulk of the mass of a molecular cloud are probably ruled out by the zeeman oh measurements summarized by crutcher ( 1998 ) . ] the small dense cores of molecular clouds that give rise to low - mass star formation are effectively isothermal ( @xcite ; @xcite )
. the situation may be different for larger regions that yield high - mass or clustered star formation .
it has often been suggested that the eos relating the gas pressure @xmath10 to the mass density @xmath11 of interstellar clouds can be represented by a polytropic relation @xmath12 with negative index @xmath0 .
@xcite pointed out the utility of this idealization within the context of the classic two - phase model of the diffuse interstellar medium [ @xcite ; @xcite ; @xcite ] , while @xcite published extensive tables analyzing the stability of non - magnetized , self - gravitating spheres of such gases .
@xcite examined the linewidth - size and density - size relations of molecular clouds , first found by @xcite and subsequently studied by many authors [ e.g. , @xcite , @xcite , @xcite , @xcite , @xcite ] .
maloney pronounced the results consistent with the properties of negative index polytropes . for a polytropic eos , the sound speed @xmath13 increases with decreasing density if @xmath14 .
the latter behaviour may be compared with the empirical linewidth - density relation for molecular clouds , @xmath15 , with @xmath16 for low - mass cores ( @xcite ) and @xmath17 for high - mass cores ( @xcite ) , implying that @xmath0 lies between @xmath18 and @xmath19 , or that a static @xmath20 lies between @xmath21 and 0.7 .
the case @xmath22 is of particular relevance for the equilibrium properties of molecular clouds .
@xcite found that the pressure of alfvn waves propagating in a stratified medium , @xmath23 , a consequence of conservation of the wave energy flux @xmath24 .
this result was later derived more rigorously by @xcite in the wkb approximation for mhd waves propagating in mildly inhomogeneous media , and , more recently by @xcite and @xcite in a specific astrophysical context . in numerical simulations of the same problem
, @xcite found indication of a much shallower relation ( @xmath25 ) for a self - gravitating medium supported by nonlinear alfvn waves . on the other hand , for the adiabatic contraction of a cloud supported by linear alfvn waves
, @xcite found a dynamic @xmath26 larger than 1 .
@xcite confirmed a similar behaviour in numerical simulations of the gravitational collapse of clouds with an initial field of hydrodynamic rather than hydromagnetic turbulence . in the limit of @xmath27 ( or @xmath28 ) , the eos becomes `` logatropic , '' @xmath29 , a form first used by @xcite to mimic the nonthermal support in molecular clouds associated with the observed supersonic linewidths .
the sound speed associated with the nonthermal contribution , @xmath30 becomes important at the low densities characteristic of molecular cloud envelopes ( as contrasted with the cloud cores ) since the thermal contribution is independent of density if the temperature @xmath31 remains constant .
this nonthermal contribution decreases with increasing density and will become subsonic at high densities as recently observed in the central regions of dense cores ( @xcite ) . @xcite and @xcite
have modeled the equilibrium and collapse properties of unmagnetized , self - gravitating , spheres with a pure logatropic eos and claim to find good agreement with observations .
adams , lizano , & shu ( in 1987 ) independently obtained the similarity solution for the gravitational collapse of an unmagnetized singular logatropic sphere ( sls ) , but they chose not to publish their findings until they had learned how to magnetize the configuration in a nontrivial way ( see the reference to this work in fuller & myers ( 1992 ) , who considered the practical modifications to the protostellar mass - infall rate introduced by `` nonthermal '' contributions to the support against self - gravity ) .
magnetization constitutes an important program to carry out if we try to justify a nonthermal eos as the result of a superposition of propagating mhd waves ( see also holliman & mckee 1993 ) . in this paper , we extend the study of li & shu ( 1996 ) to include the isopedic magnetization of pivotally self - gravitating clouds with a polytropic equation of state . as a by - product of this investigation , we obtain the unanticipated and ironic result that the only way to magnetize a singular logatropic configuration and maintain a scale - free equilibrium is to do it trivially , i.e. , by threading the sls with straight and uniform field lines ( see 6 ) .
a basic consequence of treating the turbulence as a scalar pressure , coequal to the thermal pressure except for satisfying a different eos , is that we do not change the basic topology of the magnetic field .
this assumption may require reassessment if mhd turbulence enables fast magnetic reconnection ( @xcite ) and allows the magnetic fields of highly flattened cloud cores ( mestel & strittmatter 1967 , @xcite ) or pseudodisks ( @xcite ) to disconnect from their background .
recent mhd simulations carried out in multiple spatial dimensions ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ) find turbulence in strongly magnetized media to decay almost as fast as in unmagnetized media .
such decay may be responsible for accelerating molecular cloud core formation above simple ambipolar diffusion rates ( @xcite , @xcite , @xcite ) .
although this result also cautions against treating turbulence on an equal footing as thermal pressure , we attempt a simplified first analysis that includes magnetization to assess the resulting configurational changes when we adopt an alternative eos for the pivotal state .
in particular , different power - law dependences of the radial density profile translate immediately to different time dependences in the mass - infall rate for the subsequent inside - out collapse ( @xcite , mclaughlin & pudritz 1997 ) .
the paper is organized as follows . in
2 we formulate the equations of the scale - free problem and show that each solution depends only on the polytropic exponent @xmath6 and a nondimensional parameter @xmath5 related to the cloud s morphology . in
3 we present the numerical results . in 4 , 5 , and 6
we discuss the limiting form of the solutions .
finally , in 7 we give our conclusions and discuss the possible implications of our results for star formation and the structure of giant molecular clouds .
to begin , we generalize the singular polytropic sphere in the same way that @xcite generalized the singular isothermal sphere ( sis ) . in the absence of an external boundary pressure , the only place the pressure @xmath10 enters in the equations of magnetostatic equilibrium is through a gradient .
consider then the polytropic relation @xmath32 by integrating equation ( [ dpdrho ] ) we recover for @xmath33 the isothermal eos , @xmath34 , ( where @xmath35 is the square of the isothermal sound speed ) and for @xmath36 the logatropic eos , @xmath37 .
we adopt axial symmetry in spherical coordinates and consider a poloidal magnetic field given by @xmath38 where @xmath39 is the magnetic flux .
force balance along field lines requires @xmath40 where @xmath41 is the gravitational potential and @xmath42 is the bernoulli `` constant '' along the field line @xmath43 constant .
poisson s equation now reads @xmath44+{1\over r^2\sin^2\theta } { \partial\over\partial\theta}\left[\sin\theta\left({dh\over d\phi } { \partial\phi\over\partial\theta}-k\rho^{-(2-\gamma ) } { \partial\rho\over\partial\theta}\right)\right]=4\pi g\rho;\ ] ] whereas force balance across field lines reads @xmath45 we look for scale - free solutions of the above equations by nondimensionalizing and separating variables : @xmath46 @xmath47 @xmath48 where @xmath5 is a dimensionless constant that measures the deviation from a force free magnetic field , and @xmath49 and @xmath50 are dimensionless functions of the polar angle @xmath51 . or @xmath52 . ]
these assumptions imply that the equilibria will have spatially constant mass - to - flux ratios ( see below ) .
substitution of equation ( [ nondim ] ) into equations ( [ along ] ) and ( [ across ] ) yields @xmath53\right\}=\ ] ] @xmath54,\ ] ] and @xmath55=-c_\gamma h_0r \phi^{-(2-\gamma)/(4 - 3\gamma)},\ ] ] where a prime denotes differentiation with respect to @xmath51 , and @xmath56 @xmath57 @xmath58 in particular , for @xmath59 , eq . ( [ alongn ] ) gives the dimensionless density for the non - magnetized singular polytropic sphere @xmath60^{1/(2-\gamma)},\ ] ] whereas eq .
( [ acrossn ] ) implies @xmath61 for @xmath62 , in order to satisfy the boundary conditions eq .
( [ bc ] ) . in this case
, the mass - to - flux ratio @xmath1 is infinite .
however , for @xmath63 , eq . ( [ acrossn ] ) admits also the analytic solution of @xmath64 corresponding to a straight and uniform field , while the density function is @xmath65 .
therefore , a spherical logatropic scale free cloud can be magnetized with a uniform magnetic field of any strength , and any value of the spherical mass to flux ratio is allowed.^{-1}$ ] . ] for arbitrary values of @xmath6 and @xmath5 the ordinary differential equations ( odes ) ( [ alongn ] ) and ( [ acrossn ] ) are to be integrated subject to the two - point boundary conditions ( bcs ) : @xmath66=0,\ ] ] @xmath67 the first bc implies that there is no contribution from the polar axis to the mass inside a radius @xmath68 .
the second bc comes from the definition of magnetic flux , i.e. no trapped flux at the polar axis .
the last two bcs imply no kinks at the midplane .
the equilibria are characterized by : ( _ a _ ) the spherical mass - to - flux ratio , is not defined for the polytropic scale free magnetized equilibria because the integral @xmath69 diverges since it can be shown that @xmath70 for @xmath71 . ]
@xmath72 where @xmath73 is the mass enclosed within a radius @xmath68 ; ( _ b _ ) the factor @xmath74 by which the average density is enhanced over the non - magnetized value because of the extra support provided by magnetic fields , @xmath75^{-1/(2-\gamma ) } \int_0^{\pi/2}r(\theta)\sin \theta \ , d\theta , \ ] ] which is equal to @xmath76 if @xmath59 ( see eq . [ [ nonmagr ] ] ) ; ( _ c _ ) the sound speed , @xmath77 and ( _ d _ ) the alfvn speed @xmath78 { 1 \over r(\theta ) \sin^2\theta}.\ ] ] both the sound speed and the alfvn speed scale as @xmath79 for @xmath80 , and @xmath81 for @xmath36 ; for other values of @xmath6 , the exponent of @xmath68 lies between these two values .
it is also of interest to define the ratio @xmath82 of the square of the sound speed and the square of the alfvn speed , each weighted by the density , which is a physical quantity that can be compared with observations : @xmath83 /\sin\theta
d \theta}.\ ] ] if @xmath84 represents only the thermal sound speed , then the observational summary given by fuller & myers ( 1992 ) would imply that @xmath85 in the quiet low - mass cores of gmcs , whereas @xmath86 in their envelopes .
if we include in @xmath84 , however , the turbulent contribution , then the turbulent speed is likely to be sub - alfvnic or marginally alfvnic , and @xmath87 everywhere is probably a better characterization of realistic clouds .
to obtain an equilibrium configuration for given values of @xmath6 and @xmath5 , equations ( [ alongn ] ) and ( [ acrossn ] ) are integrated numerically .
the integration is started at @xmath88 using the expansions : @xmath89 , @xmath90 , with @xmath91 , and @xmath92 a_\gamma h_0 a_0^{2(1-\gamma)/(4 - 3\gamma ) } b_0^{2-\gamma}$ ] .
the values of @xmath93 and @xmath94 are varied until the two bcs at @xmath95 ( eq . [ bc ] ) , are satisfied . for flattened equilibria ( see below ) it is more convenient to start from @xmath96 , where the bcs @xmath97 and @xmath98 are imposed , and integrate toward @xmath88 .
the values of @xmath99 and @xmath100 are then varied until a solution is found that satisfies the two bcs at @xmath101 .
figure 1 shows the resulting flux and density functions @xmath50 and @xmath49 computed for @xmath102 and values of @xmath6 between 0.2 and 1 .
we reproduce the results of @xcite for @xmath33 , which is the only case that obtains perfect toroids ( i.e. , @xmath103 = 0 $ ] ) ; models with @xmath104 have nonzero density at the polar axis .
figure 2 shows the corresponding density contours and magnetic field lines . in the limit @xmath27 , independent of @xmath5
as long as it is nonzero , the pivotal configuration becomes thin disks with an ever increasing magnetic field strength .
table 1 shows the spherical mass to flux ratio @xmath1 , the overdensity parameter @xmath74 , and the ratio of the square of the sound and alfvn speeds @xmath82 .
this table shows that , for fixed @xmath5 , @xmath82 decreases as @xmath6 decreases because the magnetic field becomes stronger .
for the same reason @xmath74 increases .
in contrast , @xmath1 goes through a minimum as @xmath6 decreases .
figures 1 and 2 demonstrate that for @xmath105 ( the logatropic limit ) , @xmath5 is not a measure of the strength of the magnetic fields since @xmath106 diverges as @xmath107 ( see 5 below )
. for fixed @xmath6 , a sequence from small @xmath5 to large @xmath5 progresses through configurations of increasing support by magnetic fields , as demonstrated explicitly for the isothermal case by @xcite .
this behavior is illustrated here for the @xmath108 case in figure 3 , which shows the density contours and magnetic field lines corresponding to values of @xmath5 from 0.05 to 1.5 .
table 2 shows the corresponding values of @xmath1 , @xmath74 , and @xmath82 . for small @xmath5 ,
the equilibria have nearly spherically symmetric isodensity contours and weak quasiuniform magnetic fields that provide little support against gravity . with increasing @xmath5
, the pivotal configurations flatten . the case @xmath109 is already quite disklike : the pole to equator density contrast is @xmath110 . for a thin disk ,
the analysis of @xcite demonstrates that magnetic tension provides virtually the sole means of horizontal support against self - gravity , with gas and magnetic pressures being important only for the vertical structure . in the limit of a completely flattened disk ( @xmath111 ) ,
@xmath112 independent of the detailed nature of the gas eos ( see next section ) .
table 2 shows the spherical mass to flux ratio @xmath1 , the overdensity parameter @xmath74 , and the ratio of the square of the sound and alfvn speeds @xmath82 . again @xmath74 increases monotonically and @xmath82 decreases monotonically as the magnetic support increases with @xmath5 , while @xmath1 goes through a minimum and tends to 1 for large @xmath5 .
since the mass - to - flux ratio @xmath1 is a fundamental quantity that will not change unless magnetic field is lost by ambipolar diffusion , in figure 4 we consider sequences where @xmath1 is held fixed , but @xmath6 is varied .
this figure shows the locus of the set of equilibria with @xmath113 and @xmath114 in the ( @xmath5 , @xmath6 ) plane .
equilibria with @xmath115 are highly flattened when @xmath105 even for small but fixed values of @xmath5 ( see 6 ) .
in fact , to obtain incompletely flattened clouds when one takes the limit @xmath105 , one also needs simultaneously to consider the limit @xmath116 .
unfortunately , because both the density and the strength of the magnetic field at the midplane diverge as the equilibria become highly flattened , we are unable to follow numerically the limit @xmath105 to verify if these sequences of constant @xmath115 will hook to a finite value in the @xmath5 axis , or will loop to @xmath117 , consistent with our demonstration in 4 that flattened disks do not exist in the logatropic limit . or large @xmath5 , it becomes necessary to determine the constants of the expansions of @xmath49 and @xmath50 near the origin with prohibitively increasing accuracy . ]
we speculate that the results for @xmath115 have the following physical interpretation . according to the theorem of shu & li ( 1997 ) ,
only if @xmath118 itself rather than @xmath1 is less than unity , the magnetic field is strong enough overall to prevent the gravitational collapse of a highly flattened cloud .
however , for moderate @xmath5 and @xmath6 when @xmath115 , even the singular equilibria are probably magnetically subcritical , since there can be little practical difference between the spherical mass - to - flux ratio @xmath1 and the `` true '' mass - to - flux ratio @xmath118 for highly flattened configurations .
the latter is formally infinite when @xmath119 only because the mass column goes to zero a little slower than the field column when we perform an integration along the central field line ( see footnote 3 ) . in this interpretation ,
scale - free clouds with @xmath115 and intermediate values of @xmath6 can become highly flattened because magnetic tension supports them laterally against their self - gravity while the soft eos does not provide much resistance in the direction along the field lines .
the squeezing of the cloud toward the midplane is compounded by the _ confining _ pressure of bent magnetic field lines that exert pinch forces in the vertical direction .
both the magnetic tension and the vertical pinch of magnetic pressure disappear when the field lines unbend , as they must to maintain the scale - free equilibria in the limit @xmath105 ( see below ) . as a consequence ,
logatropic configurations become spherical for any value of @xmath1 .
we leave as an interesting problem for future elucidation the determination whether there is still a threshold in @xmath1 below which the sls , embedded with straight and uniform field lines , will not collapse dynamically .
in the limit @xmath121 , the cloud flattens to a thin disk for any @xmath122 .
dominant balance arguments applied to the two odes of the problem reveal the following asymptotic behaviour : . see li & shu ( 1996 ) for the correct asymptotic expansion in this case . ]
@xmath123 @xmath124 to the lowest order in @xmath5 the equation of force balance along field lines ( eq . [ alongn ] ) becomes : @xmath125= \nonumber \\ & & -2\left[{4 - 3\gamma\over ( 2-\gamma)^2}s^{-(1-\gamma ) } + \left({4 - 3\gamma\over 2-\gamma}\right)b_\gamma f^{2(1-\gamma)/(4 - 3\gamma)}\right],\end{aligned}\ ] ] valid over the interval @xmath126 , plus the the integral constraint @xmath127 obtained by integrating eq .
( [ alongn ] ) from @xmath128 to @xmath129 , and applying the first bc ( eq . [ bc ] ) on the polar axis .
the constant @xmath130 is proportional to the surface density of the polytropic disks , given by @xmath131 which , for @xmath33 gives @xmath132 , as found by @xcite .
( [ acrossn ] ) expressing force balance across field lines reduces to @xmath133=-c_\gamma f^{-(2-\gamma)/(4 - 3\gamma)}r_0\delta(\theta-\pi/2),\ ] ] where the parameter @xmath134 is defined by @xmath135 this is equivalent to the equation for force free magnetic fields @xmath136 valid over the interval @xmath126 , plus the condition @xmath137 obtained integrating eq .
( [ acrossn ] ) from @xmath138 to @xmath139 , and taking the limit @xmath140 . for integer @xmath134 ,
solutions of eq .
( [ forcefree ] ) regular at @xmath88 are gegenbauer polynomials of order @xmath134 and index @xmath141 , @xmath142 ( see e.g. abramowitz & stegun 1965 ) . in general
, it can be shown ( chandrasekhar 1955 ) that any axisymmetric force free field , separable in spherical coordinates , can be expressed in terms of fundamental solutions whose radial dependence is given by a combination of bessel functions of fractional order , and the angular dependence by gegenbauer polynomials of index @xmath141 . in our case , the choice of @xmath6 determines a particular exponent of the power - law for the radial part of the flux function , and hence the corresponding value of @xmath134 ( non - integer , except for @xmath63 and 1 ) .
therefore , the magnetic field is force free everywhere except at the midplane where @xmath143 and the condition of force balance across field lines has to be satisfied . in the thin disk limit discussed here , the boundary condition @xmath144 is clearly not fullfilled : the kink of @xmath106 at the midplane provides the magnetic support against self - gravity on the midplane .
currents must exist in the disk to support these kinks . with the definitions @xmath145 eq . ( [ asymp ] ) transforms into @xmath146 which has the solution @xmath147 where @xmath148 is a solution of the homogeneous equation @xmath149 therefore , @xmath150^{-1/(1-\gamma)},\ ] ] and the integral constraint eq .
( [ int1 ] ) becomes @xmath151 the problem is thus reduced to the solution of the two homogeneous equations eq .
( [ forcefree ] ) and eq .
( [ eqq ] ) for the functions @xmath152 and @xmath148 which are determined up to an arbitrary constant .
however , the two integral constraints that would have determined these latter constants ( eqs . [ int2 ] , [ qconst ] ) , contain the additional unknown parameter @xmath130 .
the system of equations is closed by the requirement that @xmath153 substituting eq .
( [ rh0 ] ) and eq .
( [ fh0 ] ) in eq .
( [ lambdar ] ) , one obtains @xmath154 i.e. , @xmath155 which gives the remaining condition .
( [ forcefree ] ) and ( [ eqq ] ) can be solved numerically by starting the integration at @xmath88 with the series expansions : @xmath156 $ ] , and @xmath157 \theta^4 + \ldots\right\}$ ] , where @xmath158 and @xmath159 are arbitrary constants .
are of little use here : the one at @xmath88 reduces to the condition @xmath160 , trivially satisfied ; the second bc , @xmath98 can not be applied because of the @xmath161-function at @xmath162 . ] the constants @xmath163 and @xmath130 are then determined by the constraints expressed by eqs .
( [ int1 ] ) , ( [ int2]),and ( [ int3 ] ) .
figure 5 shows the functions @xmath152 and @xmath164 obtained for @xmath22 and increasing values of @xmath5 from 0.4 to 1.5 compared with the asymptotic expressions computed here .
already for @xmath165 , the actual @xmath152 and @xmath164 are very close to the corresponding asymptotic functions eq .
( [ rh0 ] ) and eq .
( [ fh0 ] ) .
table 3 shows the value of the angle @xmath166 of the magnetic field with the plane of the disk , the flux function @xmath167 evaluated at @xmath96 ( indicative of the magnetic field stength ) , and the surface density parameter @xmath130 , as functions of @xmath6 .
the angle @xmath166 ranges from @xmath168 for the isothermal case @xmath33 to @xmath169 in the logatropic case @xmath7 .
correspondingly , the magnetic flux in the disk and the surface density both diverge as @xmath105 for any large but finite value of @xmath5 .
for the isothermal case @xmath171 , @xcite have shown how the sis is recovered for @xmath172 from a family of toroids with zero density on the polar axis . for @xmath173 , in the limit @xmath170 , the asymptotic expansions are given by : @xmath174^{1/(2-\gamma ) } + p(\theta ) h_0^{(4 - 3\gamma)/(3 - 2\gamma ) } + \ldots\ ] ] @xmath175 to the lowest order in @xmath5 , eqs .
( [ alongn ] ) and ( [ acrossn ] ) become : @xmath176\right\}=\ ] ] @xmath177,\ ] ] and @xmath178=-c_\gamma \left[{4 - 3\gamma\over ( 2-\gamma)^2}\right]^{1/(2-\gamma ) } g^{-(2-\gamma)/(4 - 3\gamma)},\ ] ] the bc for the functions @xmath179 and @xmath180 are the same as those for @xmath49 and @xmath50 in eq .
( [ bc ] ) .
figure 6 shows the convergence of the solutions of the full set of equations ( [ alongn ] ) and ( [ acrossn ] ) obtained for @xmath22 and decreasing values of @xmath5 from 0.4 to 0.05 , to the asymptotic solutions obtained by integrating the equations above .
notice that @xmath181 and @xmath182 showing that the sequence of equilibria with @xmath22 originates from the corresponding unmagnetized spherical state ( eq .
[ nonmagr ] ) by reducing the density on the pole and enhancing it on the equator .
the same behaviour is found for any value of @xmath6 in the range @xmath183 . for @xmath171 ,
the function @xmath179 diverges at @xmath88 , indicating that this expansion is not appropriate in the isothermal case , as in the case @xmath121 . for the same reason
, the expansion also fails for @xmath63 , since both @xmath180 and @xmath179 diverge on the equatorial plane .
these flattened configurations are supported by magnetic and gas pressure against self - gravity .
the intensity of the magnetic field can become very high even though @xmath5 is small , because the latter parameter measures not the field strength but the deviations from a force free field ( see eq .
[ nondim ] ) .
we consider in this section the logatropic limit @xmath184 . as anticipated in 2 , for @xmath63 eq .
( [ alongn ] ) and ( [ acrossn ] ) admit the analytical solution @xmath185 and @xmath186 corresponding to a sls threaded by a straight and uniform magnetic field .
this solution represents the only possible scale - free isopedic configuration of equilibrium for a magnetized cloud with a logatropic eos . to show this
, we use the results of 4 and 5 for @xmath121 and @xmath172 to find the limit of the equilibrium configurations for @xmath27 and fixed ( small or large ) values of @xmath5 . in the limit
@xmath120 , @xmath105 , analytic solutions to equations ( [ forcefree ] ) and ( [ eqq ] ) exist .
the magnetic field tends to become uniform and straight , @xmath187 , but @xmath188 diverges , as shown in table 3 , and therefore @xmath189 also diverges ( see eq . [ sol ] ) . eq .
( [ surfden ] ) shows in this limit that the surface density @xmath190 is independent of @xmath68 , therefore , no pressure gradients can be exerted in the horizontal direction .
the value @xmath191 diverges as @xmath105 for any value of @xmath5 because @xmath192 ( see table 3 ) .
the magnetic flux threading the disk , @xmath193 , becomes infinite in order to keep the mass to flux ratio @xmath194 equal to 1 .
therefore , the limiting configuration approaches a uniform disk with infinite surface density , threaded by an infinitely strong uniform and straight magnetic field .
if we now examine the case @xmath170 , in the limit @xmath195 , it is easy to show from eq .
( [ fluxsmallh ] ) that the magnetic field tends to become uniform , @xmath196 , but @xmath197 .
consequently , the density function @xmath179 also diverges in @xmath198 , and the configuration again approaches a thin disk threaded by an uniform , infinitely strong , magnetic field .
we conclude that scale - free logatropic clouds can not exist as magnetostatic disks except in some limiting configuration . in the absence of such limits , the equilibria are spherical and can be magnetized only by straight and uniform field lines ; i.e.
, the magnetic field is force - free and therefore given by @xmath59 .
the inside - out gravitational collapse of such a sls would still proceed self - similarly as in the solution of mclaughlin & pudritz ( 1997 ) , but the frozen - in magnetic fields would yield a dependence with polar angle that eventually produces a pseudodisk ( galli & shu 1993a , b ; allen & shu 1998a ) .
we have solved the scale - free equations of magnetostatic equilibrium of isopedic self - gravitating polytropic clouds to find pivotal states that represent the initial state for the onset of dynamical collapse , as first proposed by @xcite for isothermal clouds .
compared to unmagnetized equilibria , the magnetized configurations are flattened because of magnetic support across field lines .
the degree of this support is best represented by the ratio of the square of the sound to alfvn speeds @xmath82 , or the overdensity parameter @xmath74 , since they are always monotonic functions of @xmath5 and @xmath6 .
configurations with @xmath33 become highly flattened as the parameter @xmath5 increases . when @xmath104 ( softer eos ) the equilibria get flattened even faster at the same values of @xmath5 , since along field lines there is less support from a soft eos than for a stiff one .
however , it seems that in the logatropic limit flattened disks do not exist : the singular scale - free equilibria can only be spherical uniformly magnetized clouds .
figure 7 shows a schematic picture of the @xmath199 plane indicating the topology of the solutions for scale free magnetized isopedic singular self - gravitating clouds . in self - gravitating clouds , the joint compression of matter and field
is often expressed as producing an expected relationship : @xmath200 , with different theorists expressing different preferences for the value of @xmath201 ( e.g. , mestel 1965 , fiedler & mouschovias 1993 ) .
no local ( i.e. , point by point ) relationship of the form @xmath202 holds for the scale - free equilibria studied in this paper . however , if we average the magnetic field strength and mass density over ever larger spherical volumes centered on @xmath203 , we do recover such a relationship : @xmath204 , where angular brackets denote the result of such a spatial average and @xmath205 .
we may think of the result @xmath206 as arising physically from a combination of two tendencies .
( a ) slow cloud contraction in the absence of magnetic fields and rotation tends to keep roughly one jeans mass inside every radius @xmath68 , which yields @xmath207 , or @xmath208 if @xmath209 . ( b )
slow cloud contraction in the absence of gas pressure tends to keep roughly one magnetic critical mass inside every radius @xmath68 , which yields @xmath210 = constant , or @xmath211 if gas pressure ( thermal or turbulent ) plays a comparable role to magnetic fields in cloud support .
notice that our reasoning does not rely on arguments of cloud geometry , e.g. , whether cloud cores flatten dramatically or not as they contract ; nor does it depend sensitively on the precise reason for core contraction , e.g. , because of ambipolar diffusion or turbulent decay .
crutcher ( 1998 ) claims that the observational data are consistent with @xmath212 .
if we take crutcher s conclusion at face value , we would interpret the observations as referring mostly to regions where the eos is close to being isothermal @xmath213 , which is the approximation adopted by many theoretical studies that ignore the role of cloud turbulence .
the result is not unexpected for low - mass cloud cores , but we would not naively have expected this relationship for high - mass cores and cloud envelopes , where the importance of turbulent motions is much greater .
unfortunately , the observational data refer to different clouds rather than to different ( spatially averaged ) regions of the same cloud , so there is some ambiguity how to make the proper connection to different theoretical predictions .
there may also be other mechanisms at work , e.g. , perhaps a tendency for observations to select for regions of nearly constant alfvn speed , @xmath214 constant ( bertoldi & mckee 1992 ) .
thus , we would warn the reader against drawing premature conclusions about the effective eos for molecular clouds , or the related degree to which observations can at present distinguish whether molecular clouds are magnetically supercritical or subcritical . if molecular clouds are magnetically supercritical , with @xmath1 greater than 1 by order unity ( say , @xmath215 ) , then an appreciable fraction ( say , 1/2 ) of their support against self - gravity has to come from turbulent or thermal pressure ( elmegreen 1978 , mckee et al . 1993 ,
crutcher 1998 ) . modeled as scale - free equilibria ,
such clouds with @xmath82 of order unity are not highly flattened ( see tables 1 , 2 and figs . 2 , 3 ) .
suppose we try gravitationally to extract a subunit from an unflattened massive molecular cloud , where the cloud as a whole is only somewhat supercritical , @xmath216 . if the subunit s linear size is smaller than the vertical dimension of the cloud by more than a factor of 2 , which will be the case if we consider subunits of stellar mass scales , then this subunit will not itself be magnetically supercritical .
magnetically subcritical pieces of clouds can not contract indefinitely without flux loss , so star formation in _ unflattened _ clouds , if they are not highly supercritical , needs to invoke some degree of ambipolar diffusion in order to produce small dense cores that can gravitationally separate from their surroundings . on the other hand ,
if molecular clouds are magnetically critical or subcritical , with @xmath217 , then almost all scale - free equilibria are highly flattened , with @xmath82 appreciably less than unity . on a small scale , any subunit of this cloud , even subunits with vertical dimension comparable to the cloud as a whole ,
would also be magnetically critical or subcritical .
for such a subunit to contract indefinitely , we would again need to invoke ambipolar diffusion to make a cloud core magnetically supercritical .
thus , although the decay of turbulence can accelerate the formation of cloud cores , the ultimate formation of stars from such cores may still need to rely on _ some _ magnetic flux loss ( but perhaps not more than a factor of @xmath218 ) to trigger the evolution of the cores toward gravomagneto catastrophe and a pivotal state with a formally infinite central concentration . on the large scale ,
if gmcs are modeled as flattened isopedic sheets , @xcite proved that magnetic pressure and tension are proportional to the gas pressure and force of self - gravity .
their theorems hold independently of the detailed forms of the eos or the surface density distribution . if gmcs are truly highly flattened with typical dimensions , say , of 50 pc @xmath219 50 pc @xmath219 a few pc or even less then many aspects of their magnetohydrodynamic stability and evolution become amenable to a simplified analysis through the judicious application and extension of the theorems proved by @xcite ( e.g. , see allen & shu 1998b , @xcite ) .
this exciting possibility deserves further exploration .
acknowledges support by cnr grant 97.00018.ct02 , asi grant ars-96 - 66 and ars-98 - 116 , and hospitality from unam , mxico . s.l . acknowledges support by j. s. guggenheim memorial foundation , grant dgapa - unam and conacyt , and hospitality from osservatorio di arcetri .
f.c.a . is supported by nasa grant no .
nag 5 - 2869 and by funds from the physics department at the university of michigan . the work of f.h.s .
is supported in part by an nsf grant and in part by a nasa theory grant awarded to the center for star formation studies , a consortium of the university of california at berkeley , the university of california at santa cruz , and the nasa ames research center .
shu , f. h. , allen , a. , shang , h. , ostriker , e. c. , & li , z. y. 1999 , in the physics of star formation and early stellar evolution ii , nato asi , ed .
c. j. lada and n. d. kylafis ( dordrecht : kluwer ) , in press 1 & 1.94 & 1.50 & 0.668 0.9 & 1.80 & 1.68 & 0.632 0.8 & 1.63 & 1.76 & 0.543 0.7 & 1.45 & 1.80 & 0.434 0.6 & 1.27 & 1.84 & 0.321 0.5 & 1.09 & 1.91 & 0.218 0.4 & 0.927 & 2.14 & 0.122 0.3 & 0.947 & 5.06 & 0.0165 0.2 & 0.992 & 13.8 & 0.00181 0 & @xmath220 & 1 & @xmath220 0.05 & 4.35 & 1.18 & 6.93 0.1 & 2.83 & 1.21 & 2.81 0.2 & 1.85 & 1.31 & 1.08 0.3 & 1.45 & 1.45 & 0.577 0.4 & 1.23 & 1.64 & 0.348 0.5 & 1.09 & 1.91 & 0.218 0.6 & 1.01 & 2.34 & 0.134 0.7 & 0.980 & 3.00 & 0.0786 0.8 & 0.977 & 3.91 & 0.0448 0.9 & 0.982 & 4.95 & 0.0265 1.0 & 0.987 & 6.06 & 0.0165 1.1 & 0.991 & 7.22 & 0.0108 1.2 & 0.993 & 8.44 & 0.00733 1.3 & 0.995 & 9.71 & 0.00514 1.4 & 0.996 & 11.04 & 0.00369 1.5 & 0.997 & 12.42 & 0.00272 1 & @xmath168 & 1 & 2 0.9 & @xmath221 & 1.19 & 2.65 0.8 & @xmath222 & 1.52 & 3.61 0.7 & @xmath223 & 2.05 & 5.10 0.6 & @xmath224 & 2.93 & 7.55 0.5 & @xmath225 & 4.50 & 11.9 0.4 & @xmath226 & 7.59 & 20.5 0.3 & @xmath227 & 14.7 & 40.4 0.2 & @xmath228 & 36.8 & 102 0.1 & @xmath229 & 166 & 467 0 & @xmath169 & @xmath220 & @xmath220 | we investigate the equilibrium properties of self - gravitating magnetized clouds with polytropic equations of state with negative index @xmath0 .
in particular , we consider scale - free isopedic configurations that have constant dimensionless spherical mass - to - flux ratio @xmath1 and that may constitute `` pivotal '' states for subsequent dynamical collapse to form groups or clusters of stars . for given @xmath2 , equilibria with smaller values of @xmath1 are more flattened , ranging from spherical configurations with @xmath3 to completely flattened states for @xmath4 . for a given amount of support provided by the magnetic field as measured by the dimensionless parameter @xmath5 , equilibria with smaller values of @xmath6
are more flattened .
however , logatropic ( defined by @xmath7 ) disks do not exist .
the only possible scale - free isopedic equilibria with logatropic equation of state are spherical uniformly magnetized clouds . |
the systematic study of x - ray source populations in external galaxies first became possible with the _ einstein _ observatory in 1978 .
significant advances in the field were made with subsequent satellites such as _ rosat _ , but a giant leap forward has been taken with _
chandra_. with its subarcsecond spatial resolution and high sensitivity
it is possible to resolve the vast majority of the luminous ( @xmath4 ) x - ray sources in galaxies out to distances of @xmath5 mpc .
in addition , the spectrometric capabilities of the _ chandra _ acis ccd detector allow spectral properties of sources to be extracted .
for two recent reviews of the study of x - ray sources in normal galaxies , emphasizing results from _ chandra _ , see @xcite and @xcite .
we now know that x - ray source populations in galaxies are dominated at high luminosities ( @xmath4 , the range typically explored with _ chandra
_ observations ) by x - ray binaries ( xrbs ) consisting of an accreting neutron star or black hole and a stellar companion . in addition
, galaxies usually have a few young supernova remnants ( snrs ) in this luminosity range .
unsurprisingly , the x - ray populations in early - type galaxies ( e and s0 ) seem to consist mostly of low - mass x - ray binaries ( lmxbs ) , whereas galaxies with younger stellar populations ( spiral and irregular galaxies ) typically have a much higher fraction of the shorter lived high - mass x - ray binaries ( hmxbs ) . in galaxies with a high star formation rate hmxbs
are especially common .
an important class of sources in external galaxies are the so - called ultraluminous x - ray sources ( ulxs ) , usually defined as non - nuclear sources with implied isotropic x - ray luminosities @xmath6 .
detections of ulxs with luminosities as high as @xmath7 have been reported ( see , e.g. , * ? ? ?
. the nature of ulxs is still debated , and it has been argued that at least some of them might be a new class of objects , so - called intermediate - mass black holes ( imbhs ) with masses @xmath8 ( see , e.g. , * ? ? ?
we present in this paper the x - ray source population study of the spiral galaxy ngc 6946 and the interacting irregular / spiral system ngc 4485/4490 .
a special emphasis is placed on studying the long - term ( weeks to years ) variability properties of the source populations .
these galaxies were chosen because they are nearby ( @xmath9 mpc ) and have multiple ( three or more ) long ( @xmath10 ks ) _ chandra _ acis exposures spanning a baseline of a few years .
both show an enhanced star formation rate .
the spiral ngc 6946 also has the fortunate characteristic of being observed nearly face - on , and the ngc 4485/90 system has low galactic extinction and a large number of ulxs . for more background information on the galaxies , see
[ sec : ngc6946 ] and [ sec : ngc4485 ] .
not much work has been done on the long - term variability of x - ray sources in external galaxies , since _
observations at multiple epochs are usually not available . we know from observations in our own galaxy
that time variability of various kinds , including transient outbursts , eclipses , dips , as well as less severe variations in flux , is very common among xrbs .
temporal and spectral analysis carried out for the most luminous sources in nearby galaxies shows behavior similar to that in galactic xrbs , clearly pointing to xrb populations ( see * ? ? ?
* and references therein ) .
for example , @xcite find that among 204 detected sources in m31 , 50% are variable on timescales of months and 13 are transients .
@xcite find short - timescale flares in 3 out of 157 sources in the elliptical galaxy ngc 4697 , and two of the flares have durations and luminosities similar to galactic superbursts ( thermonuclear bursts with very long ( hours ) durations and very large fluences ; see * ? ? ?
@xcite also find long - term variability in 26 out of 124 sources in ngc 4697 , and 11 of those are transient candidates . @xcite , analyzing seven _
observations of the antennae galaxies , find intensity and/or spectral variability among sources on timescales of years , months , days , and hours .
overall , @xmath11 of the sources detected in each observation show either spectral or intensity variation but do not all follow a common trend , indicating that there are various classes of sources . of the 14 ulxs in the antennae , 12 show long - term variability . in general , variability of some sort
is very common among ulxs ( see * ? ? ?
* ; * ? ? ?
despite widespread variability in luminosity among individual sources , x - ray luminosity functions ( xlfs ) seem to be remarkably stable from one observation to another , as indicated by observations of ngc 5128 @xcite , m33 @xcite , and the antennae @xcite .
the organization of the paper is as follows . in [ sec : analysis ] we describe the common analysis steps performed for both galaxies , including source detection , photometry , the construction of light curves and hardness ratios , and testing for flux and spectral variability . in [ sec : observations ] we discuss general properties and the observations of the galaxies and present the results of the source detection . in
[ sec : properties ] we present and discuss our results on various properties of the source populations in the galaxies .
finally , in [ sec : conclusions ] we briefly summarize our results .
included are tables with the source catalog for each galaxy and various observed and derived parameters for each source .
the _ chandra _ observations of both galaxies were analyzed with the ciao software , version 3.4 ( caldb ver .
3.3.0 ) , and with acis extract ( hereafter ae ) , version 3.131 .
ae is an acis point source extraction package developed at pennsylvania state university @xcite , which assists in a large variety of complex data processing tasks .
the procedures used in ae are described in @xcite and @xcite .
we note that all errors mentioned in this paper are , unless explicitly stated otherwise , poisson errors with a confidence level corresponding to 1 @xmath12 gaussian .
they were computed using the gehrels approximation of poisson limits @xcite .
the level 1 event files from the _ chandra _ data archive were reprocessed , and new level 2 event files created by following the standard acis data preparation procedure recommended by the _
chandra _ x - ray center ( cxc ) .
the data were checked for periods of unusually high background .
no such periods were found in any observation of either galaxy .
a monochromatic exposure map evaluated at 1.5 kev was created for each observation , and the event files , as well as the exposure maps , merged to create a co - added exposure covering all the observations .
a @xmath13 kev band image encompassing the whole galaxy was subsequently created from the merged event file , and ` wavdetect ` , the ciao wavelet source detection algorithm , run at nine scales ( 1 , @xmath14 , 2 , @xmath15 , 4 , @xmath16 , 8 , @xmath17 , and 16 pixels ) , using the merged exposure map ( the ` sigthresh ` parameter , which sets the threshold for identifying a pixel as belonging to a source , was set at a value of 1 over the number of pixels in the image ) .
further analysis was performed with ae .
the input to the program was the master source list from the ` wavdetect ` run and the cleaned event data for each observation .
after visual inspection of the data ( both co - added and individual exposures ) a few extra candidate sources , not detected by ` wavdetect ` , were typically added to the list . since we only perform source detection on the co - added exposure , it is conceivable that ` wavdetect ` might for example miss weak transient sources in high - background regions . in the end , only in one case did such a candidate source end up surviving the significance cuts described below .
source cxou j123038.2 + 413831 in ngc 4485/90 is in close proximity to a much brighter transient source , causing ` wavdetect ` to detect only the brighter source in the co - added exposure .
the existence of two individual sources is readily apparent when looking at individual observations , in some of which the brighter source is in quiescence .
an important characteristic of ae is that it tailors the count extraction region individually to each source in each observation based on the shape of the point spread function ( psf ) at the source location on the detector .
counts were by default extracted from a region with a psf encircled energy fraction of 0.90 , but in cases of crowded sources ae automatically reduces the fraction to prevent overlapping extraction regions .
the response functions computed by ae are corrected to take into account the flux missed by the finite extraction region in each case .
possibly spurious detections from our candidate source list were rejected based on a source significance statistic computed by ae ( for definition see * ? ? ?
* ) , along with visual inspection of the individual and merged observations , as well as the ` wavdetect ` significance statistic .
we set a conservative limit where detections with ae significance @xmath18 @xmath12 ( which typically corresponds to ` wavdetect ` significance @xmath19 @xmath12 for the co - added exposure ) were rejected .
the source positions were overlaid on an optical image of the galaxy along with the @xmath20 elliptical isophote ( obtained from * ? ? ? * ) . in general , we considered sources inside the @xmath20 isophote to be coincident with the system and removed other sources from the list .
however , in a few cases sources just outside the isophote were included based on visual inspection of the galaxy s optical extent , which can be quite irregular .
the accuracy of the source positions was refined with the help of ae . since the average off - axis angle was @xmath21 for nearly all sources , we used positions based on the centroid of the extraction regions as recommended in @xcite .
ae is capable of performing extensive broadband photometry .
for all sources an estimate of both the photon flux and the energy flux in the @xmath13 kev band was computed from the ae results for each observation , as well as for the co - added exposure . for all but the faintest sources the photon flux was computed by dividing the number of net ( i.e.
, background - subtracted ) counts by the exposure time and the mean effective area at the source location for a given energy band ( the mean_arf parameter in ae , which takes into account flux missed by the finite extraction region ) . for computing the mean effective area
the simplifying assumption of a flat incident spectrum within the given energy band was made . for an estimate of the energy flux the photon flux
was multiplied by the mean observed photon energy in the corresponding band , as computed by ae .
this was done in five adjacent energy bands spanning together the full @xmath13 kev band ( @xmath22 , @xmath23 , @xmath24 , @xmath25 , and @xmath26 kev ) , and the results from the individual bands then added together .
this is a compromise between calculating the fluxes directly for the full band ( and thus assuming a flat incident spectrum over that whole band , which gives rise to a systematic error ) and dividing the counts in each spectral channel by the corresponding value of the auxiliary response function ( arf ) , this latter method being susceptible to large poisson errors . for the ulxs in the galaxies we performed spectral fitting ( see below ) and
compared the flux results from our `` band method '' to the spectral fitting results .
we found that the values from the band method were typically @xmath27 higher than the ones from spectral fitting and that the discrepancy was usually slightly greater for the photon flux than the energy flux . for observations of sources with very few counts
it is perhaps more sensible to give an upper limit to the flux instead of an actual flux estimate .
since we do not perform source detection on individual observations , only on the co - added exposure , it is effectively meaningless for us to talk about whether a source was detected or not in an observation .
we therefore adopt a fixed threshold of five net counts , below which we give upper limits in our tables , with 3 @xmath12 gaussian confidence , instead of flux values .
we calculate the limits with the sherpa package in ciao .
they are derived from the full - band counts , assuming a power - law spectrum with a photon index of 1.7 ( characteristic of best - fit spectra for sources in our sample ) and galactic absorption , and using the response functions at the source location .
we note that for the purpose of computing the flux variability parameter described in
[ sec : variability ] below , we always use actual flux estimates , although for less than 5 net counts those estimates are calculated using sherpa ( as described above ) rather than our band method . in the case of the ulxs
the flux estimates were improved on by fitting eight simple spectral models to each of the _ chandra _ observations in xspec , and using in each case the best - fitting model to derive the fluxes .
the models considered were absorbed power - law , bremsstrahlung , blackbody , and multicolor disk blackbody models , as well as two - component models with one of the previous components plus an additional power - law component ( both equally absorbed ) .
the absorption parameter was allowed to vary freely with a minimum value equal to the galactic absorption column .
after converting the energy fluxes to luminosities based on our adopted distance to each galaxy long - term light curves ( spanning all observations ) were created for each source .
it should be noted that all fluxes and luminosities quoted in this paper are absorbed values ( i.e. , they are not corrected for absorption along the line of sight ) , and they are , unless otherwise noted , given for the @xmath13 kev range .
the error bars shown in light curves and other plots only take into account the poisson errors in the counts .
lcccc 2001 sep 7 & acis - s & 1043 & 58.29 & f + 2002 nov 25 & acis - s & 4404 & 29.92 & f + 2004 oct 22 & acis - s & 4631 & 29.70 & f + 2004 nov 6 & acis - s & 4632 & 27.94 & f + 2004 dec 3 & acis - s & 4633 & 26.62 & f + hardness ratios were chosen as our primary means of investigating the spectral properties of the source populations , since a large portion of the sources had too few counts to be amenable to spectral fitting .
two kinds of hardness ratios were computed for each source in each observation , as well as for the co - added exposure .
our soft color is defined as @xmath28 and the hard color as @xmath29 where s , m , h , and t are the net counts in the @xmath22 kev ( soft ) , @xmath30 kev ( medium ) , @xmath31 kev ( hard ) , and @xmath13 kev ( total ) bands , respectively .
the counts used to calculate the hardness ratios were in each case divided by the mean value of the effective area over the given energy band at the location of the source .
this was done to account for differences in effective areas between different energy bands , and differences due to variations in source locations on the detector .
the errors in the hardness ratios were computed using standard error propagation . although it is not possible to conclusively classify an individual source based on its hardness ratios alone , it is possible to draw conclusions about bulk trends within a source population , as shown by @xcite .
they propose a classification scheme based on hardness ratios , in which they identify the regions in a color - color diagram where sources of a given type are most likely to fall .
we utilize this method to draw conclusions about the source populations ( see [ sec : spec_class ] ) . it should be noted that our hardness ratio definitions differ slightly from the ones used by @xcite .
they use an upper limit of 8 kev to their hard and total bands , in contrast to our using 7 kev .
however , for the vast majority of sources only a very small fraction of the counts in the @xmath32 kev band is between 7 and 8 kev , so this should have a negligible effect .
they also use a slightly different method to account for variations in effective area . as @xcite discuss in their paper , absorption will tend to blur the distinction in a color - color diagram between the different source types
. however , this effect should not be very severe for the galaxies studied in this paper ; although ngc 6946 suffers from substantial extinction , it is observed face - on , and the absorption column in the direction of ngc 4485/90 is low ( see [ sec : observations ] ) .
we compute a significance parameter for long - term flux variability for each source : @xmath33 where @xmath34 denotes the photon flux of a source in the @xmath35 _ chandra _ observation , and @xmath36 is the corresponding error .
we define a source as variable if @xmath37 .
we also quantify the variability by computing the ratio @xmath38 .
we consider a source to be a transient candidate if its measured flux is consistent with zero ( within @xmath39 @xmath12 error ) in at least one observation , and if the source is also clearly variable ( @xmath37 ) .
we test for short - term flux variability ( within single observations ) by performing a one - sided kolmogorov - smirnov test for each source in each observation , comparing event times to a uniform count - rate model .
we define a source as showing evidence for variability if the significance of the computed k - s statistic is less than @xmath40 ( i.e. , if the null hypothesis that the count rate is uniform can be rejected at the 99.7% level or more , corresponding to approximately 3 @xmath12 gaussian confidence ) . to search for long - term spectral variability ,
we compute significance parameters for the variability in the hardness ratios .
they are defined analogously to the flux variability parameter above .
again , we require 3 @xmath12 confidence to classify a source as variable . apart from _ chandra _ , there are three other x - ray telescopes that have had good enough spatial resolution to possibly be of significant value to this study .
the _ einstein _
hri , the _ rosat _ hri , and the _ xmm - newton _ epic cameras all had or have an on - axis resolution of @xmath41 .
however , the _ einstein _ and _ rosat _ hris were only sensitive in the @xmath42 and @xmath43 kev bands and had negligible energy resolution .
none of the galaxies analyzed in this paper was observed with the _ einstein _ hri . for the ulxs that were well resolved from neighboring bright sources
the long - term light curves were extended by adding data points from _ rosat _ and _ xmm - newton _ observations .
it was usually necessary to use small extraction regions to minimize the contamination from nearby sources and from the diffuse emission in which many sources are embedded .
lcc 1994 may 14 & rh600501n00 & 59.89 + 1995 aug 13 & rh600718n00 & 21.51 + 1996 aug 10 & rh500476n00 & 8.36 + _ rosat _ count rates were derived from the rdf ( rationalized data format ) files available from the _ rosat _ archive .
the event data files were corrected for the hri aspect error ( see , e.g. , * ? ? ?
* ) using the correction script developed at the smithsonian astrophysical observatory ( sao ) .
background - subtracted count rates were extracted using the asterix software package ( ver .
2.3-b2 ) . the
` xrtsort ` , ` xrtsub ` , and ` xrtcorr ` tasks were used , which automatically make dead - time , vignetting , and quantum efficiency corrections . in all cases counts
were extracted from a circle of radius @xmath44 . for point sources with an off - axis angle @xmath21 the psf encircled energy fraction for such
a circle is @xmath45 @xcite .
the count rates were therefore divided by this number to correct for the counts missed by the finite extraction region . to facilitate comparison with the _ chandra _ results count rates from _ rosat _ were converted to energy flux in the @xmath13 kev band using pimms . to determine the spectral model to use for the conversion the eight models described in [ sec : light_curves ] were fitted to the source spectrum from the co - added _ chandra _ exposure , and the best - fitting model used . for _ xmm - newton _ data a new analysis was performed for each of three epic detectors using the sas software , versions 7.0.0 and 7.1.0 , and the latest available calibration files at the time of the analysis .
starting with the original data files ( odfs ) from the _ xmm - newton _ archive , mos and pn data were reprocessed using the ` emproc ` and ` epproc ` tasks . after excluding periods of high background
, source spectra were extracted with the ` evselect ` task .
the ` eregionanalyse ` task was used to center the extraction radius on the centroid of the counts distribution for each source .
counts were extracted from circles with radii in the range @xmath46 to @xmath47 as deemed appropriate in each case .
the event selection criteria used for the spectral extraction were @xmath48 and @xmath49 for the mos instruments , and @xmath50 and @xmath51 for the pn , as recommended in the analysis threads from the _ xmm - newton _ science operations centre .
response files were created using the ` rmfgen ` and ` arfgen ` tasks , incorporating a correction for the flux missed by the finite extraction regions .
the spectra were fitted with the eight spectral models described in [
sec : light_curves ] .
the spectra for each of the three detectors were fitted simultaneously , allowing only an overall normalization factor to vary between detectors to account for differences in calibration . using the best - fitting model for each source , an energy flux in the @xmath13 kev band
was then computed .
the individual fluxes for the three instruments were combined into a single flux result by taking the mean weighted with the inverse variance ( i.e. , squared error ) of each individual flux .
ngc 6946 is one of the closest large spiral galaxies and is seen nearly face - on .
it is classified by @xcite as type sab(rs)cd .
it has been extensively studied in almost all wavelength bands and has been shown to have strong nuclear starburst activity @xcite .
high star formation activity in its disk has also been observed @xcite , and in agreement with this ngc 6946 has been host to nine recorded supernovae in the last 91 years ( three in the past 7 years ) , currently the highest known number of historical supernovae in any galaxy .
the global star formation rate in ngc 6946 for stars in the mass range @xmath52 is estimated by @xcite to be @xmath53 , and @xcite estimate the dynamical mass of the galaxy to be @xmath54 .
the distance to ngc 6949 is given by @xcite as 5.1 mpc and is listed as 5.5 mpc in @xcite , assuming a hubble constant value of @xmath55 .
more recently , @xcite derive a distance of @xmath56 based on a type ii supernova , and @xcite find a distance of @xmath57 based on photometric observations of the brightest stars in the galaxy . finally , @xcite estimate the distance to the ngc 6946 galaxy group ( consisting of ngc 6946 and seven other small galaxies ) to be @xmath58 , based on observations of the brightest blue giants in the galaxies .
we choose to adopt a distance of 5.9 mpc , as do @xcite , @xcite , and @xcite .
the galactic absorption column in the direction of ngc 6946 is @xmath59 @xcite .
lccccccccc[b ] 2004 jun 09 & 0200670101 & 3.95 & ff - me & & 4.05 & ff - me & & 2.50 & ff - me + 2004 jun 13 & 0200670301 & 11.47 & ff - me & & 11.68 & ff - me & & 8.93 & ff - me + 2004 jun 25 & 0200670401 & 7.69 & ff - me & & 8.49 & ff - me & & 4.81 & ff - me + 2006 jun 02 & 0401360201 & 4.40 & ff - me & & 4.31 & ff - me & & 3.81 & ff - me + 2006 jun 18 & 0401360301 & 5.28 & ff - me & & 5.60 & ff - me & & 3.58 & ff - me + ngc 6946 was first observed in x - rays with the _ einstein _ satellite ; three observations with the _ einstein _ ipc were made in 19791981 . analyzing these observations , @xcite resolve the x - ray emission from the galaxy into two peaks .
ngc 6946 was observed with the _ rosat _ pspc in 1992 , twice with the _ asca _ satellite in 1993 and 1994 , and three times with the _ rosat _ hri in 199496 .
@xcite analyzes the _ rosat _ pspc observation and resolves emission from the galaxy into nine point sources , as well as finding evidence for diffuse emission . analyzing the first of the three _ rosat _ hri observations and the second of the two _ asca _ observations @xcite detect 14 point sources and diffuse emission from the galaxy .
most recently , ngc 6946 has been observed five times with _ chandra _ between 2001 and 2004 , and 11 times with _ xmm - newton _ in 20032006 .
the most detailed studies so far of the x - ray emission components of ngc 6946 are presented in a pair of papers based on the first of the five _ chandra _ observations .
@xcite study the point source population and find 72 sources ; the diffuse emission is extensively studied in @xcite .
the _ chandra _ observations of ngc 6946 are summarized in table [ obs6946chandra ] .
these are all acis - s full - frame imaging observations taken in timed exposure mode .
the frame time was in all cases set at the standard value of 3.2 s , and events were telemetered in the faint format .
the observations span a period of about 39 months and have a cumulative good live exposure time of 172.5 ks . in each observation
the majority of sources falls on the back - illuminated s3 chip , and a smaller portion falls on the front - illuminated s2 and/or s4 chips .
none of the observations covers the whole @xmath20 isophote .
the data were analyzed as described in
[ sec : analysis ] .
source detection was performed on a @xmath60 rectangle encompassing the system .
a circular region of radius @xmath61 at the center of the galaxy ( circle centered at @xmath62 , @xmath63 ) was excluded from further analysis , since the sources there are too crowded and embedded in too much localized diffuse emission for reliable background - subtracted count rates to be extracted .
this central region seems to contain at least four and possibly six point sources . outside this region
a total of 90 reliable source detections down to luminosities of a few times @xmath0 ( depending on the location ) were found to be coincident with the galaxy ( see fig . [
fig : dss ] ) .
the source catalog , including selected source parameters , is presented in table [ source_prop6946 ] .
we choose to give photon fluxes for each source in each observation , as opposed to counts or count rates , since the latter two quantities are not directly comparable between different sources and observations due to variations in exposure and effective area .
an adaptively smoothed image based on the co - added _ chandra _ exposure is shown in figure [ fig : xray ] .
the ` csmooth ` task in ciao was used to create smoothed images for the @xmath22 , @xmath30 , and @xmath31 kev bands .
the same smoothing scales were used for all three images .
the scales were calculated by adaptively smoothing a full - band image with a significance between 3.5 @xmath12 and 5 @xmath12 above the local background .
each of the three images was also exposure corrected using exposure maps smoothed with the same scales as the images .
the images were then combined to create a color - coded full - band image with red representing the soft band , green the medium band , and blue the hard band .
finally , to enhance the point sources , the combined image was smoothed with a gaussian with a kernel radius of three pixels .
of the 90 sources , 65 were previously detected by @xcite and 25 were not .
@xcite also report two sources within the central region excluded from our analysis , as well as one that falls outside the extent of the galaxy as we define it .
moreover , three of their 72 detected sources fall below our significance threshold for inclusion , and , finally , one of their sources almost certainly suffers from a typographical error in its position ( sources 35 and 36 in their table 1 have the same position to within an acis pixel size ) .
we associate five of our 90 detected sources with foreground stars ( see table [ source_prop6946 ] ) . comparing our source locations to the usno - a2.0 catalogue @xcite , available online through vizier
, we find that five sources are coincident with foreground stars in the catalog to within @xmath64 .
we estimate the likelihood of a chance coincidence to be low . for a crude test we shifted the source positions northeast by @xmath65 .
this resulted in no matches with sources in the catalog to within @xmath66 .
positions of sources were compared with two existing lists of snrs in ngc 6946 : a list of 27 optically identified snrs from @xcite and a list of 35 radio - selected snr candidates from @xcite .
interestingly , the two lists only have two sources in common .
we find one counterpart to a snr in the former list and five counterparts to snr candidates in the latter list , as well as a counterpart to one of the two common sources .
two of the nine historical supernovae in the galaxy took place within the timespan covered by the _ chandra _ observations , and they are both detected .
in addition to these , two of the older historical supernovae are detected .
we estimate how many of our detected sources can be expected to be cosmic background sources ( cbss ; primarily active galactic nuclei and star - forming galaxies ) by using the results of @xcite from the _ chandra _ deep fields .
they give formulae estimating the number of cbss per square degree of the sky above a given limiting flux in two energy bands , @xmath67 and @xmath68 kev .
performing source detection on a merged @xmath67 kev image from all five observations , 75 of our 85 nonforeground sources were detected with a ` wavdetect ` significance level above 5 @xmath12 . in the @xmath68 kev band ,
53 of the 85 sources were detected , five of which were not detected in the soft band . since the total exposure varies widely over the extent of the galaxy , so does the limiting count rate for detection . for our soft band detections , we estimate that a typical limiting count rate is @xmath69 in the @xmath13 kev band , and @xmath70 for our hard band detections .
we convert these count rates to energy fluxes in the soft and hard bands , respectively , assuming a power law with a spectral index of 1.4 ( as done by * ? ? ?
we take into account absorption both in our own galaxy and in ngc 6946 . using the results of @xcite
, we estimate that a typical absorption column through ngc 6946 is @xmath71 .
adding galactic absorption , this yields a total absorption column for cbss of @xmath72 . using our derived soft and hard band flux limits in the formulae of @xcite
, we get 22.4 sources in the soft band over the angular area of the @xmath20 isophote of ngc 6946 ( corresponding to 30% of our 75 soft band detections ) , and 23.6 sources in the hard band , or 45% of our 53 detections .
overall , we estimate that @xmath73 of our total 85 detections in the @xmath13 kev band can be expected to arise from cbss .
the _ rosat _ hri and _ xmm - newton _
epic observations of ngc 6946 are summarized in tables [ obs6946rosat ] and [ obs6946xmm ] .
a total of 11 _ xmm - newton _
epic observations have been made , all of which suffer from periods of high background .
six of them are completely unusable and were therefore not analyzed further , and they are not included in table [ obs6946xmm ] .
ngc 4485 and ngc 4490 are a closely interacting pair of galaxies , variously classified as spiral or irregular systems ( see fig .
[ fig : dss ] ) . in @xcite
ngc 4490 is classified as type sb(s)d , and the smaller companion ngc 4485 as type ib(s)m .
both galaxies show signs of tidal disruption .
@xcite gives separate distances of 7.8 mpc ( ngc 4490 ) and 9.3 mpc ( ngc 4485 ) to the two galaxies , assuming a hubble constant value of @xmath55 .
these distances are adopted by @xcite and @xcite .
such disparate distances are , however , unphysical for such a closely interacting system .
@xcite , @xcite , and @xcite therefore assume a distance of 7.8 mpc to both galaxies .
a consistent distance of 8 mpc to the pair is adopted by @xcite , @xcite , @xcite , and @xcite , based on the suspected membership of the system in the cvn ii cloud @xcite , and this is the value we used .
the galactic absorption column in the direction of ngc 4485/4490 is @xmath74 @xcite .
[ fig:6946_xray ] [ fig:4485_xray ] lcccc 2000 nov 3 & acis - s & 1579 & 19.52 & f + 2004 jul 27 & acis - s & 4725 & 38.41 & vf + 2004 nov 20 & acis - s & 4726 & 39.59 & vf + ngc 4485/90 has been studied at radio , infrared , and optical wavelengths and shows evidence for enhanced star formation @xcite .
@xcite estimate a star formation rate in ngc 4490 of @xmath75 and a dynamical mass of @xmath76 within 8 kpc ( the companion ngc 4485 being less massive by about a factor of 8) .
the system was first observed in x - rays with the _ rosat _ pspc in 1991 . analyzing those data
, @xcite detected four point - like sources , three in ngc 4490 and one in ngc 4485 .
the galaxy pair was subsequently observed three times with the _ rosat _ hri in 1995 and 1996 . analyzing two of those observations
, @xcite only found one additional point source , an apparently transient source coincident with ngc 4490 .
@xcite analyzed all three _ rosat _ observations as part of their study of ulxs using the _ rosat _ hri archive .
they found the same sources as @xcite .
most recently , the system was observed three times with _
chandra _ between 2000 and 2004 and once with _ xmm - newton _ in 2002 .
analyzing the first of the _ chandra _ observations , @xcite found 31 point sources , 29 of those being coincident with ngc 4490 , one coincident with ngc 4485 , and one source located in a bridge between the galaxies .
the _ chandra _ observations of the ngc 4485/90 system are summarized in table [ obs4485chandra ] .
these are all acis - s full - frame imaging observations taken in timed exposure mode .
the frame time was in all cases set at the standard value of 3.2 s , and events were telemetered in either the faint or very faint formats .
the observations span a period of about 4 yr and have a cumulative good live exposure time of 97.5 ks . in all three observations all the sources coincident with ngc 4485/90 fall within the s3 chip .
the data were analyzed as described in
[ sec : analysis ] .
source detection was performed on a @xmath77 rectangle encompassing the system .
a total of 38 reliable source detections down to a luminosity of @xmath1 were found to be coincident with the two galaxies . of these , 33 are coincident with ngc 4490 , three coincident with ngc 4485 , and two more are located in the bridge between the galaxies ( see fig .
[ fig : dss ] ) .
the source catalog is presented in table [ source_prop4485 ] . an adaptively smoothed color - coded image based on the co - added _ chandra _
exposure is shown in figure [ fig : xray ] .
of the 38 sources , 27 had previously been detected by @xcite and 11 had not been seen before .
in addition , four of the sources reported in @xcite fell below our threshold of detection significance and are therefore not included in our tables . comparing our source locations to the usno - a2.0 catalogue @xcite
, we find no close matches between the locations of foreground stars and our detected sources .
there have been two historical supernovae in ngc 4490 : sn 1982f , which is not detected , and recently sn 2008ax .
lcc[b ! ] 1996 jun 18 & rh600855n00 & 26.70 + 1996 nov 26 & rh600855a01 & 24.58 + we estimate how many of our detections can be expected to arise from cbss using the same procedure as the one described for ngc 6946 in
[ sec : ngc6946 ] .
we use the results of @xcite to estimate that a typical absorption column through the ngc 4485/90 system is @xmath78 , giving a total absorption column of @xmath79 for cbss behind ngc 4485/90 . applying the formulae of @xcite
then yields an estimate of @xmath80 for the expected number of cbss among our 38 detected sources . for comparison , we note that @xcite , using the results of @xcite , attribute @xmath81 of their detections to cbss . the _ rosat _ hri and _ xmm - newton
_ epic observations of ngc 4485/90 are summarized in tables [ obs4485rosat ] and [ obs4485xmm ] .
the first of the _ rosat _ observations is too short to be of significant use to this study , and it is not included in table [ obs4485rosat ] .
lccccccccc 2002 may 27 & 0112280201 & 16.34 & ff - me & & 16.37 & ff - me & & 11.17 & eff - me +
the variability properties of the sources in ngc 6946 are summarized in table [ var_prop6946 ] .
based on the source flux variability parameter @xmath82 defined in
[ sec : variability ] we classify 25 of the 85 nonforeground sources , or 29% , as variable on timescales of weeks , months , or years . of these
, 17 sources are seen to vary by more than a factor of 2 , and we classify 11 sources ( or 13% of the 85 ) as transient candidates ( two of those are recent supernovae ) .
evidence for short - term ( hours ) variability , as defined in
[ sec : variability ] , is seen in four , or 5% , of the 85 sources .
two of those also show long - term variability .
unsurprisingly , none of the sources associated with snrs from the lists of @xcite and @xcite shows variability of any kind above a level of 2 @xmath12 , with the exception of the ulx ( see [ sec : ulx_6946 ] ) .
the variability properties of the sources in ngc 4485/90 are summarized in table [ var_prop4485 ] .
we classify 15 of the 38 sources , or 39% , as showing variability on timescales of months or years . of these
, 10 sources vary by more than a factor of 2 , and four sources ( or 11% of the 38 ) are transient candidates .
evidence for short - term variability is seen in four of the 38 sources .
all of those also show long - term variability . due to the limited number of counts within each observation for most of the sources showing short - term variability , it is very difficult ( with one marginal exception ) to draw any concrete conclusions about the nature of the variability ( e.g. , to distinguish between burst - like behavior and more continuous decline or rise in count rate ) .
the brightest of these sources , cxou j203500.1 + 600908 in ngc 6946 , seems in the first _
observation to show a burst - like increase in count rate on a timescale of @xmath83 ks superposed on a steady decline in count rate throughout the whole observation . where possible , the baseline of the long - term light curves of the ulxs was extended by adding data points from _ rosat _ hri and _ xmm - newton _ observations as described in
[ sec : rosat_xmm ] .
long - term light curves of the 10 brightest sources in ngc 6946 are shown in figures [ fig : lc_203500.74 + 601130.5 ] and [ fig : lc_6946 ] , and of the 10 brightest sources in ngc 4485/90 in figures [ fig : lc_4485a ] and [ fig : lc_4485b ] .
we emphasize that the error bars in the light curves contain only the 1 @xmath12 poisson errors in the counts .
the _ rosat _ and _ xmm - newton _ data are generally of much lower quality than the _ chandra _ data , and comparisons of luminosities between different instruments should be regarded with some caution .
various factors ( e.g. , cross - calibration issues and uncertainties in instrument responses and in assumed spectral shapes for energy conversions ) are likely to add to the errors . to give some idea of the magnitude of additional errors we note that the _ chandra _ team estimates that present 1 @xmath12 calibration uncertainties for acis are 0.3% in the absolute energy scale , 5% in the absolute effective area , and a few percent for the relative efficiency ( i.e. , for variations in quantum efficiency over the area of the detector ) . when choosing a spectral model to use for calculating ulx fluxes ( as described in
[ sec : light_curves ] ) the scatter in flux values between several well - fitting models was typically of the order of a few percent
. however , sometimes none of the eight simple spectral models considered provided a particularly good fit , in which case the errors added are likely greater . as for cross - calibration uncertainties between _ chandra _ and _ xmm - newton _
, we expect that the relative cross - calibration error between our _ chandra _ and _ xmm - newton _ fluxes should likely be @xmath84 ( h. marshall , private communication , 2008 ) .
the observed fractions here of @xmath85 of sources showing flux variability is similar to what has been found in previous studies of late - type galaxies .
@xcite survey 11 nearby face - on spiral galaxies with two _ chandra _ observations each and find that at least 27% of the total number of sources exhibit variability on either long or short timescales .
@xcite analyze three observations of m33 and find that 25% of the detected sources exhibit long - term variability .
they also classify 17% of their sources as candidate transients , similar to what is seen in our galaxies .
overall , the abundant variability observed among the sources in ngc 6946 and ngc 4485/90 clearly points to source populations dominated by accreting xrbs .
it is interesting to compare the number of transient candidates seen in ngc 6946 and ngc 4485/90 to the number in our own galaxy .
inspection of the data archives of the _ rossi x - ray timing explorer s all - sky monitor _
( rxte asm ) reveals that on average @xmath86 transients ( including recurrent ones ) with luminosities above @xmath87 go off every year in the galaxy ( r. remillard , private communication , 2008 ) .
we see 11 transient candidates in ngc 6946 and four in ngc 4485/90 ( all with maximum luminosities above @xmath88 ) . due to the close monitoring of our galaxy by the asm
, we would expect it to reveal a larger fraction of the actual number of transients among the x - ray sources than a handful of _ chandra _ observations do in ngc 6946 and ngc 4485/90 .
however , a large fraction of the transient population of the milky way is presumably obscured by gas and dust . to put these numbers in context ,
we compare the estimated masses and star formation rates of these systems .
a typical quoted value for the dynamical mass of the milky way within @xmath89 kpc is a few times @xmath90 ( see , e.g. , * ? ? ?
* ) , and a recent estimate of its star formation rate is @xmath53 @xcite .
corresponding numbers for ngc 6946 and ngc 4485/90 are given in
[ sec : ngc6946 ] and [ sec : ngc4485 ] . in short ,
the three systems are estimated to have similar star formation rates , and the milky way and ngc 6946 have comparable estimates for their mass , whereas ngc 4485/90 seems to be about an order of magnitude less massive . as mentioned in
[ sec : ngc6946 ] , ngc 6946 has been host to nine historical supernovae , the most recent one being sn 2008s .
the last three of the five _ chandra _ observations were aimed at type iip sn 2004et and observed it 30 , 45 , and 72 days after the explosion ( see light curve in fig .
[ fig : lc_6946 ] ) . a multiwavelength study of this object in x - ray , optical , and radio
is presented in @xcite .
study of sn 2004et is presented in @xcite .
the second _ chandra _ observation of ngc 6946 , on 2002 november 25 , was aimed at type iip sn 2002hh and observed it @xmath73 days after the explosion , whose time is constrained to have been between 2002 october 26.1 and 31.1 ut @xcite .
the measured luminosity at our adopted distance was @xmath91 , with hardness ratios of @xmath92 and @xmath93 .
no source was detected at the position of the supernova in the first _
observation in 2001 september . in the last three _
chandra _ observations , made around two years after the explosion , the source is seen to be emitting at a luminosity of @xmath94 ( see light curve in fig . [
fig : lc_6946 ] ) .
there is not a measurable difference in the soft color compared to the 2002 observation .
however , the hard color exhibits significant softening between the 2002 and 2004 observations , being measured at @xmath95 , @xmath96 , and @xmath97 in 2004 .
we note that sn 2004et exhibits similar spectral behavior , the soft color remaining constant but the hard color decreasing ( see entries in tables [ spec_prop6946a ] and [ spec_prop6946b ] ) , although the hard color is never seen to be as high as in the first observation of sn 2002hh .
this is in agreement with the finding of @xcite for sn 2004et that , while the @xmath68 kev flux decays , the @xmath67 kev flux remains roughly constant .
type iil sn 1980k fell outside the detector in the first _
chandra _ observation but is seen to be emitting steadily at @xmath94 in the other four observations .
we note that it was observed by the _
ipc 35 and 82 days after maximum light ( which was @xmath98 days after the explosion ) . from those observations
@xcite report absorption corrected luminosities of @xmath99 and @xmath100 in the @xmath101 kev band ( after scaling their values to our adopted distance ) .
sn 1980k was also observed by the _ rosat _
pspc in 1992 june .
@xcite gives a luminosity estimate of @xmath102 in the @xmath67 kev range ( scaled to our adopted distance ) .
type ii sn 1968d is seen in all five _ chandra _ observations and has a luminosity of @xmath102 .
the other four historical supernovae ( sn 1917a , sn 1939c , sn 1948b , and sn 1969p ) are not detected .
the only ultraluminous x - ray source in ngc 6946 , often called mf16 @xcite , has been extensively studied .
it was first detected with the _ einstein _ ipc in 19791981 @xcite and was later observed with the _ rosat _ pspc in 1992 . analyzing the _ rosat _ observation , @xcite derived an x - ray luminosity of @xmath103 in the @xmath67 kev band ( scaled to our adopted distance ) , and identified the source as a supernova remnant based on an optical counterpart discovered by @xcite .
the remnant appears to be quite evolved , perhaps @xmath104 yr old @xcite . _
hubble space telescope _
images presented by @xcite reveal a three loop structure with an extent of @xmath105 ( corresponding to @xmath106 pc at our adopted distance ) .
various models have been put forward to try to explain this unusually high luminosity from such an evolved snr .
@xcite propose that ejecta from a young supernova might be interacting with the dense shell of an older snr ; @xcite suggest instead that a single supernova blast wave is interacting with a dense circumstellar nebula from a high - mass progenitor and that a hard spectral component in the _ rosat _ pspc data comes from a pulsar wind nebula .
@xcite , analyzing the first _
observation , argue from spectral considerations that neither of these scenarios can explain the data . @xcite
analyze the first two _ chandra _ observations of the object and speculate that the luminosity of mf16 arises from a black hole x - ray binary ( bhxrb ) within the snr .
they argue this based on ( 1 ) the point - like nature of the x - ray source , ( 2 ) the close resemblance of the source spectrum to spectra from other ulxs thought to be bhxrbs , ( 3 ) the short - timescale ( minutes ) variability of the hard component ( @xmath107 kev ) of the flux , and ( 4 ) the long - timescale drop of @xmath108 in count rate between the first and second _ chandra _ observations ( @xmath109 thereof stemming from a degradation in effective area ) .
our k - s test does not detect short - timescale variations in the full @xmath13 kev band for any of the _ chandra _ observations .
we do , however , see definite long - term variability in the _ chandra _ data ( see light curve in fig . [
fig : lc_203500.74 + 601130.5 ] ) .
there is a drop in luminosity of @xmath110 between the first and second observations , and then a much larger drop between the second and third observations ; the luminosity in the third observation is @xmath111 of the original first observation value .
two weeks later , in the fourth _ chandra _ observation , the luminosity has again increased to @xmath112 of its original value .
we also see clear evidence of variability in the _ xmm - newton _ data , most notably a @xmath113 jump in luminosity in the 12 days between two observations in 2004 june . finally , we see variations in the soft color with a significance of 2.9 @xmath12 ( see fig . [
fig : color - lum ] ) . these spectral variations do not seem to have a clear correlation with flux .
there is spectral softening associated with the drop in flux between the second and third _ chandra _ observations , but there is no detectable hardening when the flux subsequently increases again .
overall , this observed variability considerably strengthens the case for an accreting object as the source of at least part of the x - ray emission from mf16 .
the ngc 4485/90 system contains a high number of ulxs , a total of eight . in a comprehensive survey of nearby galaxies observed with the _ rosat _ hri , ( * ? ? ?
* see also @xcite ) find an average rate of @xmath114 ulxs per galaxy .
the rate is higher when only considering late - type galaxies and still higher for starburst galaxies ( @xmath115 ) .
considerably higher numbers , however , are not unusual for interacting galaxies ( see , e.g. , * ? ? ?
* ) , an extreme case being the antennae , with 14 known ulxs @xcite .
seven of the ulxs in the ngc 4485/90 system are in ngc 4490 and one is in ngc 4485 .
we numerate them here in order of right ascension ( see table [ source_prop4485 ] ) .
two of them , ulx-5 and ulx-7 , have not been identified as being ultraluminous before , their luminosities having entered the ulx range only in the most recent _ chandra _ observation .
the maximum luminosities of the ulxs range from @xmath116 to @xmath117 .
these sources show a lot of variability .
long - term flux variability above the 3 @xmath12 level is exhibited by all but one of them ( see light curves in figs . [
fig : lc_4485a ] and [ fig : lc_4485b ] ) .
ulx-7 is a transient , ulx-5 varies by a factor of @xmath118 and ulx-6 by a factor of @xmath80 .
the others vary by less than a factor of 2 .
ulx-5 shows short - term flux variability .
two sources , ulx-6 and ulx-8 , have significant variations in their hard color . in both cases
, there is a clear transition from a harder higher luminosity state to a softer lower luminosity one and back ( see fig .
[ fig : color - lum ] ) .
@xcite use the infrared spectrograph on the _ spitzer space telescope _ to do mid - infrared spectral diagnostics on six of the ulxs ( all except ulx-5 and ulx-7 ) .
they find that five of those show ionization features that they claim indicate accretion , whereas the sixth source , ulx-1 , seems to be more consistent with a snr .
as pointed out by @xcite , ulx-1 is coincident with a radio source , first j123029.4 + 413927 , which also points towards a snr origin for ulx-1 .
however , we see a @xmath119 drop in luminosity in the four months between the second and third _ chandra _ observations behavior that is more typical of an accreting xrb than a snr . this is reminiscent of the ulx in ngc 6946 , which has also been associated with a snr but
whose x - ray variability indicates an accreting xrb .
the abundance of variability among the ulxs in ngc 4485/90 is similar to what is seen in the antennae .
@xcite find that two of the ulxs in the antennae show variability on a timescale of a few days .
twelve out of 14 ulxs show long - term variability , one being a transient , and four of the ulxs occasionally have luminosities below the ulx limit . in ngc 4485/90
we find that four of the eight ulxs have a luminosity below @xmath120 in at least one of the three _ chandra _ observations .
it is therefore clear that a single observation of a galaxy is by no means guaranteed to discover the galaxy s full ulx population .
surveys such as that of @xcite , where many of the galaxies are only observed once or twice , can only give a lower limit to the average number of sources in a galaxy with the potential of being ultraluminous at one time or other .
the hardness ratios of the sources in ngc 6946 , both for each observation and for the co - added exposure , are given in tables [ spec_prop6946a ] and [ spec_prop6946b ] .
the hardness ratios for the ngc 4485/90 sources are similarly given in table [ spec_prop4485 ] .
significance parameters for variability in the hardness ratios between observations are given in tables [ var_prop6946 ] ( ngc 6946 ) and [ var_prop4485 ] ( ngc 4485/90 ) .
variability above 3 @xmath12 in either of the ratios is only detected in two of the sources in ngc 6946 and three of the sources in ngc 4485/90 .
in addition , the ulx in ngc 6946 shows a 2.9 @xmath12 variation in its hard color .
we note that a much larger fraction of the sources probably undergoes some spectral variation , but most of the sources have too few counts for the hardness ratios to be very sensitive to spectral changes .
color - luminosity plots for the sources exhibiting significant hardness ratio variability are shown in figure [ fig : color - lum ] .
four of these sources ( one of those being sn 2002hh ) seem to show a clear correlation between higher luminosity and increased hardness , whereas the other two show more complicated behavior .
previous studies of external galaxies have routinely found sources with spectral variability , in some cases with spectral behavior reminiscent of galactic xrbs ; flux - color transitions have also been observed in ulxs ( * ? ? ? * and references therein ) .
@xcite find signs of variability in hardness ratios of 21 of their @xmath121 detected sources in the antennae ; some of these are ulxs . as in our case , they find that the variability does not follow the same pattern for all sources .
they see sources that become harder with higher luminosity , as we do , but also sources that soften with increased luminosity .
in addition , some of the sources show no regular pattern , or no significant variation in luminosity .
two sources have especially soft spectra and are strong candidates for supersoft sources ( ssss ; see , e.g. , * ? ? ?
cxou j203426.2 + 600914 in ngc 6946 has all of its 68 source counts below 1 kev , and cxou j123034.5 + 413834 in ngc 4490 has 32 of its total 35 source counts below 1 kev , 25 counts being below 0.65 kev .
both sources have a luminosity of @xmath122 , and neither one shows evidence for variability .
color - color diagrams for the sources in ngc 6946 and ngc 4485/90 , based on the co - added exposures , are shown in figure [ fig : spec ] .
the fact that most of the sources do not show significant variation in their hardness ratios between observations provides justification for using hardness ratios calculated from the co - added exposure to describe the source populations spectral properties .
the classification scheme of @xcite discussed in [
sec : hardness_ratios ] , is overlaid . as mentioned before ,
although a source s position in a color - color diagram does not give conclusive information about the nature of the source , the overall distribution of sources should give us some information about the population as a whole .
also indicated in the color - color diagrams is the physically allowed region , covering all combinations of hr1 and hr2 possible if the net counts in all the energy bands ( s , m , h , t ; see
[ sec : hardness_ratios ] ) are non - negative . due to background subtraction in cases of very few source counts ,
some of the hardness ratios fall outside this region .
they are , however , all consistent with the physically allowed region within their 1 @xmath12 errors .
@xcite note that , while thermal snrs are in general expected to have @xmath123 , as indicated in the classification scheme , snrs whose spectra are dominated by nonthermal components will have somewhat harder spectra and are likely to be located in the lmxb or even absorbed sources part of the color - color diagram .
we see in figure [ fig : spec ] that the locations of the sources associated with snrs in ngc 6946 are in good agreement with this . for ngc 4485/90 ,
all but two of the 38 sources fall in the xrb regions of the color - color diagram , with roughly equal portions falling in the lmxb and hmxb regions . in ngc 6946
there seems to be a much larger snr component to the population .
in addition to the four detected historical supernovae and the six sources associated with optically or radio identified snrs , @xmath80 sources have colors consistent with thermal snrs .
the rest of the sources in ngc 6946 ( apart from the sss candidate ) have colors consistent with xrbs , again with roughly equal portions in the lmxb and hmxb regions .
we note that a large portion of the sources has colors consistent with both regions .
the large presence of hmxbs in these systems , indicated by the color - color diagrams , is not surprising given the high level of star formation observed .
all the ulxs have colors consistent with xrbs .
the ulx in ngc 6946 falls firmly in the lmxb region , whereas six of the ulxs in ngc 4485/90 have colors more consistent with hmxbs .
variable and transient sources are equally divided between the lmxb and hmxb regions in ngc 4485/90 . for ngc 6946 , these sources seem to slightly favor the lmxb region , but this is hardly a significant difference . the most variable sources ( those that vary by a factor of 10 or more )
are also seen to be roughly equally divided between the lmxb and hmxb regions in both galaxies .
we construct x - ray luminosity functions ( xlfs ) for each observation , as well as the co - added exposure , of both ngc 6946 and ngc 4485/90 .
we use luminosities from the full @xmath124 band . in the case of ngc 6946 , where parts of the galaxy fell outside the _ chandra _
detectors in each observation , only sources that fell within the detectors in all five observations were considered . hence , 24 of the 85 nonforeground sources were discarded .
this was done to make sure that when comparing the xlfs from different observations we are comparing the same source population . to avoid any effects of incompleteness at the low - luminosity end of the xlf
, we set a conservative completeness limit for both galaxies and consider only the portion of the xlf above that limit . to facilitate comparisons between different observations we use the same completeness limit for all observations of each galaxy .
we fit a simple power law to the unbinned distribution for each observation using the cash statistic @xcite . in order to test the null hypothesis that the xlfs of different observations of the same galaxy are derived from the same parent distribution
, we compare the xlfs in pairs using a two - sided kolmogorov - smirnov test .
cc 1043 & @xmath125 + 4404 & @xmath126 + 4631 & @xmath127 + 4632 & @xmath128 + 4633 & @xmath129 + co - added & @xmath130 + the xlfs for ngc 6946 are shown in figure [ fig : xlf ] .
we adopt a completeness limit of @xmath131 .
judging from the figure , the xlf of the galaxy does not seem to vary much between observations .
this is borne out by the results of the pairwise k - s tests between the five individual observations , which yield significance levels ranging from 0.52 to 0.99 and thus do not warrant rejection of the null hypothesis of a common parent distribution .
interestingly , performing the k - s test between the co - added exposure and each of the observations , we get lower significance levels , ranging from 0.13 to 0.77 .
the best - fit cumulative slopes of the xlfs are shown in table [ xlf6946 ] .
they are in all cases @xmath2 and are in good agreement with each other .
we note that @xcite derive a slope of @xmath132 for the first _
observation , in agreement with our value of @xmath125 .
the xlfs for ngc 4485/90 are shown in figure [ fig : xlf ] . here
we adopt a completeness limit of @xmath133 .
again , there does not seem to be significant variation between individual observations .
this is supported by pairwise k - s tests between the three observations , which yield significance levels from 0.58 to 0.97 . performing the k - s test between the co - added exposure and individual observations yields higher significance levels here ( between 0.93 and 0.98 ) , in contrast to ngc 6946 .
the best - fit cumulative slopes of the xlfs shown in table [ xlf4485 ] agree well with each other and are all @xmath134 .
we note that @xcite derive a slope of @xmath135 for the first _
observation , in agreement with our value of @xmath136 .
there is an indication in figure [ fig : xlf ] of a high - luminosity break or cutoff in the xlfs , which steepen considerably for the @xmath137 most luminous sources .
we therefore also tried fitting a broken power law to the xlfs .
this typically led to a best - fit break luminosity of @xmath138 and a slope for the low - luminosity portion that is @xmath139 lower than for the simple unbroken power law ( i.e. , @xmath3 ) .
the slope of the high - luminosity portion was , however , in all cases very poorly constrained .
we use the results of @xcite to assess whether cbss are likely to have a significant effect on the xlfs . in ngc 6946
we expect @xmath140 cbss among the @xmath141 sources typically above the completeness limit and within the region covered by all five _ chandra _ observations . in ngc 4485/90
we estimate that @xmath137 of the @xmath142 sources typically above the completeness limit can be expected to be cbss . based on these numbers alone
, one might expect that cbss could have a significant effect in the case of ngc 6946 but probably only a small effect for ngc 4485/90 . as it turns out ,
the cumulative number - flux slopes for cbss derived in @xcite are quite similar to the ones for the xlfs : 0.55 for the @xmath67 kev band and 0.56 for the @xmath68 kev band .
contamination of our sample by cbss is therefore likely to have only a small effect in both cases .
the true slopes for ngc 4485/90 are probably slightly flatter than those derived above and slightly steeper for ngc 6946 . however , the change in xlf slopes due to cbss is almost certainly well within the uncertainties quoted in tables [ xlf6946 ] and [ xlf4485 ] .
our results agree well with previous results for other galaxies . as mentioned in
[ sec : intro ] , xlfs have been seen to be remarkably stable in spite of variability among individual sources , and ngc 6946 and ngc 4485/90 do not seem to be an exception .
single - observation snapshots of their xlfs therefore seem to represent rather well their shape averaged over longer times .
their cumulative xlf slopes fall firmly in the regime of starburst galaxies , whose slopes are typically between 0.4 and 0.8 @xcite another clear indication of the young stellar populations in these two systems .
cc 1579 & @xmath136 + 4725 & @xmath143 + 4726 & @xmath144 + co - added & @xmath145 +
we have presented a study of the x - ray source populations in the spiral galaxy ngc 6946 and the irregular / spiral interacting galaxy pair ngc 4485/4490 .
we analyzed data from five _ chandra _ observations of ngc 6946 and from three _ chandra _ observations of ngc 4485/90 .
we detect 90 point sources coincident with ngc 6946 ( excluding a small circumnuclear region ) down to luminosities of a few times @xmath0 .
we associate five of these with foreground stars .
seven sources are coincident with optically or radio identified snrs , and four are supernovae from the past 40 years .
we detect 38 sources coincident with ngc 4485/90 down to a luminosity of @xmath1 . in ngc 6946 , 25 of the 85 sources exhibit long - term ( weeks to years ) variability in luminosity
; 11 are transient candidates .
we detect short - term ( hours ) variability in four sources .
three sources show significant variability in their hardness ratios .
we find that the single ulx in ngc 6946 , the snr known as mf16 , exhibits long - term variability in flux and color that strongly supports speculations of its high luminosity arising from an xrb within the remnant . in ngc 4485/90 , 15 of the 38 sources
show flux variability on timescales of months to years ; four are transient candidates .
short - term variability is seen for four sources , and three show variability in their hardness ratios .
ngc 4485/90 contains eight ulxs .
two of these have not been identified as ulxs before ; this shows the value of long - term monitoring in identifying ultraluminous sources .
all but one of the ulxs exhibit long - term variability , and one of them is a transient .
one of the ulxs , which has been associated with a snr based on spectral characteristics and its spatial coincidence with a radio source , shows clear long - term variability pointing to an accreting xrb .
we detect short - term variability for one ulx , and two have significant variations in color .
this abundant variability among the ulxs clearly points to accretion as the source of luminosity .
consistent with the widespread variability seen within the source populations , the x - ray colors of the sources indicate that the populations are dominated by xrbs .
the source population of ngc 6946 also seems to contain a sizable snr component , perhaps @xmath146 of the detected sources .
the distribution of colors among the sources in both ngc 6946 and ngc 4485/90 indicates roughly equal fractions of lmxbs and hmxbs .
such a significant presence of hmxbs is consistent with the fact that both systems show high star formation activity .
the shape of the xlfs of ngc 6946 and ngc 4485/90 does not change significantly between observations , and seems to be well represented by single - observation snapshots .
the xlf of ngc 6946 ( above @xmath131 ) can be described by a power law with cumulative slope @xmath2 .
the xlf of ngc 4485/90 can be described by a power law with cumulative slope @xmath3 between @xmath147 and @xmath148 . above @xmath149 ,
the luminosity function drops off rather sharply .
the flatness of the slopes ( especially for ngc 4485/90 ) is another testament to the systems high star formation rate and young stellar population .
this work was supported by _
award ar6 - 7009x .
use was made of data obtained from the high energy astrophysics science archive research center ( heasarc ) , provided by nasa s goddard space flight center .
we thank the referee for thorough reading of the paper and constructive comments .
broos , p. s. , townsley , l. k. , getman , k. , & bauer , f. e. 2002 , acis extract , an acis point source extraction package ( university park : pennsylvania state univ . ) , http://www.astro.psu.edu/xray/docs/tara/ae_users_guide.html sivakoff , g. r. , jordan , a. , juett , a. m. , sarazin , c. l. , & irwin , j. m. 2005a , in iau symp .
230 , populations of high energy sources in galaxies , ed .
e. j. a. meurs & g. fabbiano ( cambridge : cambridge univ . press ) , 210 zimmermann , u. , boese , g. , becker , w. , belloni , t. , doebereiner , s. , izzo , c. , kahabka , p. , & schwentker , o. 1998 , exsas user s guide ( garching : _ rosat _ scientific data center ) , http://www.mpe.mpg.de/xray/wave/rosat/doc/users-guide/exsasguide.php ccccccccccccccl 1 & 203419.0 + 601007 & 308.57942 & 60.16884 & 7.2 & @xmath150 & @xmath151 & & & @xmath152 & @xmath153 & & 2.2 & 2.5 & + 2 & 203421.4 + 601017 & 308.58923 & 60.17162 & 7.6 & @xmath154 & @xmath155 & & & @xmath156 & @xmath157 & & 5.5 & 8.1 & + 3 & 203423.4 + 601039 & 308.59770 & 60.17773 & 4.8 & @xmath158 & @xmath159 & & & @xmath160 & @xmath161 & & 1.1 & 3.4 & t , n + 4 & 203424.7 + 601038 & 308.60311 & 60.17737 & 7.9 & @xmath162 & @xmath163 & & & @xmath164 & @xmath165 & & 6.6 & 8.4 & + 5 & 203425.0 + 600906 & 308.60426 & 60.15171 & 13.3 & @xmath166 & @xmath167 & & & @xmath168 & @xmath169 & & 7.1 & 12 & t + 6 & 203426.1 + 600909 & 308.60889 & 60.15262 & 32.1 & @xmath170 & @xmath171 & & & @xmath172 & @xmath173 & & 32 & 44 & + 7 & 203426.2 + 600914 & 308.60940 & 60.15414 & 7.1 & @xmath174 & @xmath175 & & & @xmath176 & @xmath177 & & 0.9 & 1.0 & sss + 8 & 203429.1 + 601051 & 308.62137 & 60.18085 & 6.5 & @xmath178 & @xmath179 & & & @xmath180 & @xmath181 & & 2.2 & 4.8 & t + 9 & 203431.5 + 600638 & 308.63130 & 60.11069 & 3.1 & @xmath182 & @xmath183 & & & & @xmath184 & & 1.1 & 1.6 & + 10 & 203431.5 + 601056 & 308.63156 & 60.18237 & 5.8 & @xmath185 & @xmath186 & & & @xmath187 & @xmath188 & & 1.2 & 6.1 & t , n + 11 & 203432.0 + 601207 & 308.63347 & 60.20214 & 4.1 & @xmath189 & @xmath190 & & @xmath191 & @xmath192 & @xmath193 & & 0.9 & 3.2 & n + 12 & 203432.4 + 600821 & 308.63541 & 60.13942 & 5.8 & @xmath194 & @xmath195 & & & @xmath196 & @xmath197 & & 2.3 & 4.8 & t + 13 & 203434.4 + 601032 & 308.64362 & 60.17578 & 15.5 & @xmath198 & @xmath199 & & @xmath200 & @xmath201 & @xmath202 & & 4.7 & 7.0 & s + 14 & 203435.3 + 600742 & 308.64720 & 60.12853 & 4.6 & @xmath203 & @xmath204 & & & @xmath205 & @xmath206 & & 0.4 & 0.7 & + 15 & 203436.1 + 600839 & 308.65064 & 60.14428 & 7.8 & @xmath207 & @xmath208 & & @xmath209 & @xmath210 & @xmath211 & & 1.1 & 1.2 & ld + 16 & 203436.4 + 600558 & 308.65168 & 60.09952 & 7.0 & @xmath212 & @xmath213 & & & & @xmath214 & & 4.2 & 5.7 & + 17 & 203436.5 + 600930 & 308.65220 & 60.15844 & 46.4 & @xmath215 & @xmath216 & & @xmath217 & @xmath218 & @xmath219 & & 31 & 42 & + 18 & 203437.2 + 600954 & 308.65533 & 60.16503 & 4.2 & @xmath220 & @xmath221 & & @xmath222 & @xmath223 & @xmath224 & & 1.8 & 2.5 & + 19 & 203437.2 + 601019 & 308.65535 & 60.17200 & 3.5 & @xmath225 & @xmath226 & & @xmath227 & @xmath228 & @xmath229 & & 0.5 & 0.6 & + 20 & 203437.9 + 600434 & 308.65798 & 60.07618 & 10.7 & @xmath230 & @xmath231 & @xmath232 & & & @xmath233 & & 8.7 & 9.8 & + 21 & 203439.8 + 600822 & 308.66593 & 60.13967 & 8.4 & @xmath234 & @xmath235 & @xmath236 & @xmath237 & @xmath238 & @xmath239 & & 3.3 & 4.7 & + 22 & 203440.5 + 600946 & 308.66916 & 60.16302 & 4.2 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & @xmath244 & @xmath245 & & 1.0 & 1.9 & n + 23 & 203441.3 + 600846 & 308.67245 & 60.14620 & 5.1 & @xmath246 & @xmath247 & @xmath248 & @xmath249 & @xmath250 & @xmath251 & & 0.9 & 1.7 & ld + 24 & 203442.0 + 600529 & 308.67535 & 60.09158 & 5.1 & @xmath252 & @xmath253 & @xmath254 & @xmath255 & @xmath256 & @xmath257 & & 0.7 & 1.5 & + 25 & 203443.1 + 600652 & 308.67997 & 60.11457 & 4.1 & @xmath258 & @xmath259 & @xmath260 & @xmath261 & @xmath262 & @xmath263 & & 0.9 & 1.6 & + 26 & 203444.2 + 600719 & 308.68432 & 60.12203 & 12.6 & @xmath264 & @xmath265 & @xmath266 & @xmath267 & @xmath268 & @xmath269 & & 6.6 & 29 & sn 2002hh , t , n + 27 & 203445.9 + 601049 & 308.69159 & 60.18035 & 5.8 & @xmath270 & @xmath271 & @xmath272 & @xmath273 & @xmath274 & @xmath275 & & 0.6 & 0.8 & + 28 & 203446.3 + 601328 & 308.69310 & 60.22451 & 10.9 & @xmath276 & @xmath277 & @xmath278 & @xmath279 & @xmath280 & @xmath281 & & 5.3 & 6.8 & + 29 & 203447.1 + 600850 & 308.69626 & 60.14740 & 4.0 & @xmath282 & @xmath283 & @xmath284 & @xmath285 & @xmath286 & @xmath287 & & 0.5 & 1.8 & t + 30 & 203447.5 + 601003 & 308.69803 & 60.16751 & 8.6 & @xmath288 & @xmath289 & @xmath290 & @xmath291 & @xmath292 & @xmath293 & & 1.1 & 1.5 & + 31 & 203447.6 + 600932 & 308.69840 & 60.15905 & 13.0 & @xmath294 & @xmath295 & @xmath296 & @xmath297 & @xmath298 & @xmath299 & & 3.4 & 9.3 & + 32 & 203448.4 + 600811 & 308.70193 & 60.13651 & 22.9 & @xmath300 & @xmath301 & @xmath302 & @xmath303 & @xmath304 & @xmath305 & & 12 & 19 & + 33 & 203448.6 + 600941 & 308.70264 & 60.16155 & 4.4 & @xmath306 & @xmath307 & @xmath308 & @xmath309 & @xmath310 & @xmath311 & & 0.9 & 2.5 & t + 34 & 203448.8 + 600554 & 308.70341 & 60.09854 & 26.7 & @xmath312 & @xmath313 & @xmath314 & @xmath315 & @xmath316 & @xmath317 & & 9.5 & 11 & s + 35 & 203449.0 + 601217 & 308.70445 & 60.20483 & 24.4 & @xmath318 & @xmath319 & @xmath320 & @xmath321 & @xmath322 & @xmath323 & & 14 & 16 & + 36 & 203449.4 + 601342 & 308.70621 & 60.22846 & 5.6 & @xmath324 & @xmath325 & @xmath326 & @xmath327 & @xmath328 & @xmath329 & & 1.9 & 2.6 & + 37 & 203449.7 + 601011 & 308.70717 & 60.16990 & 7.1 & @xmath330 & @xmath331 & @xmath332 & @xmath333 & @xmath334 & @xmath335 & & 1.8 & 2.8 & + 38 & 203449.7 + 600806 & 308.70736 & 60.13504 & 4.6 & @xmath336 & @xmath337 & @xmath338 & @xmath339 & @xmath340 & @xmath341 & & 0.4 & 0.7 & ld + 39 & 203449.7 + 601020 & 308.70745 & 60.17228 & 5.5 & @xmath342 & @xmath343 & @xmath344 & @xmath345 & @xmath340 & @xmath346 & & 1.1 & 2.2 & + 40 & 203450.0 + 600924 & 308.70851 & 60.15685 & 7.9 & @xmath347 & @xmath348 & @xmath349 & @xmath350 & @xmath168 & @xmath351 & & 2.9 & 3.6 & + 41 & 203450.4 + 601016 & 308.71030 & 60.17112 & 5.1 & @xmath352 & @xmath353 & @xmath354 & @xmath355 & @xmath356 & @xmath251 & & 1.1 & 2.2 & + 42 & 203450.4 + 600724 & 308.71034 & 60.12337 & 13.5 & @xmath357 & @xmath358 & @xmath359 & @xmath360 & @xmath361 & @xmath362 & & 4.9 & 8.7 & t , n + 43 & 203450.7 + 600747 & 308.71149 & 60.12997 & 5.0 & @xmath363 & @xmath364 & @xmath365 & @xmath366 & @xmath367 & @xmath368 & & 0.4 & 0.6 & + 44 & 203450.7 + 601207 & 308.71165 & 60.20214 & 4.9 & @xmath369 & @xmath370 & @xmath371 & @xmath372 & @xmath373 & @xmath374 & & 1.0 & 2.4 & + 45 & 203450.9 + 601020 & 308.71215 & 60.17243 & 10.3 & @xmath375 & @xmath376 & @xmath377 & @xmath378 & @xmath379 & @xmath380 & & 1.3 & 1.6 & ld + 46 & 203451.5 + 601248 & 308.71462 & 60.21335 & 5.4 & @xmath381 & @xmath382 & @xmath383 & @xmath384 & @xmath385 & @xmath386 & & 1.5 & 2.5 & + 47 & 203451.5 + 600909 & 308.71470 & 60.15253 & 4.1 & @xmath387 & @xmath388 & @xmath223 & @xmath389 & @xmath390 & @xmath391 & & 0.4 & 0.7 & mf , n + 48 & 203453.3 + 601055 & 308.72237 & 60.18207 & 6.3 & @xmath392 & @xmath393 & @xmath394 & @xmath395 & @xmath334 & @xmath396 & & 0.4 & 0.7 & + 49 & 203453.4 + 600719 & 308.72259 & 60.12208 & 8.5 & @xmath397 & @xmath398 & @xmath399 & @xmath400 & @xmath401 & @xmath402 & & 3.2 & 13 & + 50 & 203453.7 + 600910 & 308.72416 & 60.15286 & 7.2 & @xmath403 & @xmath404 & @xmath405 & @xmath406 & @xmath407 & @xmath408 & & 1.8 & 2.2 & n + 51 & 203455.6 + 600930 & 308.73203 & 60.15856 & 7.4 & @xmath409 & @xmath410 & @xmath411 & @xmath412 & @xmath413 & @xmath414 & & 1.2 & 2.7 & + 52 & 203456.5 + 600834 & 308.73547 & 60.14281 & 30.6 & @xmath415 & @xmath416 & @xmath417 & @xmath418 & @xmath419 & @xmath420 & & 26 & 33 & + 53 & 203456.5 + 600819 & 308.73562 & 60.13879 & 7.7 & @xmath421 & @xmath422 & @xmath423 & @xmath424 & @xmath425 & @xmath426 & & 0.8 & 1.5 & + 54 & 203456.8 + 600532 & 308.73680 & 60.09235 & 10.1 & @xmath427 & @xmath428 & @xmath429 & @xmath430 & @xmath431 & @xmath432 & & 3.0 & 4.3 & + 55 & 203456.9 + 601209 & 308.73725 & 60.20274 & 4.4 & @xmath433 & @xmath434 & @xmath435 & @xmath436 & @xmath437 & @xmath438 & & 0.9 & 1.5 & s , n + 56 & 203456.9 + 600826 & 308.73727 & 60.14064 & 6.4 & @xmath439 & @xmath440 & @xmath441 & @xmath442 & @xmath443 & @xmath444 & & 1.2 & 3.8 & n + 57 & 203457.2 + 600733 & 308.73857 & 60.12598 & 3.3 & @xmath445 & @xmath446 & @xmath447 & @xmath448 & @xmath449 & @xmath450 & & 0.2 & 0.6 & s , n + 58 & 203457.7 + 600948 & 308.74058 & 60.16356 & 27.7 & @xmath451 & @xmath452 & @xmath453 & @xmath454 & @xmath455 & @xmath456 & & 13 & 18 & + 59 & 203458.4 + 600934 & 308.74341 & 60.15954 & 9.3 & @xmath457 & @xmath458 & @xmath459 & @xmath460 & @xmath461 & @xmath462 & & 1.9 & 2.4 & sn 1968d + 60 & 203500.1 + 600908 & 308.75057 & 60.15223 & 63.5 & @xmath463 & @xmath464 & @xmath465 & @xmath466 & @xmath467 & @xmath468 & & 47 & 57 & + 61 & 203500.4 + 600859 & 308.75177 & 60.14976 & 6.0 & @xmath469 & @xmath470 & @xmath471 & @xmath472 & @xmath473 & @xmath474 & & 1.4 & 2.6 & n + 62 & 203500.6 + 601250 & 308.75257 & 60.21402 & 10.8 & @xmath475 & @xmath476 & @xmath477 & @xmath478 & @xmath479 & @xmath480 & & 3.3 & 7.8 & + 63 & 203500.7 + 601130 & 308.75312 & 60.19183 & 145.3 & @xmath481 & @xmath482 & @xmath483 & @xmath484 & @xmath485 & @xmath486 & & 310 & 330 & ulx , ld , mf + 64 & 203500.9 + 601010 & 308.75403 & 60.16972 & 33.0 & @xmath487 & @xmath488 & @xmath489 & @xmath490 & @xmath491 & @xmath492 & & 26 & 32 & + 65 & 203501.1 + 600923 & 308.75482 & 60.15642 & 3.6 & @xmath493 & @xmath494 & @xmath399 & @xmath495 & @xmath496 & @xmath497 & & 0.1 & 0.4 & n + 66 & 203501.9 + 601003 & 308.75818 & 60.16768 & 20.1 & @xmath498 & @xmath499 & @xmath500 & @xmath501 & @xmath502 & @xmath503 & & 10 & 12 & + 67 & 203504.6 + 601117 & 308.76934 & 60.18821 & 11.6 & @xmath504 & @xmath505 & @xmath506 & @xmath507 & @xmath508 & @xmath509 & & 1.8 & 2.4 & + 68 & 203504.6 + 600918 & 308.76952 & 60.15512 & 5.4 & @xmath510 & @xmath511 & @xmath228 & @xmath512 & @xmath513 & @xmath514 & & 0.5 & 0.6 & + 69 & 203505.3 + 600623 & 308.77223 & 60.10640 & 15.8 & @xmath515 & @xmath516 & @xmath517 & @xmath518 & @xmath519 & @xmath520 & & 7.1 & 7.9 & + 70 & 203505.5 + 600742 & 308.77306 & 60.12843 & 4.8 & @xmath521 & @xmath522 & @xmath523 & @xmath524 & @xmath525 & @xmath526 & & 1.0 & 1.6 & n + 71 & 203506.5 + 601040 & 308.77742 & 60.17781 & 9.1 & @xmath527 & @xmath528 & @xmath529 & @xmath530 & @xmath531 & @xmath532 & & 3.0 & 6.1 & + 72 & 203508.1 + 601113 & 308.78403 & 60.18696 & 7.4 & @xmath533 & @xmath534 & @xmath535 & @xmath536 & @xmath537 & @xmath538 & & 1.2 & 1.7 & ld + 73 & 203510.1 + 601000 & 308.79234 & 60.16692 & 5.0 & @xmath539 & @xmath540 & @xmath541 & @xmath542 & @xmath543 & @xmath544 & & 1.0 & 1.3 & n + 74 & 203510.9 + 600856 & 308.79543 & 60.14912 & 5.2 & @xmath545 & @xmath546 & @xmath547 & @xmath548 & @xmath549 & @xmath550 & & 0.3 & 0.6 & n + 75 & 203512.5 + 600837 & 308.80213 & 60.14377 & 3.7 & @xmath551 & @xmath552 & @xmath553 & @xmath554 & @xmath555 & @xmath556 & & 0.7 & 1.1 & n + 76 & 203512.7 + 600731 & 308.80321 & 60.12551 & 22.5 & @xmath557 & @xmath558 & @xmath559 & @xmath560 & @xmath322 & @xmath561 & & 12 & 17 & + 77 & 203513.1 + 600907 & 308.80461 & 60.15207 & 4.4 & @xmath562 & @xmath393 & @xmath563 & @xmath564 & @xmath565 & @xmath566 & & 1.3 & 2.3 & n + 78 & 203513.6 + 600522 & 308.80690 & 60.08950 & 6.6 & & @xmath567 & @xmath568 & @xmath569 & @xmath570 & @xmath571 & & 1.8 & 3.8 & n + 79 & 203516.2 + 601046 & 308.81773 & 60.17959 & 16.0 & @xmath572 & @xmath573 & @xmath574 & @xmath575 & @xmath576 & @xmath577 & & 8.0 & 23 & t + 80 & 203517.3 + 600813 & 308.82214 & 60.13719 & 4.7 & @xmath578 & @xmath579 & @xmath580 & @xmath581 & @xmath582 & @xmath526 & & 1.0 & 1.8 & n + 81 & 203518.7 + 601056 & 308.82830 & 60.18237 & 30.7 & @xmath583 & @xmath584 & @xmath585 & @xmath586 & @xmath587 & @xmath588 & & 19 & 24 & + 82 & 203520.0 + 600933 & 308.83336 & 60.15940 & 4.8 & @xmath589 & @xmath310 & @xmath590 & @xmath591 & @xmath592 & @xmath593 & & 0.5 & 1.0 & + 83 & 203521.1 + 600951 & 308.83813 & 60.16437 & 6.0 & @xmath594 & @xmath595 & @xmath596 & @xmath597 & @xmath598 & @xmath599 & & 0.8 & 1.1 & + 84 & 203525.3 + 600717 & 308.85565 & 60.12156 & 21.7 & & @xmath600 & @xmath601 & @xmath602 & @xmath603 & @xmath604 & & 19 & 30 & sn 2004et , t , n + 85 & 203525.5 + 600958 & 308.85651 & 60.16629 & 12.4 & @xmath605 & @xmath606 & @xmath607 & @xmath608 & @xmath609 & @xmath610 & & 2.9 & 3.8 & + 86 & 203527.8 + 600920 & 308.86604 & 60.15559 & 9.3 & @xmath611 & @xmath612 & @xmath613 & @xmath614 & @xmath615 & @xmath616 & & 3.2 & 3.7 & + 87 & 203528.6 + 600605 & 308.86937 & 60.10157 & 3.8 & & @xmath617 & @xmath618 & @xmath619 & @xmath442 & @xmath620 & & 0.7 & 1.1 & n + 88 & 203530.1 + 600623 & 308.87569 & 60.10648 & 11.7 & & @xmath621 & @xmath622 & @xmath623 & @xmath624 & @xmath625 & & 3.0 & 3.4 & sn 1980k , n + 89 & 203530.6 + 600737 & 308.87770 & 60.12714 & 6.4 & & @xmath626 & @xmath627 & @xmath628 & @xmath629 & @xmath630 & & 1.0 & 1.1 & s , n + 90 & 203532.2 + 601018 & 308.88442 & 60.17180 & 8.9 & @xmath631 & & @xmath632 & @xmath633 & @xmath634 & @xmath635 & & 3.8 & 5.2 & n + ccccccccccccl 1 & 123025.2 + 413924 & 187.60506 & 41.65691 & 15.6 & @xmath636 & @xmath637 & @xmath638 & @xmath639 & & 22 & 23 & + 2 & 123026.7 + 413821 & 187.61165 & 41.63942 & 5.2 & @xmath640 & @xmath641 & @xmath642 & @xmath643 & & 2.2 & 3.3 & t , n + 3 & 123027.1 + 413814 & 187.61330 & 41.63724 & 8.7 & @xmath644 & @xmath645 & @xmath646 & @xmath647 & & 9.3 & 10 & + 4 & 123027.6 + 413941 & 187.61528 & 41.66161 & 4.1 & @xmath648 & @xmath649 & @xmath650 & @xmath651 & & 1.5 & 2.8 & n + 5 & 123028.2 + 413958 & 187.61779 & 41.66622 & 4.2 & @xmath652 & @xmath653 & @xmath654 & @xmath655 & & 1.3 & 3.2 & + 6 & 123028.5 + 413926 & 187.61913 & 41.65731 & 5.6 & @xmath656 & @xmath657 & @xmath658 & @xmath659 & & 5.3 & 8.0 & + 7 & 123028.7 + 413757 & 187.61969 & 41.63254 & 9.5 & @xmath660 & @xmath661 & @xmath662 & @xmath663 & & 9.8 & 17 & + 8 & 123029.0 + 414046 & 187.62122 & 41.67961 & 6.9 & @xmath664 & @xmath665 & @xmath666 & @xmath667 & & 3.7 & 4.8 & + 9 & 123029.4 + 413927 & 187.62278 & 41.65771 & 31.6 & @xmath668 & @xmath669 & @xmath670 & @xmath671 & & 100 & 117 & ulx-1 + 10 & 123029.4 + 414058 & 187.62281 & 41.68293 & 3.3 & @xmath672 & @xmath673 & @xmath674 & @xmath675 & & 0.8 & 1.4 & n + 11 & 123030.3 + 413853 & 187.62630 & 41.64811 & 22.5 & @xmath676 & @xmath677 & @xmath678 & @xmath679 & & 43 & 64 & + 12 & 123030.3 + 414126 & 187.62662 & 41.69067 & 6.2 & @xmath680 & @xmath681 & @xmath682 & @xmath683 & & 3.6 & 5.8 & t , n + 13 & 123030.4 + 414142 & 187.62698 & 41.69509 & 67.8 & @xmath684 & @xmath685 & @xmath686 & @xmath687 & & 280 & 410 & ulx-2 + 14 & 123030.7 + 413911 & 187.62807 & 41.65326 & 55.8 & @xmath688 & @xmath689 & @xmath690 & @xmath691 & & 288 & 295 & ulx-3 + 15 & 123031.0 + 413837 & 187.62937 & 41.64376 & 5.6 & @xmath692 & @xmath693 & @xmath694 & @xmath695 & & 4.5 & 6.8 & + 16 & 123031.3 + 413901 & 187.63048 & 41.65053 & 6.4 & @xmath696 & @xmath697 & @xmath698 & @xmath699 & & 4.0 & 5.3 & + 17 & 123031.6 + 414140 & 187.63184 & 41.69471 & 28.2 & @xmath254 & @xmath700 & @xmath701 & @xmath702 & & 48 & 83 & t , n + 18 & 123032.1 + 413918 & 187.63411 & 41.65508 & 56.8 & @xmath703 & @xmath704 & @xmath705 & @xmath706 & & 282 & 347 & ulx-4 + 19 & 123032.8 + 413845 & 187.63667 & 41.64587 & 4.0 & @xmath707 & @xmath708 & @xmath709 & @xmath651 & & 1.0 & 1.7 & n + 20 & 123032.9 + 414014 & 187.63719 & 41.67064 & 6.8 & @xmath710 & @xmath711 & @xmath712 & @xmath713 & & 6.3 & 7.6 & n + 21 & 123034.1 + 413859 & 187.64218 & 41.64984 & 5.2 & @xmath714 & @xmath715 & @xmath716 & @xmath717 & & 3.4 & 4.7 & n + 22 & 123034.2 + 413805 & 187.64263 & 41.63484 & 6.5 & @xmath718 & @xmath719 & @xmath720 & @xmath721 & & 6.4 & 8.5 & + 23 & 123034.3 + 413850 & 187.64320 & 41.64724 & 19.1 & @xmath722 & @xmath723 & @xmath724 & @xmath725 & & 25 & 39 & + 24 & 123034.5 + 413834 & 187.64377 & 41.64290 & 4.3 & @xmath726 & @xmath727 & @xmath728 & @xmath729 & & 0.9 & 1.2 & sss , n + 25 & 123035.1 + 413846 & 187.64648 & 41.64638 & 31.7 & @xmath730 & @xmath731 & @xmath732 & @xmath733 & & 67 & 103 & ulx-5 + 26 & 123035.8 + 413832 & 187.64926 & 41.64247 & 6.0 & @xmath734 & @xmath735 & @xmath736 & @xmath737 & & 3.4 & 5.2 & + 27 & 123036.2 + 413837 & 187.65101 & 41.64388 & 47.2 & @xmath738 & @xmath739 & @xmath740 & @xmath741 & & 154 & 229 & ulx-6 + 28 & 123037.7 + 413823 & 187.65748 & 41.63991 & 3.9 & @xmath742 & @xmath743 & @xmath744 & @xmath745 & & 2.2 & 3.4 & n + 29 & 123038.2 + 413831 & 187.65926 & 41.64198 & 5.0 & @xmath746 & @xmath747 & @xmath748 & @xmath749 & & 1.6 & 2.4 & + 30 & 123038.4 + 413831 & 187.66012 & 41.64218 & 27.4 & @xmath750 & @xmath449 & @xmath751 & @xmath752 & & 59 & 139 & ulx-7 , t , n + 31 & 123038.4 + 413743 & 187.66026 & 41.62863 & 12.2 & @xmath753 & @xmath754 & @xmath755 & @xmath756 & & 9.1 & 11 & + 32 & 123038.8 + 413810 & 187.66207 & 41.63614 & 17.1 & @xmath757 & @xmath758 & @xmath759 & @xmath760 & & 28 & 45 & + 33 & 123040.3 + 413813 & 187.66806 & 41.63714 & 22.1 & @xmath761 & @xmath762 & @xmath763 & @xmath764 & & 40 & 48 & + 34 & 123043.0 + 413756 & 187.67928 & 41.63248 & 5.7 & @xmath765 & @xmath766 & @xmath767 & @xmath768 & & 1.4 & 1.8 & + 35 & 123043.1 + 413818 & 187.67985 & 41.63854 & 63.7 & @xmath769 & @xmath770 & @xmath771 & @xmath772 & & 279 & 332 & ulx-8 + 36 & 123045.5 + 413640 & 187.68984 & 41.61114 & 9.3 & @xmath773 & @xmath774 & @xmath775 & @xmath776 & & 7.1 & 9.8 & + 37 & 123047.7 + 413807 & 187.69894 & 41.63547 & 5.0 & @xmath777 & @xmath539 & @xmath778 & @xmath779 & & 2.3 & 3.5 & + 38 & 123047.7 + 413727 & 187.69906 & 41.62433 & 7.6 & @xmath780 & @xmath781 & @xmath782 & @xmath783 & & 5.0 & 9.0 & + cccccccccccccl 1 & 203419.0 + 601007 & 0.12 & 0.94 & & & 0.83 & & 2.2 & 2.4 & & 0.5 & 0.8 & + 2 & 203421.4 + 601017 & 0.96 & 0.98 & & & 0.39 & & 2.0 & 3.9 & & 0.6 & 0.2 & + 3 & 203423.4 + 601039 & & 0.66 & & & 0.13 & & 4.0 & 110 & & 0.7 & 0.5 & l + 4 & 203424.7 + 601038 & 0.65 & 0.68 & & & 0.29 & & 2.5 & 3.0 & & 0.9 & 1.4 & + 5 & 203425.0 + 600906 & 0.30 & & & & 0.15 & & 11.9 & 73.0 & & 0.8 & 0.5 & l + 6 & 203426.1 + 600909 & 0.48 & 0.69 & & & 0.77 & & 10.9 & 2.7 & & 0.6 & 1.6 & l + 7 & 203426.2 + 600914 & 0.58 & 0.47 & & & 0.49 & & 1.2 & 1.5 & & 0.4 & 0.3 & + 8 & 203429.1 + 601051 & 0.33 & 0.85 & & & & & 3.2 & @xmath784 & & 0.5 & 0.9 & l + 9 & 203431.5 + 600638 & 0.25 & & & & & & 1.9 & 7.1 & & 0.4 & 0.3 & + 10 & 203431.5 + 601056 & & & & & 0.46 & & 5.5 & 78.3 & & 0.3 & 0.7 & l + 11 & 203432.0 + 601207 & 0.15 & 0.39 & & 0.33 & & & 1.6 & @xmath785 & & 1.6 & 1.7 & + 12 & 203432.4 + 600821 & 0.79 & & & & & & 6.0 & @xmath786 & & 0.0 & 0.0 & l + 13 & 203434.4 + 601032 & 0.03 & 0.65 & & 0.82 & 0.13 & & 3.3 & 1.8 & & 1.5 & 0.9 & l , star + 14 & 203435.3 + 600742 & 0.22 & 0.44 & & & 0.17 & & 1.7 & 4.0 & & 0.3 & 0.3 & + 15 & 203436.1 + 600839 & 0.67 & 0.74 & & 0.23 & 0.55 & & 1.6 & 3.5 & & 1.1 & 0.7 & + 16 & 203436.4 + 600558 & 0.30 & 0.41 & & & & & 1.7 & 1.7 & & 0.4 & 0.4 & + 17 & 203436.5 + 600930 & 0.21 & 0.69 & & 0.93 & 0.85 & & 8.2 & 1.6 & & 3.5 & 1.7 & l , sp + 18 & 203437.2 + 600954 & 0.02 & 0.92 & & 0.16 & 0.90 & & 1.8 & 6.8 & & 1.1 & 1.0 & + 19 & 203437.2 + 601019 & 0.97 & 0.45 & & 0.75 & 0.57 & & 0.7 & 2.3 & & 1.1 & 1.0 & + 20 & 203437.9 + 600434 & 0.42 & 0.46 & 0.18 & & & & 0.7 & 1.2 & & 0.9 & 1.3 & + 21 & 203439.8 + 600822 & 0.86 & 0.95 & 0.81 & 0.43 & 0.11 & & 2.6 & 3.0 & & 1.3 & 1.1 & + 22 & 203440.5 + 600946 & 0.18 & & 0.82 & 0.87 & 0.91 & & 1.4 & 7.9 & & 1.0 & 1.1 & + 23 & 203441.3 + 600846 & 0.22 & 0.44 & 0.15 & 0.44 & 0.58 & & 1.3 & 2.6 & & 0.8 & 0.5 & + 24 & 203442.0 + 600529 & 0.07 & 0.35 & 0.37 & & 0.29 & & 1.5 & 4.8 & & 1.2 & 1.4 & + 25 & 203443.1 + 600652 & 0.34 & & 0.91 & 0.13 & & & 1.7 & 7.2 & & 0.3 & 0.6 & + 26 & 203444.2 + 600719 & & 0.29 & 0.92 & 0.60 & 0.44 & & 10.2 & 225 & & 0.6 & 3.5 & l , sp + 27 & 203445.9 + 601049 & 0.01 & 0.39 & 0.88 & 0.16 & 0.97 & & 2.1 & 5.0 & & 0.5 & 0.8 & + 28 & 203446.3 + 601328 & 0.36 & 0.02 & 0.13 & 0.15 & 0.80 & & 1.4 & 1.7 & & 1.4 & 1.5 & + 29 & 203447.1 + 600850 & 0.65 & & & & & & 4.0 & @xmath787 & & 0.0 & 0.0 & l + 30 & 203447.5 + 601003 & 0.61 & 0.97 & 0.20 & 0.73 & 0.29 & & 1.5 & 2.0 & & 2.3 & 1.1 & + 31 & 203447.6 + 600932 & 0.95 & 0.58 & 0.31 & 0.90 & 0.42 & & 8.4 & 13.6 & & 1.6 & 1.2 & l + 32 & 203448.4 + 600811 & 0.01 & 0.72 & 0.30 & 0.10 & 0.30 & & 5.7 & 3.0 & & 1.4 & 1.7 & l + 33 & 203448.6 + 600941 & 0.81 & 0.99 & 0.75 & & 0.44 & & 3.8 & @xmath787 & & 0.4 & 0.7 & l + 34 & 203448.8 + 600554 & 0.18 & @xmath788 & 0.75 & 0.01 & 0.07 & & 3.5 & 1.5 & & 1.2 & 1.5 & s , l , star + 35 & 203449.0 + 601217 & 0.85 & 0.33 & 0.47 & 0.50 & 0.50 & & 2.8 & 1.5 & & 1.9 & 1.8 & + 36 & 203449.4 + 601342 & 0.80 & 0.23 & 0.57 & @xmath789 & 0.32 & & 1.7 & 2.6 & & 1.4 & 0.9 & s + 37 & 203449.7 + 601011 & 0.46 & 0.54 & 0.66 & 0.37 & 0.12 & & 0.9 & 1.5 & & 1.0 & 1.2 & + 38 & 203449.7 + 600806 & 0.26 & 0.30 & & 0.83 & 0.28 & & 1.5 & 5.2 & & 0.7 & 0.6 & + 39 & 203449.7 + 601020 & 0.95 & 0.92 & 0.64 & 0.09 & 0.07 & & 1.5 & 2.8 & & 0.9 & 1.6 & + 40 & 203450.0 + 600924 & 0.43 & 0.34 & 0.34 & 0.87 & 0.33 & & 1.0 & 1.6 & & 1.3 & 1.1 & + 41 & 203450.4 + 601016 & 0.95 & 0.96 & 0.10 & @xmath788 & & & 1.8 & 8.5 & & 0.4 & 1.0 & s + 42 & 203450.4 + 600724 & & 0.82 & 0.34 & 0.40 & 0.99 & & 6.9 & @xmath790 & & 1.1 & 1.6 & l + 43 & 203450.7 + 600747 & 0.69 & 0.04 & 0.22 & 0.23 & 0.15 & & 0.6 & 1.8 & & 0.4 & 0.6 & + 44 & 203450.7 + 601207 & 0.70 & 0.21 & @xmath791 & 0.84 & 0.90 & & 1.5 & 8.5 & & 0.7 & 1.1 & + 45 & 203450.9 + 601020 & 0.65 & 0.05 & 0.12 & 0.83 & 0.61 & & 1.4 & 1.5 & & 1.1 & 1.5 & + 46 & 203451.5 + 601248 & 0.13 & 0.36 & 0.98 & 0.72 & 0.31 & & 1.5 & 2.9 & & 1.0 & 1.1 & + 47 & 203451.5 + 600909 & 0.13 & 0.81 & 0.12 & 0.89 & 0.77 & & 1.3 & 12.3 & & 0.6 & 1.1 & + 48 & 203453.3 + 601055 & 0.50 & 0.15 & 0.19 & 0.63 & 0.15 & & 1.2 & 2.3 & & 1.3 & 1.1 & + 49 & 203453.4 + 600719 & 0.55 & 0.26 & & 0.20 & 0.59 & & 6.1 & 25.8 & & 0.8 & 1.5 & l + 50 & 203453.7 + 600910 & 0.07 & 0.59 & 0.78 & 0.93 & 0.30 & & 0.9 & 1.7 & & 1.5 & 0.5 & + 51 & 203455.6 + 600930 & 0.30 & 0.28 & 0.41 & 0.34 & 0.11 & & 3.2 & 8.2 & & 1.4 & 0.8 & l + 52 & 203456.5 + 600834 & 0.01 & 1.00 & 0.82 & 0.50 & 0.06 & & 4.2 & 1.7 & & 1.6 & 0.8 & l + 53 & 203456.5 + 600819 & 0.01 & 0.57 & 0.83 & 0.70 & 0.38 & & 1.2 & 2.0 & & 1.5 & 1.0 & + 54 & 203456.8 + 600532 & 0.77 & 0.87 & 0.81 & 0.01 & 0.62 & & 2.0 & 2.4 & & 1.6 & 0.7 & + 55 & 203456.9 + 601209 & 0.37 & 0.25 & 0.60 & 0.23 & 0.16 & & 1.2 & 3.1 & & 1.1 & 0.7 & star + 56 & 203456.9 + 600826 & 0.08 & 0.08 & 0.69 & 0.46 & 0.02 & & 3.6 & 12.8 & & 0.4 & 1.0 & l + 57 & 203457.2 + 600733 & 0.70 & 0.06 & & & & & 2.1 & @xmath792 & & 1.0 & 1.0 & star + 58 & 203457.7 + 600948 & 0.35 & 0.10 & 0.40 & 0.88 & 0.08 & & 5.4 & 1.8 & & 1.2 & 1.0 & l + 59 & 203458.4 + 600934 & 0.19 & 0.64 & 0.01 & 0.53 & 0.83 & & 1.2 & 1.6 & & 0.5 & 1.5 & + 60 & 203500.1 + 600908 & @xmath793 & 0.24 & 0.50 & 0.76 & 0.80 & & 12.2 & 1.8 & & 1.2 & 0.9 & s , l + 61 & 203500.4 + 600859 & 0.18 & 0.77 & 0.34 & 0.37 & 0.48 & & 2.2 & 4.6 & & 0.8 & 1.1 & + 62 & 203500.6 + 601250 & 0.08 & 0.58 & 0.80 & 0.77 & 0.95 & & 3.7 & 4.3 & & 1.3 & 0.8 & l + 63 & 203500.7 + 601130 & 0.48 & 0.67 & 0.25 & 0.31 & 0.61 & & 8.9 & 1.2 & & 2.9 & 1.8 & l + 64 & 203500.9 + 601010 & 0.86 & 0.40 & 0.11 & 0.54 & 0.64 & & 2.9 & 1.3 & & 0.7 & 2.1 & + 65 & 203501.1 + 600923 & 0.96 & 0.34 & & & 0.76 & & 1.2 & 4.7 & & 0.8 & 0.5 & + 66 & 203501.9 + 601003 & 0.98 & 0.04 & 0.26 & 0.13 & 0.96 & & 2.3 & 1.5 & & 0.7 & 1.9 & + 67 & 203504.6 + 601117 & 0.48 & 0.59 & 0.81 & 0.84 & 0.36 & & 1.8 & 1.9 & & 1.4 & 0.6 & + 68 & 203504.6 + 600918 & 0.26 & 0.69 & 0.26 & 0.63 & 0.43 & & 1.1 & 2.3 & & 1.0 & 0.9 & + 69 & 203505.3 + 600623 & 0.12 & 0.22 & 0.57 & 0.51 & 0.91 & & 0.8 & 1.2 & & 1.5 & 1.6 & + 70 & 203505.5 + 600742 & 0.73 & & 0.20 & 0.36 & 0.58 & & 1.2 & 2.6 & & 1.2 & 0.6 & + 71 & 203506.5 + 601040 & 0.28 & 0.41 & 0.02 & 0.85 & 0.20 & & 2.5 & 2.4 & & 1.3 & 1.7 & + 72 & 203508.1 + 601113 & 0.68 & 0.76 & 0.96 & 0.35 & 0.89 & & 1.7 & 2.1 & & 1.6 & 1.0 & + 73 & 203510.1 + 601000 & 0.47 & 0.33 & 0.70 & 0.34 & 0.58 & & 1.1 & 2.2 & & 1.2 & 0.8 & + 74 & 203510.9 + 600856 & 0.48 & 0.18 & 0.57 & 0.40 & 0.39 & & 0.9 & 2.0 & & 1.7 & 0.8 & + 75 & 203512.5 + 600837 & 0.02 & 0.38 & 0.46 & 0.90 & 0.22 & & 0.6 & 1.9 & & 0.7 & 0.9 & + 76 & 203512.7 + 600731 & 0.63 & 0.71 & 0.71 & 0.06 & 0.74 & & 3.3 & 1.8 & & 1.7 & 0.9 & l + 77 & 203513.1 + 600907 & 0.46 & 0.21 & 0.47 & & 0.16 & & 1.6 & 4.4 & & 1.3 & 1.2 & + 78 & 203513.6 + 600522 & & 0.96 & 0.52 & 0.94 & 0.19 & & 2.8 & 5.8 & & 0.9 & 0.7 & + 79 & 203516.2 + 601046 & @xmath794 & 0.80 & 0.19 & & 0.06 & & 10.9 & 118 & & 1.5 & 2.3 & s , l + 80 & 203517.3 + 600813 & 0.40 & 0.09 & 0.37 & 0.72 & 0.30 & & 1.1 & 4.8 & & 1.1 & 0.9 & + 81 & 203518.7 + 601056 & 0.51 & 0.98 & 0.47 & 0.95 & 0.53 & & 5.2 & 1.7 & & 1.3 & 1.6 & l + 82 & 203520.0 + 600933 & 0.51 & & 0.76 & & 0.84 & & 2.4 & 9.2 & & 0.4 & 1.2 & + 83 & 203521.1 + 600951 & 0.42 & 0.88 & 0.90 & 0.52 & 0.28 & & 1.1 & 2.2 & & 0.5 & 0.9 & + 84 & 203525.3 + 600717 & & & 0.41 & 0.12 & 0.76 & & 12.9 & 112 & & 0.6 & 2.2 & l + 85 & 203525.5 + 600958 & 0.92 & 0.98 & 0.07 & 0.30 & 0.96 & & 2.2 & 1.8 & & 1.0 & 1.2 & + 86 & 203527.8 + 600920 & 0.34 & 0.66 & 0.63 & 0.63 & 1.00 & & 0.9 & 1.3 & & 0.7 & 1.0 & + 87 & 203528.6 + 600605 & & 0.08 & 0.63 & 0.14 & & & 0.9 & 2.4 & & 0.9 & 1.0 & + 88 & 203530.1 + 600623 & & 0.85 & 0.22 & 0.65 & 0.39 & & 0.6 & 1.2 & & 0.6 & 1.0 & + 89 & 203530.6 + 600737 & & 0.07 & 0.82 & 0.02 & 0.67 & & 0.6 & 1.3 & & 0.8 & 0.6 & star + 90 & 203532.2 + 601018 & 0.10 & & 0.31 & 0.24 & 0.01 & & 0.9 & 1.4 & & 1.8 & 1.7 & + cccccccccccl 1 & 123025.2 + 413924 & 0.35 & 0.50 & 0.61 & & 0.6 & 1.1 & & 1.1 & 1.8 & + 2 & 123026.7 + 413821 & & 0.25 & 1.00 & & 3.1 & @xmath795 & & 1.2 & 1.6 & l + 3 & 123027.1 + 413814 & 0.84 & 0.72 & 0.35 & & 0.3 & 1.1 & & 0.5 & 1.3 & + 4 & 123027.6 + 413941 & & 0.35 & 0.77 & & 2.3 & 4.5 & & 0.8 & 0.8 & + 5 & 123028.2 + 413958 & 0.10 & 0.73 & 0.42 & & 2.3 & 5.4 & & 1.3 & 0.5 & + 6 & 123028.5 + 413926 & 0.82 & 0.05 & 0.94 & & 1.2 & 1.9 & & 1.3 & 0.8 & + 7 & 123028.7 + 413757 & 0.87 & 0.84 & 0.90 & & 5.1 & 4.9 & & 1.3 & 2.0 & l + 8 & 123029.0 + 414046 & 0.22 & 0.62 & 0.57 & & 1.3 & 1.6 & & 0.5 & 0.3 & + 9 & 123029.4 + 413927 & 0.84 & 0.89 & 0.68 & & 5.4 & 1.5 & & 1.0 & 0.4 & l + 10 & 123029.4 + 414058 & & 0.55 & 0.80 & & 0.7 & 2.0 & & 0.6 & 1.2 & + 11 & 123030.3 + 413853 & @xmath796 & @xmath797 & 0.18 & & 12.0 & 4.6 & & 1.3 & 1.8 & s , l + 12 & 123030.3 + 414126 & & 0.83 & 0.88 & & 4.2 & @xmath787 & & 1.3 & 1.2 & l + 13 & 123030.4 + 414142 & 0.60 & 0.50 & 0.94 & & 12.1 & 1.6 & & 2.1 & 2.6 & l + 14 & 123030.7 + 413911 & 0.46 & 0.84 & 0.40 & & 2.1 & 1.1 & & 0.6 & 0.4 & + 15 & 123031.0 + 413837 & 0.31 & 0.11 & 0.06 & & 1.6 & 2.3 & & 0.5 & 1.4 & + 16 & 123031.3 + 413901 & 0.60 & 0.92 & 0.93 & & 2.1 & 2.4 & & 0.8 & 0.8 & + 17 & 123031.6 + 414140 & & 0.30 & 0.14 & & 22.8 & 721 & & 0.2 & 1.8 & l + 18 & 123032.1 + 413918 & 0.66 & 0.76 & 0.50 & & 11.9 & 1.8 & & 1.2 & 1.9 & l + 19 & 123032.8 + 413845 & 0.15 & 0.60 & 0.24 & & 1.4 & 2.4 & & 0.9 & 0.7 & + 20 & 123032.9 + 414014 & 0.30 & 0.51 & 0.74 & & 2.0 & 2.6 & & 0.8 & 0.6 & + 21 & 123034.1 + 413859 & & 0.24 & 0.84 & & 2.5 & 30.4 & & 0.8 & 0.2 & + 22 & 123034.2 + 413805 & 0.62 & 0.13 & 0.72 & & 2.2 & 2.8 & & 0.8 & 1.0 & + 23 & 123034.3 + 413850 & 0.82 & 0.49 & @xmath798 & & 5.7 & 2.5 & & 2.2 & 3.5 & s , l , sp + 24 & 123034.5 + 413834 & 0.03 & 0.79 & 0.77 & & 0.7 & 1.7 & & 0.5 & 0.7 & + 25 & 123035.1 + 413846 & 0.34 & @xmath799 & 0.57 & & 19.5 & 10.6 & & 1.6 & 1.1 & s , l + 26 & 123035.8 + 413832 & 0.64 & 0.44 & 0.95 & & 1.4 & 1.8 & & 1.5 & 2.4 & + 27 & 123036.2 + 413837 & 0.82 & 0.06 & 0.22 & & 26.0 & 4.5 & & 0.9 & 4.0 & l , sp + 28 & 123037.7 + 413823 & 0.38 & 0.74 & 0.68 & & 0.7 & 1.7 & & 0.7 & 1.3 & + 29 & 123038.2 + 413831 & 0.81 & 0.45 & 0.19 & & 1.8 & 2.9 & & 1.2 & 0.9 & + 30 & 123038.4 + 413831 & & & 0.87 & & 27.2 & @xmath800 & & 0.2 & 0.2 & l + 31 & 123038.4 + 413743 & 0.43 & 0.67 & 0.83 & & 1.3 & 1.3 & & 1.1 & 0.7 & + 32 & 123038.8 + 413810 & @xmath801 & 0.03 & @xmath802 & & 6.3 & 2.4 & & 1.2 & 2.3 & s , l + 33 & 123040.3 + 413813 & 0.21 & 0.27 & 0.39 & & 2.0 & 1.2 & & 1.1 & 0.5 & + 34 & 123043.0 + 413756 & 0.75 & 0.67 & 0.30 & & 0.7 & 1.5 & & 1.2 & 0.5 & + 35 & 123043.1 + 413818 & 0.51 & 0.47 & 0.31 & & 8.1 & 1.4 & & 1.0 & 3.6 &
l , sp + 36 & 123045.5 + 413640 & 0.52 & 0.13 & 0.34 & & 1.4 & 1.4 & & 1.1 & 1.4 & + 37 & 123047.7 + 413807 & 0.21 & 0.36 & 0.21 & & 0.8 & 1.7 & & 1.0 & 0.7 & + 38 & 123047.7 + 413727 & 0.88 & 0.27 & 0.85 & & 3.4 & 3.9 & & 0.9 & 0.7 & l + cccccccccc 1 & 203419.0 + 601007 & @xmath803 & @xmath804 & & @xmath805 & @xmath806 & & & + 2 & 203421.4 + 601017 & @xmath807 & @xmath808 & & @xmath809 & @xmath810 & & & + 3 & 203423.4 + 601039 & & & & @xmath811 & @xmath812 & & & + 4 & 203424.7 + 601038 & @xmath813 & @xmath814 & & @xmath815 & @xmath816 & & & + 5 & 203425.0 + 600906 & @xmath817 & @xmath818 & & @xmath819 & @xmath820 & & & + 6 & 203426.1 + 600909 & @xmath821 & @xmath822 & & @xmath823 & @xmath824 & & & + 7 & 203426.2 + 600914 & @xmath825 & @xmath826 & & @xmath827 & @xmath828 & & & + 8 & 203429.1 + 601051 & @xmath829 & @xmath830 & & @xmath831 & @xmath832 & & & + 9 & 203431.5 + 600638 & @xmath833 & @xmath834 & & @xmath835 & @xmath836 & & & + 10 & 203431.5 + 601056 & @xmath837 & @xmath838 & & @xmath839 & @xmath840 & & & + 11 & 203432.0 + 601207 & @xmath841 & @xmath842 & & @xmath843 & @xmath844 & & & + 12 & 203432.4 + 600821 & @xmath845 & @xmath846 & & & & & & + 13 & 203434.4 + 601032 & @xmath847 & @xmath848 & & @xmath849 & @xmath850 & & & + 14 & 203435.3 + 600742 & @xmath851 & @xmath852 & & @xmath853 & @xmath854 & & & + 15 & 203436.1 + 600839 & @xmath855 & @xmath856 & & @xmath857 & @xmath858 & & & + 16 & 203436.4 + 600558 & @xmath859 & @xmath860 & & @xmath861 & @xmath862 & & & + 17 & 203436.5 + 600930 & @xmath863 & @xmath864 & & @xmath865 & @xmath866 & & & + 18 & 203437.2 + 600954 & @xmath867 & @xmath868 & & @xmath869 & @xmath870 & & & + 19 & 203437.2 + 601019 & @xmath871 & @xmath872 & & @xmath873 & @xmath874 & & & + 20 & 203437.9 + 600434 & @xmath875 & @xmath876 & & @xmath833 & @xmath877 & & @xmath878 & @xmath879 + 21 & 203439.8 + 600822 & @xmath880 & @xmath881 & & @xmath882 & @xmath883 & & @xmath884 & @xmath885 + 22 & 203440.5 + 600946 & @xmath886 & @xmath887 & & @xmath888 & @xmath889 & & @xmath890 & @xmath891 + 23 & 203441.3 + 600846 & @xmath892 & @xmath893 & & @xmath894 & @xmath895 & & @xmath896 & @xmath897 + 24 & 203442.0 + 600529 & @xmath898 & @xmath899 & & @xmath900 & @xmath901 & & @xmath902 & @xmath903 + 25 & 203443.1 + 600652 & @xmath904 & @xmath905 & & @xmath906 & @xmath907 & & @xmath908 & @xmath909 + 26 & 203444.2 + 600719 & @xmath910 & @xmath911 & & @xmath912 & @xmath913 & & @xmath914 & @xmath915 + 27 & 203445.9 + 601049 & @xmath916 & @xmath917 & & @xmath918 & @xmath919 & & @xmath920 & @xmath921 + 28 & 203446.3 + 601328 & @xmath922 & @xmath923 & & @xmath924 & @xmath925 & & @xmath926 & @xmath927 + 29 & 203447.1 + 600850 & @xmath928 & @xmath929 & & & & & & + 30 & 203447.5 + 601003 & @xmath930 & @xmath931 & & @xmath932 & @xmath933 & & @xmath934 & @xmath935 + 31 & 203447.6 + 600932 & @xmath936 & @xmath937 & & @xmath938 & @xmath939 & & @xmath940 & @xmath941 + 32 & 203448.4 + 600811 & @xmath942 & @xmath943 & & @xmath944 & @xmath945 & & @xmath946 & @xmath947 + 33 & 203448.6 + 600941 & @xmath948 & @xmath949 & & @xmath950 & @xmath951 & & @xmath952 & @xmath953 + 34 & 203448.8 + 600554 & @xmath954 & @xmath955 & & @xmath956 & @xmath957 & & @xmath958 & @xmath959 + 35 & 203449.0 + 601217 & @xmath960 & @xmath961 & & @xmath962 & @xmath963 & & @xmath947 & @xmath964 + 36 & 203449.4 + 601342 & @xmath965 & @xmath966 & & @xmath967 & @xmath968 & & @xmath969 & @xmath970 + 37 & 203449.7 + 601011 & @xmath879 & @xmath971 & & @xmath972 & @xmath973 & & @xmath974 & @xmath975 + 38 & 203449.7 + 600806 & @xmath976 & @xmath977 & & @xmath978 & @xmath979 & & @xmath953 & @xmath980 + 39 & 203449.7 + 601020 & @xmath981 & @xmath982 & & @xmath983 & @xmath984 & & @xmath985 & @xmath986 + 40 & 203450.0 + 600924 & @xmath987 & @xmath988 & & @xmath989 & @xmath990 & & @xmath991 & @xmath992 + 41 & 203450.4 + 601016 & @xmath993 & @xmath994 & & @xmath995 & @xmath996 & & @xmath997 & @xmath998 + 42 & 203450.4 + 600724 & & & & @xmath845 & @xmath999 & & @xmath1000 & @xmath1001 + 43 & 203450.7 + 600747 & @xmath1002 & @xmath1003 & & @xmath1004 & @xmath1005 & & @xmath1006 & @xmath1007 + 44 & 203450.7 + 601207 & @xmath1008 & @xmath1009 & & @xmath1010 & @xmath1011 & & @xmath1012 & @xmath1013 + 45 & 203450.9 + 601020 & @xmath1014 & @xmath848 & & @xmath1015 & @xmath1016 & & @xmath1017 & @xmath1018 + 46 & 203451.5 + 601248 & @xmath1019 & @xmath1020 & & @xmath1021 & @xmath1022 & & @xmath1023 & @xmath1024 + 47 & 203451.5 + 600909 & @xmath1025 & @xmath1026 & & @xmath1027 & @xmath1028 & & & + 48 & 203453.3 + 601055 & @xmath1029 & @xmath1030 & & @xmath1031 & @xmath1032 & & @xmath1033 & @xmath1034 + 49 & 203453.4 + 600719 & @xmath1035 & @xmath1036 & & @xmath880 & @xmath1037 & & @xmath1038 & @xmath1039 + 50 & 203453.7 + 600910 & @xmath1040 & @xmath1041 & & @xmath1042 & @xmath1043 & & @xmath1044 & @xmath1045 + 51 & 203455.6 + 600930 & @xmath1046 & @xmath1047 & & @xmath1048 & @xmath1049 & & @xmath1050 & @xmath1051 + 52 & 203456.5 + 600834 & @xmath1052 & @xmath1053 & & @xmath1054 & @xmath1055 & & @xmath1056 & @xmath1057 + 53 & 203456.5 + 600819 & @xmath1058 & @xmath1059 & & @xmath1060 & @xmath1061 & & @xmath1062 & @xmath1063 + 54 & 203456.8 + 600532 & @xmath1064 & @xmath1065 & & @xmath1066 & @xmath1067 & & @xmath1068 & @xmath1069 + 55 & 203456.9 + 601209 & @xmath1070 & @xmath1071 & & @xmath1072 & @xmath1073 & & @xmath1074 & @xmath1075 + 56 & 203456.9 + 600826 & @xmath1076 & @xmath1077 & & @xmath1078 & @xmath1079 & & @xmath1080 & @xmath1081 + 57 & 203457.2 + 600733 & @xmath1082 & @xmath1083 & & @xmath1084 & @xmath1085 & & @xmath1086 & @xmath1087 + 58 & 203457.7 + 600948 & @xmath1088 & @xmath1089 & & @xmath1090 & @xmath939 & & @xmath1091 & @xmath1092 + 59 & 203458.4 + 600934 & @xmath1093 & @xmath1094 & & @xmath1095 & @xmath1096 & & @xmath1097 & @xmath1098 + 60 & 203500.1 + 600908 & @xmath1099 & @xmath1100 & & @xmath1101 & @xmath1102 & & @xmath1103 & @xmath866 + 61 & 203500.4 + 600859 & @xmath844 & @xmath1104 & & @xmath1105 & @xmath1106 & & @xmath1107 & @xmath1108 + 62 & 203500.6 + 601250 & @xmath1109 & @xmath1110 & & @xmath1111 & @xmath1112 & & @xmath1113 & @xmath1114 + 63 & 203500.7 + 601130 & @xmath1115 & @xmath1116 & & @xmath1117 & @xmath1118 & & @xmath1119 & @xmath1120 + 64 & 203500.9 + 601010 & @xmath1121 & @xmath1122 & & @xmath813 & @xmath1123 & & @xmath1124 & @xmath942 + 65 & 203501.1 + 600923 & @xmath1125 & @xmath1126 & & @xmath1127 & @xmath1128 & & @xmath1129 & @xmath1130 + 66 & 203501.9 + 601003 & @xmath817 & @xmath944 & & @xmath815 & @xmath1131 & & @xmath1132 & @xmath1133 + 67 & 203504.6 + 601117 & @xmath1134 & @xmath1135 & & @xmath1136 & @xmath1137 & & @xmath1138 & @xmath1139 + 68 & 203504.6 + 600918 & @xmath1140 & @xmath1141 & & @xmath1142 & @xmath1143 & & @xmath1144 & @xmath1145 + 69 & 203505.3 + 600623 & @xmath1146 & @xmath1147 & & @xmath1148 & @xmath829 & & @xmath1149 & @xmath1109 + 70 & 203505.5 + 600742 & @xmath1150 & @xmath1151 & & @xmath1152 & @xmath1153 & & @xmath1154 & @xmath1155 + 71 & 203506.5 + 601040 & @xmath1156 & @xmath1157 & & @xmath1158 & @xmath1159 & & @xmath1160 & @xmath1161 + 72 & 203508.1 + 601113 & @xmath1162 & @xmath1163 & & @xmath1164 & @xmath1165 & & @xmath1166 & @xmath1167 + 73 & 203510.1 + 601000 & @xmath1168 & @xmath1169 & & @xmath1170 & @xmath1171 & & @xmath1172 & @xmath1173 + 74 & 203510.9 + 600856 & @xmath1174 & @xmath1175 & & @xmath1176 & @xmath1177 & & @xmath1178 & @xmath1179 + 75 & 203512.5 + 600837 & @xmath1180 & @xmath1181 & & @xmath1182 & @xmath1183 & & @xmath1184 & @xmath1185 + 76 & 203512.7 + 600731 & @xmath1186 & @xmath1187 & & @xmath1188 & @xmath1189 & & @xmath947 & @xmath1190 + 77 & 203513.1 + 600907 & @xmath1191 & @xmath1192 & & @xmath1193 & @xmath1194 & & @xmath1195 & @xmath1196 + 78 & 203513.6 + 600522 & & & & @xmath1197 & @xmath1198 & & @xmath1199 & @xmath1200 + 79 & 203516.2 + 601046 & @xmath1201 & @xmath1202 & & @xmath1203 & @xmath1204 & & @xmath938 & @xmath1205 + 80 & 203517.3 + 600813 & @xmath1206 & @xmath1207 & & @xmath1208 & @xmath1209 & & @xmath1210 & @xmath1211 + 81 & 203518.7 + 601056 & @xmath1212 & @xmath1213 & & @xmath1214 & @xmath1215 & & @xmath1053 & @xmath1216 + 82 & 203520.0 + 600933 & @xmath1217 & @xmath1218 & & @xmath1219 & @xmath1220 & & @xmath1221 & @xmath1222 + 83 & 203521.1 + 600951 & @xmath1223 & @xmath1224 & & @xmath1225 & @xmath1226 & & @xmath1227 & @xmath1228 + 84 & 203525.3 + 600717 & & & & @xmath1229 & @xmath1230 & & @xmath1231 & @xmath1232 + 85 & 203525.5 + 600958 & @xmath1233 & @xmath1234 & & @xmath1235 & @xmath1236 & & @xmath1237 & @xmath1238 + 86 & 203527.8 + 600920 & @xmath1239 & @xmath1240 & & @xmath1241 & @xmath1242 & & @xmath1243 & @xmath1244 + 87 & 203528.6 + 600605 & & & & @xmath1245 & @xmath1246 & & @xmath1247 & @xmath1248 + 88 & 203530.1 + 600623 & & & & @xmath1249 & @xmath1250 & & @xmath1251 & @xmath1252 + 89 & 203530.6 + 600737 & & & & @xmath1253 & @xmath1254 & & @xmath1255 & @xmath1256 + 90 & 203532.2 + 601018 & @xmath1257 & @xmath1258 & & & & & @xmath1259 & @xmath1260 + cccccccccc 1 & 203419.0 + 601007 & & & & @xmath1261 & @xmath1262 & & @xmath1263 & @xmath1264 + 2 & 203421.4 + 601017 & & & & @xmath1265 & @xmath1266 & & @xmath1267 & @xmath1268 + 3 & 203423.4 + 601039 & & & & @xmath1269 & @xmath1270 & & @xmath1271 & @xmath1272 + 4 & 203424.7 + 601038 & & & & @xmath1273 & @xmath1274 & & @xmath1275 & @xmath1276 + 5 & 203425.0 + 600906 & & & & @xmath1277 & @xmath1278 & & @xmath817 & @xmath1279 + 6 & 203426.1 + 600909 & & & & @xmath1280 & @xmath1281 & & @xmath821 & @xmath1282 + 7 & 203426.2 + 600914 & & & & @xmath1283 & @xmath1284 & & @xmath1285 & @xmath1286 + 8 & 203429.1 + 601051 & & & & & & & @xmath1263 & @xmath1189 + 9 & 203431.5 + 600638 & & & & & & & @xmath1287 & @xmath1288 + 10 & 203431.5 + 601056 & & & & @xmath877 & @xmath1289 & & @xmath1290 & @xmath1291 + 11 & 203432.0 + 601207 & @xmath1292 & @xmath1293 & & & & & @xmath1294 & @xmath1295 + 12 & 203432.4 + 600821 & & & & & & & @xmath803 & @xmath1296 + 13 & 203434.4 + 601032 & @xmath1297 & @xmath1298 & & @xmath1299 & @xmath1300 & & @xmath1301 & @xmath1302 + 14 & 203435.3 + 600742 & & & & @xmath1303 & @xmath1304 & & @xmath1305 & @xmath1306 + 15 & 203436.1 + 600839 & @xmath1307 & @xmath1308 & & @xmath1309 & @xmath1310 & & @xmath1311 & @xmath1312 + 16 & 203436.4 + 600558 & & & & & & & @xmath1313 & @xmath1314 + 17 & 203436.5 + 600930 & @xmath1315 & @xmath1316 & & @xmath1317 & @xmath1318 & & @xmath1319 & @xmath1320 + 18 & 203437.2 + 600954 & @xmath1321 & @xmath1322 & & @xmath1323 & @xmath1324 & & @xmath1325 & @xmath1326 + 19 & 203437.2 + 601019 & @xmath1327 & @xmath1328 & & @xmath840 & @xmath1329 & & @xmath1330 & @xmath1331 + 20 & 203437.9 + 600434 & & & & & & & @xmath1332 & @xmath1333 + 21 & 203439.8 + 600822 & @xmath1248 & @xmath1334 & & @xmath1335 & @xmath1336 & & @xmath1337 & @xmath1338 + 22 & 203440.5 + 600946 & @xmath1339 & @xmath1340 & & @xmath1341 & @xmath1342 & & @xmath1343 & @xmath1344 + 23 & 203441.3 + 600846 & @xmath1345 & @xmath1346 & & @xmath1347 & @xmath1348 & & @xmath1349 & @xmath1113 + 24 & 203442.0 + 600529 & @xmath1350 & @xmath1351 & & @xmath1352 & @xmath1353 & & @xmath833 & @xmath1354 + 25 & 203443.1 + 600652 & @xmath1355 & @xmath1356 & & @xmath1357 & @xmath1358 & & @xmath1359 & @xmath1360 + 26 & 203444.2 + 600719 & @xmath1361 & @xmath1362 & & @xmath1363 & @xmath1364 & & @xmath1275 & @xmath1365 + 27 & 203445.9 + 601049 & @xmath1366 & @xmath1367 & & @xmath1368 & @xmath1369 & & @xmath1370 & @xmath1371 + 28 & 203446.3 + 601328 & @xmath932 & @xmath1372 & & @xmath1373 & @xmath1374 & & @xmath1056 & @xmath1375 + 29 & 203447.1 + 600850 & & & & & & & @xmath830 & @xmath1157 + 30 & 203447.5 + 601003 & @xmath1376 & @xmath1377 & & @xmath1378 & @xmath1349 & & @xmath1379 & @xmath1380 + 31 & 203447.6 + 600932 & @xmath1381 & @xmath1382 & & @xmath1383 & @xmath1384 & & @xmath1385 & @xmath1386 + 32 & 203448.4 + 600811 & @xmath1387 & @xmath1064 & & @xmath1388 & @xmath1055 & & @xmath1052 & @xmath912 + 33 & 203448.6 + 600941 & & & & & & & @xmath1259 & @xmath1389 + 34 & 203448.8 + 600554 & @xmath1390 & @xmath1391 & & @xmath1392 & @xmath1393 & & @xmath1394 & @xmath1395 + 35 & 203449.0 + 601217 & @xmath859 & @xmath1396 & & @xmath1397 & @xmath1398 & & @xmath912 & @xmath822 + 36 & 203449.4 + 601342 & @xmath1399 & @xmath1400 & & @xmath1401 & @xmath1402 & & @xmath1403 & @xmath1404 + 37 & 203449.7 + 601011 & @xmath1405 & @xmath1406 & & @xmath1407 & @xmath1104 & & @xmath1408 & @xmath1409 + 38 & 203449.7 + 600806 & @xmath1410 & @xmath1411 & & @xmath1412 & @xmath1413 & & @xmath1414 & @xmath852 + 39 & 203449.7 + 601020 & @xmath1415 & @xmath1416 & & @xmath1417 & @xmath1418 & & @xmath1419 & @xmath1420 + 40 & 203450.0 + 600924 & @xmath1421 & @xmath1422 & & @xmath1423 & @xmath1424 & & @xmath1425 & @xmath1426 + 41 & 203450.4 + 601016 & @xmath1427 & @xmath1428 & & & & & @xmath1429 & @xmath1430 + 42 & 203450.4 + 600724 & @xmath1431 & @xmath1432 & & @xmath1433 & @xmath1434 & & @xmath1275 & @xmath1279 + 43 & 203450.7 + 600747 & @xmath1435 & @xmath1436 & & @xmath1437 & @xmath1438 & & @xmath1439 & @xmath1440 + 44 & 203450.7 + 601207 & @xmath1441 & @xmath1442 & & @xmath1443 & @xmath1444 & & @xmath1445 & @xmath1446 + 45 & 203450.9 + 601020 & @xmath1447 & @xmath1448 & & @xmath1449 & @xmath1450 & & @xmath1451 & @xmath1452 + 46 & 203451.5 + 601248 & @xmath1453 & @xmath1454 & & @xmath1455 & @xmath1456 & & @xmath1457 & @xmath829 + 47 & 203451.5 + 600909 & @xmath1458 & @xmath1459 & & @xmath1460 & @xmath1461 & & @xmath1462 & @xmath1354 + 48 & 203453.3 + 601055 & @xmath1463 & @xmath1464 & & @xmath1465 & @xmath1466 & & @xmath1467 & @xmath1468 + 49 & 203453.4 + 600719 & @xmath1389 & @xmath1469 & & @xmath1470 & @xmath1471 & & @xmath1472 & @xmath1473 + 50 & 203453.7 + 600910 & @xmath1474 & @xmath1475 & & @xmath1476 & @xmath1477 & & @xmath1000 & @xmath1478 + 51 & 203455.6 + 600930 & @xmath1479 & @xmath1480 & & @xmath988 & @xmath1481 & & @xmath1482 & @xmath1264 + 52 & 203456.5 + 600834 & @xmath1483 & @xmath1057 & & @xmath1484 & @xmath1485 & & @xmath1486 & @xmath1487 + 53 & 203456.5 + 600819 & @xmath1488 & @xmath1489 & & @xmath1490 & @xmath1491 & & @xmath1492 & @xmath1380 + 54 & 203456.8 + 600532 & @xmath1493 & @xmath1494 & & @xmath1495 & @xmath1008 & & @xmath1496 & @xmath1057 + 55 & 203456.9 + 601209 & @xmath1497 & @xmath1498 & & @xmath1499 & @xmath1328 & & @xmath1500 & @xmath1501 + 56 & 203456.9 + 600826 & @xmath1502 & @xmath1503 & & @xmath1035 & @xmath1504 & & @xmath1505 & @xmath1506 + 57 & 203457.2 + 600733 & & & & & & & @xmath1507 & @xmath1508 + 58 & 203457.7 + 600948 & @xmath944 & @xmath1509 & & @xmath1510 & @xmath1496 & & @xmath1511 & @xmath1512 + 59 & 203458.4 + 600934 & @xmath1513 & @xmath1514 & & @xmath1515 & @xmath1516 & & @xmath1517 & @xmath1203 + 60 & 203500.1 + 600908 & @xmath1518 & @xmath1102 & & @xmath1519 & @xmath1316 & & @xmath1520 & @xmath1100 + 61 & 203500.4 + 600859 & @xmath1021 & @xmath991 & & @xmath1521 & @xmath1522 & & @xmath1523 & @xmath1524 + 62 & 203500.6 + 601250 & @xmath1405 & @xmath1224 & & @xmath1525 & @xmath1526 & & @xmath1527 & @xmath1528 + 63 & 203500.7 + 601130 & @xmath1529 & @xmath1530 & & @xmath1529 & @xmath1116 & & @xmath1531 & @xmath1532 + 64 & 203500.9 + 601010 & @xmath1533 & @xmath1534 & & @xmath1535 & @xmath1536 & & @xmath1537 & @xmath1538 + 65 & 203501.1 + 600923 & @xmath1539 & @xmath1540 & & @xmath1541 & @xmath1542 & & @xmath1543 & @xmath1544 + 66 & 203501.9 + 601003 & @xmath1545 & @xmath1546 & & @xmath938 & @xmath1547 & & @xmath1548 & @xmath1549 + 67 & 203504.6 + 601117 & @xmath1364 & @xmath1550 & & @xmath993 & @xmath1551 & & @xmath1552 & @xmath1553 + 68 & 203504.6 + 600918 & @xmath1554 & @xmath1555 & & @xmath1556 & @xmath1557 & & @xmath1558 & @xmath1559 + 69 & 203505.3 + 600623 & @xmath803 & @xmath1560 & & @xmath1561 & @xmath1562 & & @xmath1563 & @xmath1564 + 70 & 203505.5 + 600742 & @xmath1565 & @xmath1566 & & @xmath1567 & @xmath1568 & & @xmath1569 & @xmath1335 + 71 & 203506.5 + 601040 & @xmath1570 & @xmath1571 & & @xmath1572 & @xmath1573 & & @xmath1574 & @xmath1575 + 72 & 203508.1 + 601113 & @xmath1576 & @xmath1577 & & @xmath996 & @xmath1578 & & @xmath1579 & @xmath1580 + 73 & 203510.1 + 601000 & @xmath1581 & @xmath1582 & & @xmath1583 & @xmath1584 & & @xmath1585 & @xmath1586 + 74 & 203510.9 + 600856 & @xmath1587 & @xmath1588 & & @xmath1589 & @xmath1590 & & @xmath1591 & @xmath1592 + 75 & 203512.5 + 600837 & @xmath1593 & @xmath1594 & & @xmath1595 & @xmath906 & & @xmath1331 & @xmath1596 + 76 & 203512.7 + 600731 & @xmath1597 & @xmath859 & & @xmath1598 & @xmath1599 & & @xmath1600 & @xmath1052 + 77 & 203513.1 + 600907 & @xmath1601 & @xmath1602 & & @xmath1603 & @xmath1604 & & @xmath1605 & @xmath1606 + 78 & 203513.6 + 600522 & @xmath1607 & @xmath1608 & & @xmath1609 & @xmath1610 & & @xmath1611 & @xmath1612 + 79 & 203516.2 + 601046 & @xmath1613 & @xmath1614 & & @xmath1615 & @xmath1616 & & @xmath1617 & @xmath1618 + 80 & 203517.3 + 600813 & @xmath1619 & @xmath1620 & & @xmath1621 & @xmath1622 & & @xmath1623 & @xmath1624 + 81 & 203518.7 + 601056 & @xmath1625 & @xmath1552 & & @xmath1626 & @xmath1627 & & @xmath1628 & @xmath1600 + 82 & 203520.0 + 600933 & @xmath1629 & @xmath1630 & & @xmath1631 & @xmath1632 & & @xmath1633 & @xmath1434 + 83 & 203521.1 + 600951 & @xmath1634 & @xmath1635 & & @xmath1636 & @xmath1637 & & @xmath1638 & @xmath1639 + 84 & 203525.3 + 600717 & @xmath962 & @xmath1333 & & @xmath1187 & @xmath1527 & & @xmath1600 & @xmath1640 + 85 & 203525.5 + 600958 & @xmath1364 & @xmath1641 & & @xmath1642 & @xmath1643 & & @xmath1644 & @xmath1645 + 86 & 203527.8 + 600920 & @xmath1646 & @xmath1647 & & @xmath1162 & @xmath1648 & & @xmath1517 & @xmath1649 + 87 & 203528.6 + 600605 & @xmath1650 & @xmath1651 & & @xmath1652 & @xmath1653 & & @xmath1654 & @xmath1655 + 88 & 203530.1 + 600623 & @xmath1139 & @xmath1656 & & @xmath1139 & @xmath1657 & & @xmath1658 & @xmath1659 + 89 & 203530.6 + 600737 & @xmath1660 & @xmath1661 & & @xmath1662 & @xmath917 & & @xmath1663 & @xmath1301 + 90 & 203532.2 + 601018 & @xmath1664 & @xmath1665 & & @xmath1666 & @xmath1331 & & @xmath1599 & @xmath1667 + ccccccccccccc 1 & 123025.2 + 413924 & @xmath1668 & @xmath1669 & & @xmath1552 & @xmath1670 & & @xmath880 & @xmath1671 & & @xmath1213 & @xmath1672 + 2 & 123026.7 + 413821 & & & & @xmath1673 & @xmath1674 & & @xmath1675 & @xmath1676 & & @xmath831 & @xmath1677 + 3 & 123027.1 + 413814 & @xmath1678 & @xmath1679 & & @xmath1572 & @xmath878 & & @xmath1263 & @xmath1680 & & @xmath946 & @xmath1681 + 4 & 123027.6 + 413941 & @xmath1682 & @xmath1683 & & @xmath1684 & @xmath1685 & & @xmath1686 & @xmath1687 & & @xmath1688 & @xmath1689 + 5 & 123028.2 + 413958 & @xmath1159 & @xmath1690 & & @xmath1691 & @xmath1692 & & @xmath1693 & @xmath1694 & & @xmath1695 & @xmath1696 + 6 & 123028.5 + 413926 & @xmath1697 & @xmath1698 & & @xmath1699 & @xmath1700 & & @xmath982 & @xmath1701 & & @xmath1702 & @xmath1703 + 7 & 123028.7 + 413757 & @xmath1704 & @xmath1705 & & @xmath1706 & @xmath1707 & & @xmath1137 & @xmath1708 & & @xmath1300 & @xmath1709 + 8 & 123029.0 + 414046 & @xmath1364 & @xmath1648 & & @xmath812 & @xmath1710 & & @xmath1489 & @xmath1711 & & @xmath1712 & @xmath1713 + 9 & 123029.4 + 413927 & @xmath1714 & @xmath1715 & & @xmath1716 & @xmath1717 & & @xmath1718 & @xmath1719 & & @xmath1720 & @xmath1721 + 10 & 123029.4 + 414058 & @xmath1722 & @xmath1723 & & @xmath1724 & @xmath1222 & & @xmath1725 & @xmath1726 & & @xmath1727 & @xmath1689 + 11 & 123030.3 + 413853 & @xmath942 & @xmath1425 & & @xmath1728 & @xmath1729 & & @xmath1730 & @xmath1713 & & @xmath1121 & @xmath1484 + 12 & 123030.3 + 414126 & & & & @xmath1731 & @xmath1035 & & @xmath948 & @xmath1732 & & @xmath1547 & @xmath1109 + 13 & 123030.4 + 414142 & @xmath1537 & @xmath1733 & & @xmath1734 & @xmath1735 & & @xmath821 & @xmath1736 & & @xmath1737 & @xmath1738 + 14 & 123030.7 + 413911 & @xmath1739 & @xmath1740 & & @xmath1741 & @xmath1742 & & @xmath1743 & @xmath1742 & & @xmath1741 & @xmath1744 + 15 & 123031.0 + 413837 & @xmath1405 & @xmath1745 & & @xmath1746 & @xmath1138 & & @xmath1404 & @xmath1747 & & @xmath1748 & @xmath1404 + 16 & 123031.3 + 413901 & @xmath1749 & @xmath1750 & & @xmath1751 & @xmath1752 & & @xmath1753 & @xmath1493 & & @xmath1280 & @xmath1445 + 17 & 123031.6 + 414140 & @xmath1754 & @xmath1755 & & @xmath1315 & @xmath1617 & & @xmath1756 & @xmath1757 & & @xmath1758 & @xmath1759 + 18 & 123032.1 + 413918 & @xmath1760 & @xmath1484 & & @xmath1761 & @xmath1762 & & @xmath1763 & @xmath1764 & & @xmath1763 & @xmath1765 + 19 & 123032.8 + 413845 & @xmath1766 & @xmath1767 & & @xmath1195 & @xmath1768 & & @xmath1138 & @xmath1769 & & @xmath1770 & @xmath1771 + 20 & 123032.9 + 414014 & @xmath1772 & @xmath1773 & & @xmath1387 & @xmath1774 & & @xmath1775 & @xmath932 & & @xmath1776 & @xmath1777 + 21 & 123034.1 + 413859 & @xmath1778 & @xmath1779 & & @xmath1780 & @xmath1781 & & @xmath1782 & @xmath1783 & & @xmath1784 & @xmath1785 + 22 & 123034.2 + 413805 & @xmath1623 & @xmath1786 & & @xmath1787 & @xmath1788 & & @xmath1789 & @xmath1790 & & @xmath1332 & @xmath1791 + 23 & 123034.3 + 413850 & @xmath880 & @xmath1792 & & @xmath1793 & @xmath1794 & & @xmath1795 & @xmath1796 & & @xmath1797 & @xmath1798 + 24 & 123034.5 + 413834 & @xmath1799 & @xmath1800 & & @xmath1801 & @xmath1802 & & @xmath1803 & @xmath1287 & & @xmath1804 & @xmath1793 + 25 & 123035.1 + 413846 & @xmath948 & @xmath1805 & & @xmath1806 & @xmath1728 & & @xmath1548 & @xmath1807 & & @xmath1808 & @xmath1809 + 26 & 123035.8 + 413832 & @xmath1810 & @xmath1811 & & @xmath1812 & @xmath1813 & & @xmath897 & @xmath1814 & & @xmath1815 & @xmath1387 + 27 & 123036.2 + 413837 & @xmath1761 & @xmath1816 & & @xmath1817 & @xmath1818 & & @xmath1819 & @xmath1820 & & @xmath1720 & @xmath1733 + 28 & 123037.7 + 413823 & @xmath1821 & @xmath1822 & & @xmath1648 & @xmath1823 & & @xmath1824 & @xmath1166 & & @xmath1825 & @xmath1826 + 29 & 123038.2 + 413831 & @xmath1827 & @xmath1828 & & @xmath1829 & @xmath1830 & & @xmath1831 & @xmath1832 & & @xmath1833 & @xmath1661 + 30 & 123038.4 + 413831 & & & & & & & @xmath1739 & @xmath1714 & & @xmath1739 & @xmath1487 + 31 & 123038.4 + 413743 & @xmath1834 & @xmath1835 & & @xmath1836 & @xmath1837 & & @xmath1838 & @xmath1839 & & @xmath1252 & @xmath1840 + 32 & 123038.8 + 413810 & @xmath1482 & @xmath1429 & & @xmath1574 & @xmath1841 & & @xmath1842 & @xmath1843 & & @xmath1844 & @xmath1845 + 33 & 123040.3 + 413813 & @xmath1846 & @xmath1847 & & @xmath1848 & @xmath1849 & & @xmath1549 & @xmath1850 & & @xmath1088 & @xmath1851 + 34 & 123043.0 + 413756 & @xmath1852 & @xmath1853 & & @xmath1854 & @xmath1855 & & @xmath1856 & @xmath937 & & @xmath1857 & @xmath1858 + 35 & 123043.1 + 413818 & @xmath1716 & @xmath1859 & & @xmath1762 & @xmath1809 & & @xmath1720 & @xmath1819 & & @xmath1720 & @xmath1763 + 36 & 123045.5 + 413640 & @xmath977 & @xmath1860 & & @xmath1431 & @xmath1387 & & @xmath1861 & @xmath1862 & & @xmath1332 & @xmath1092 + 37 & 123047.7 + 413807 & @xmath1863 & @xmath1864 & & @xmath1865 & @xmath1866 & & @xmath1867 & @xmath1868 & & @xmath1869 & @xmath1493 + 38 & 123047.7 + 413727 & @xmath1771 & @xmath1064 & & @xmath1481 & @xmath1870 & & @xmath1513 & @xmath1871 & & @xmath917 & @xmath815 + | we analyze data from five _ chandra _ observations of the spiral galaxy ngc 6946 and from three _ chandra _ observations of the irregular / spiral interacting galaxy pair ngc 4485/4490 , with an emphasis on investigating the long - term variability exhibited by the source populations .
we detect 90 point sources coincident with ngc 6946 down to luminosities of a few times @xmath0 , and 38 sources coincident with ngc 4485/90 down to a luminosity of @xmath1 .
twenty - five ( 15 ) sources in ngc 6946 ( ngc 4485/90 ) exhibit long - term ( i.e. , weeks to years ) variability in luminosity ; 11 ( 4 ) are transient candidates . the single ultraluminous x - ray source ( ulx ) in ngc 6946 and all but one of the eight ulxs in ngc 4485/90 exhibit long - term flux variability .
two of the ulxs in ngc 4485/90 have not been identified before as ultraluminous sources .
the widespread variability in both systems is indicative of the populations being dominated by x - ray binaries , and this is supported by the x - ray colors of the sources .
the distribution of colors among the sources indicates a large fraction of high - mass x - ray binaries in both systems .
the shapes of the x - ray luminosity functions of the galaxies do not change significantly between observations and can be described by power laws with cumulative slopes @xmath2 ( ngc 6946 ) and @xmath3 ( ngc 4485/90 ) . |
in the present paper , a particular approach to understanding lepton mixing , proposed in @xcite and further studied in @xcite , will be examined in more detail .
first , we briefly review the approach .
of course , the standard model interaction term for @xmath1 decay or @xmath2 includes the leptonic piece : @xmath3 the object @xmath4 is now known @xcite-@xcite to be a linear combination of neutrino mass eigenstates , @xmath5 : @xmath6 where , in a basis with the charged leptons diagonal , the full lepton mixing matrix is written as : @xmath7 as has been discussed by many authors @xcite-@xcite the results of neutrino oscillation experiments are ( neglecting possible phases to be discussed later ) consistent with the tribimaximal mixing " matrix : @xmath8 many different approaches have been used to explain the form of @xmath9 .
natural",and often investigated one uses the parallel three generation structure of the fundamental fermion families as a starting point .
an underlying discrete symmetry @xmath10 , the permutation group on three objects , is then assumed .
@xcite-@xcite the permutation matrices @xmath11 are , @xmath12,\hspace{.3 in } s^{(12)}= \left [ \begin{array}{ccc } 0&1 & 0 \\ 1&0&0\\ 0&0&1 \end{array } \right],\hspace{.3 in } s^{(13)}= \left [ \begin{array}{ccc } 0&0&1 \\ 0&1&0\\ 1&0&0 \end{array } \right ] , \nonumber \\ s^{(23)}&= & \left [ \begin{array}{ccc } 1&0&0 \\ 0&0&1\\ 0&1&0 \end{array } \right],\hspace{.3 in } s^{(123)}= \left [ \begin{array}{ccc } 0&0&1 \\
1&0&0\\ 0&1&0 \end{array } \right],\hspace{.3 in } s^{(132)}= \left [ \begin{array}{ccc } 0&1&0 \\ 0&0&1 \\ 1&0&0 \end{array } \right ] , \label{s3matrices}\end{aligned}\ ] ] this defining representation is not irreducible . the 3-dimensional space breaks up into irreducible 2-dimensonal and 1-dimensional spaces .
one may note that the tribimaximal matrix , @xmath13 is an example of the transformation which relates the given basis to the irreducible one .
this fact provides our motivation for further investigating the @xmath14 symmetry , even though many other interesting approaches exist .
of course , the symmetry requirement reads , @xmath15=0 , \label{commutator}\ ] ] where @xmath11 stands for any of the six matrices in eq.([s3matrices ] ) and @xmath16 is the neutrino mass matrix . by explicitly evaluating the commutators one obtains the solution : @xmath17+\beta \left [ \begin{array}{ccc } 1&1&1\\ 1&1&1\\ 1&1&1\\ \end{array } \right ] \equiv \alpha { \bf 1}+\beta d .
\label{solution}\ ] ] @xmath18 and @xmath1 are , in general , complex numbers for the case of majorana neutrinos while @xmath19 is usually called the democratic " matrix .
it is easy to verify that this @xmath20 may be brought to diagonal ( but not necessarily real ) form by the real orthogonal matrix , @xmath21 defined above : @xmath22 .
\label{complexeigenvalues}\ ] ] @xmath23 may be written in terms of the eigenvectors of @xmath20 as : @xmath24 , \label{vecr}\ ] ] for example , @xmath25 is the first column of the tribimaximal matrix , eq.([r ] ) .
physically one can assign different masses to the mass eigenstate @xmath26 in the 1-dimensional basis and to the ( doubly degenerate ) eigenstates @xmath25 and @xmath27 in the 2-dimensional basis . at first glance
this sounds ideal since it is well known that the three neutrino masses are grouped into two almost degenerate ones ( solar neutrinos `` ) and one singlet , with different values .
however , since we are demanding that r be taken as the tribimaximal form , the physical identification requires @xmath25 and @xmath26 to be the ' ' solar " neutrino eigenstates rather than the degenerate ones @xmath25 and @xmath27 .
this had been considered a serious objection to the present approach since often a scenario is pictured in which the mass eigenvalue for @xmath28 is considerably larger than the roughly degenerate masses associated with @xmath29 and @xmath30 .
a way out was suggested in @xcite where it was noted that , for values of @xmath31 larger than around 0.3 ev , the neutrino spectrum would actually be approximately degenerate .
this may be seen in detail by consulting the chart in table 1 of @xcite wherein the neutrino masses are tabulated as a function of an assumed value of the third neutrino mass , @xmath32 .
actually it is seen that there is also a region around @xmath33 0.04 ev and @xmath34 where an assumed initial degeneracy may be reasonable . to make physical sense out of such a scenario , it was suggested that the neutrino mass matrix be written as , @xmath35 where @xmath36 has the full @xmath14 invariance and has degenerate ( at least approximately ) eigenvalues .
furthermore , the smaller @xmath37 is invariant under a particular @xmath38 subgroup of @xmath14 and breaks the degeneracy .
finally , @xmath39 is invariant under a different @xmath38 subgroup of @xmath14 and is assumed to be smaller still .
the strengths are summarized as : @xmath40 this is inspired by the pre - qcd flavor perturbation theory of the strong interaction which works quite well . in that case the initially unknown strong interaction hamiltonian is expanded as @xmath41 here @xmath42 is the dominant @xmath43 flavor invariant piece , @xmath44 is the smaller gell - mann okubo perturbation @xcite which transforms as the eighth component of a flavor octet representation and breaks the symmetry to su(2 ) and @xmath45 , which transforms as a different component of the octet representation and breaks the symmetry further to the hypercharge u(1 ) , is smaller still .
there is a possible immediate objection to the assumption that the neutrino mass eigenvalues be degenerate in the initial s@xmath46 invariant approximation ; after all eq.([complexeigenvalues ] ) shows that there are two different eigenvalues @xmath18 and @xmath47 .
this was overcome by recognizing that these are both complex numbers and that they could both have the same magnitude but different directions .
having the same magnitude guarantees that all three physical masses will be the same .
this introduces a physical phase @xmath0 corresponding to the angle between @xmath18 and @xmath47 . in the strong interaction case , the initial su(3 )
invariance was found to be reasonably well obeyed .
it is thus natural to ask what predictions may exist in the initial @xmath14 invariant approximation in our neutrino model .
it was found @xcite that the leptonic factor for neutrinoless double beta decay , @xmath48 could be predicted in this limit to be , @xmath49 where @xmath50 is the degenerate neutrino mass and @xmath0is the majorana type phase mentioned above .
this led to the inequality @xmath51 the next step in the program is to consider the effect of the perturbation @xmath37 .
many authors @xcite-@xcite have suggested that a @xmath52-@xmath53 symmetry ( ( 23 ) symmetry in the present language ) is associated with tribimaximal mixing in the neutrino sector .
thus it is a natural @xmath38 symmetry choice for @xmath37 .
recently , chen and wolfenstein @xcite applied this type of perturbation to our present model with the additional assumption that the majorana phase @xmath0 takes the value @xmath54 .
this corresponds to cp conservation .
their result for @xmath55 is in agreement with the lower limit in eq.([inequality ] ) . here
we will investigate the first perturbed case without assuming that special value of @xmath0 . before going on to this
we will present an amusing argument to show that the ( 23 ) perturbation is naturally associated with the tribimaximal form ( modulo the majorana type phase @xmath0 ) rather than a tribimaximal form multiplied by a rotation in the two dimensional degenerate subspace ( which is physically irrelevant at the @xmath14 invariant level ) .
this is based on the fact that degenerate perturbation theory must be employed , which leads to a stability condition .
further we will show that other @xmath38 perturbations are mathematically consistent but do not lead to the desired tribimaximal form .
in the present framework there are three different possible perturbations , each characterized by the @xmath38 subgroup which remains invariant .
let us first consider the favored perturbation which leaves invariant the s@xmath56 subgroup , consisting of @xmath57 and @xmath58 .
apart from a piece which may be reabsorbed in eq.([solution ] ) , such a perturbation has the form , @xmath59 where @xmath60 and @xmath61 are parameters .
it is convenient to adopt the language of ordinary quantum mechanics perturbation theory .
we should then work in a basis like eq.([complexeigenvalues ] ) where @xmath20 in eq.([solution ] ) is diagonal .
however , because of the double degeneracy between the eigenvectors @xmath25 and @xmath27 in eq.([r ] ) , the matrix @xmath23 is not the unique one which diagonalizes @xmath20 .
we should really use the more general matrix @xmath62 where @xmath63 is given by : @xmath64 in this basis @xmath65 has the form : @xmath66 here , @xmath67 and @xmath68 . note that , before adding a perturbation , the @xmath14 symmetry predicts the lepton mixing matrix to be @xmath62 rather than the desired tribimaximal form , @xmath23 .
in perturbation theory , the first correction to the @xmath69 eigenvector involves the ratio @xmath70 . for degenerate perturbation theory it is of course necessary that the numerator vanishes for those states with @xmath71 .
here we simply require for the ( 13 ) matrix element : @xmath72 this yields in general , @xmath73 .
the solution with @xmath74 is the desired tribimaximal form .
the solution with @xmath75 just changes the signs of the first and third columns .
however , the solutions with @xmath76 and @xmath77 interchange the first and third columns , which does not agree with experiment .
thus , apart from a discrete ambiguity , the tribimaximal form is uniquely chosen when a smooth connection with the ( 23)-type perturbation is required .
of course , the smooth connection corresponds to choosing the correct initial states for the perturbation treatment .
it is easy to see that perturbations which leave the other two @xmath38 subgroups invariant , do not lead to mixing matrices of the desired tribimaximal form .
the perturbation which commutes with @xmath78 is : @xmath79 similarly , the perturbation which commutes with @xmath80 has the form : @xmath81 the stability condition for obtaining the tribimaximal mixing for the @xmath82 pertubation would require the matrix element @xmath83 to vanish ; instead it takes the value @xmath84 .
similarly , the stability condition for the @xmath85 pertubation does not work since the matrix element @xmath86 takes the generally non - zero value @xmath87 .
while we have seen that the stability condition for ( 23 ) invariant perturbations enforces the experimentally plausible tribimaximal mixing , the underlying @xmath14 symmetry should allow characteristic stable mixing matrices to emerge for either the ( 12 ) invariant or ( 13 ) invariant perturbations .
what are their forms ? in the case of a ( 12 ) perturbation , the stability condition associated with degenerate
perturbation theory reads : @xmath88 here the characteristic mixing matrix emerges as @xmath89 for a suitable value of @xmath90 .
the solution is easily seen to have the form : @xmath91 in the case of a ( 13 ) invariant perturbation , the stability condition associated with degenerate perturbation theory reads : @xmath92 here the characteristic stable mixing matrix turns out to be : @xmath93 the situation is summarized in table [ firsttable ] .
mathematically , any of the three perturbations will result in a stable mixing matrix .
however , only the ( 23 ) perturbation gives the experimentally allowed tribimaximal form .
for example ,
we see that the zero value of @xmath94 , in good present agreement with experiment , only holds for the @xmath65 [ ( 23)-type ] perturbation . .characteristic , stable mixing matrices for each @xmath38 invariant perturbation . [ cols="^,^",options="header " , ]
in order to go further we adopt convenient conventions for the , in general , complex parameters @xmath18 and @xmath1 defined in eq.([solution ] ) .
the goal is to adjust a phase , @xmath0 in order that the zeroth order spectrum has three exactly degenerate neutrinos . as shown in fig .
[ triangle ] , we take the 2-vector 3@xmath1 to be real positive . then the 2-vector @xmath18 lies in the third quadrant as : @xmath95 where the physical phase @xmath0 lies in the range : @xmath96 finally @xmath97 is related to @xmath1 by , @xmath98 in the limiting case @xmath99 , @xmath18 takes the real value , @xmath100
for simplicity we will consider the parameters @xmath60 and @xmath61 in eq.([23pert ] ) to be real rather than complex .
the entire neutrino mass matrix to first order is @xmath101 . since we are working in a basis where the zeroth order piece is diagonalized by the tribimaximal matrix , @xmath23 , we must diagonalize the matrix : @xmath102 diagonalizing the upper left 2 x 2 sub - matrix yields the three , in general , complex eigenvalues : @xmath103 where we introduced the abbreviation , @xmath104 .
the indicated approximations to the exact eigenvalues correspond to working to first order in the parameters t and u. remember that according to our original setup , t and u are supposed to be small compared to @xmath97 and @xmath1 . since fig .
[ triangle ] shows that generally @xmath105 , it is sufficient that @xmath106 and @xmath107 be small compared to @xmath1 . in this approximation
the corresponding eigenvectors are the columns of , @xmath108 the entire diagonalization may be presented as , @xmath109 here @xmath110 , @xmath111 and @xmath32 are the three ( positive ) neutrino masses and @xmath112 is the full neutrino mixing matrix ( in a basis where the charged leptons are diagonal ) .
the neutrino masses , to order @xmath113 , are seen to be : @xmath114 , \nonumber \\ m_2&\approx & \frac{3\beta}{2}csc(\frac{\psi}{2})\left[1 + \frac{4}{9\beta } ( t+u)sin^2(\frac{\psi}{2 } ) \right ] , \nonumber \\ m_3&\approx & \frac{3\beta}{2}csc(\frac{\psi}{2})\left[1- \frac{6}{9\beta } ( t - u)sin^2(\frac{\psi}{2})\right ] .
\label{numasses}\end{aligned}\ ] ] these mass parameters were made real , positive by the introduction of the phase matrix : @xmath115 where , @xmath116 \nonumber \\ \sigma&\approx & \pi-\frac{1}{2}tan^{-1 } [ \frac{cot(\psi/2)}{1+\frac{4(t+u)}{9\beta } } ] \nonumber \\ \rho&\approx & \frac{\pi}{2}+\frac{1}{2}tan^{-1 } [ \frac{cot(\psi/2)}{1-\frac{2(t - u)}{3\beta } } ] .
\label{majphases}\end{aligned}\ ] ] to compare with experiment , we have important information from neutrino oscillation experiments @xcite-@xcite .
it is known that @xcite @xmath117 also , constraints on cosmological structure formation yield @xcite a rough bound , @xmath118 the two allowed spectrum types are : @xmath119 .
now , from eq.([numasses ] ) we see to leading order : @xmath120 the quantities @xmath121 and @xmath122 may thus be obtained for a type 1 spectrum as : @xmath123 where the central experimental values were used . in the type 2 spectrum case
, we should change @xmath124 in the above to find , @xmath125 thus the , assumed real , @xmath14 violation parameters @xmath121 and @xmath122 are now known for each spectrum type .
information about the quantity @xmath1 may in principle be obtained from the perturbed lepton mixing matrix given in eq .
( [ fullmixingmat ] ) : @xmath126 with a usual parameterization @xcite the matrix with zero ( 13 ) element takes the form , @xmath127 where @xmath128 is short for @xmath129 for example .
this amounts to the predictions , @xmath130 notice that , when the perturbation is absent , this agrees with the tribimaximal form used here if both @xmath131 and @xmath132 lie in the second quadrant .
the results of a recent study @xcite of neutrino oscillation experiments are : @xmath133 one immediately notices that the prediction , @xmath134=1/2 is unchanged from its tribimaximal value by the perturbation and agrees with the new analysis . on the other hand
the tribimaximal prediction , @xmath135=1/3 is slightly changed from its tribimaximal value and actually lies slightly above the upper experimental error bar .
this is probably not a serious disagreement but it might be instructive to try to fix it using the predicted perturbation in the present model : @xmath136 for either the type 1 or type 2 assumed spectrum , the perturbation is seen to be in the correct direction to lower the value of @xmath137 , as desired .
however , because of the large cancellation between @xmath138 and @xmath122 , this effect is extremely small for a reasonable value of @xmath139 ; even with @xmath1 as small as 0.05 ev , @xmath135 is only lowered to 0.332 .
it is also interesting to discuss the absolute masses of the neutrinos rather than just the differences of their squares .
since the differences are known , let us focus on one of them , say @xmath32 : @xmath140 notice that the first term on the right hand side is , using eq.([alphabeta ] ) , simply the zeroth order degenerate mass , @xmath97 while the second term represents the correction . also note that ( see fig .
[ triangle ] ) the point @xmath141 is not allowed . considering @xmath0 as a parameter ( related to the strength of neutrinoless double beta decay )
, this equation represents a quadratic formula giving @xmath1 in terms of the absolute mass @xmath32 for any assumed @xmath0 . in fig .
[ pregion ] , adopting the criterion that @xmath142 and @xmath143 be less than 1/5 for perturbative behavior , we display the perturbative region in the @xmath144 plane .
in contrast to the case of @xmath32 , @xmath110 and @xmath111 are seen to have small corrections since they of course depend on @xmath145 rather than @xmath146 .
the characteristic physical novelty of of a theory with majorana type neutrinos is the prediction of a small , but non - zero , rate for the neutrinoless double beta decay of a nucleus : ( a , z ) @xmath147 ( a , z+2 ) + 2@xmath148 .
the appropriate leptonic factor describing the amplitude for this process is , @xmath149 substituting in the neutrino masses to order @xmath113 from eq.([numasses ] as well as eq.([phasematrix ] ) yields : @xmath150 e^{2i(\tau-\sigma})| . \label{meetwo}\ ] ] the needed intermediate quantity @xmath151 $ ] may be easily obtained from eqs.([majphases ] ) by construction of a suitable right triangles to be : @xmath152 \approx cos \psi -\frac{(t+u)sin\psi}{9\beta}. \nonumber\ ] ] we then find , correct to first order in @xmath113 , @xmath153 which is just the zeroth order result . the experimental bound on @xmath55 is given @xcite as , @xmath154 which is small enough so that there is hope the possibility of a majorana neutrino might be settled in the near future . since the correction to @xmath55 has been seen to be zero in this model we can take over the zeroth order inequality in eq.([inequality ] ) .
this means that the existence of the majorana phase , @xmath0 can alter the amplitude for neutrinoless double beta decay by a factor of three for given ( approximately degenerate ) neutrino masses .
in some ways the problem of flavor " in the standard model is reminiscent of that in strong interaction physics before the quark model . at that time
it was realized that , as a precursor to detailed dynamics , group theory might give important clues .
then the strong interactions were postulated to be su(3 ) flavor invariant with a weaker piece having just the the su(2 ) isospin ( times hypercharge ) invariance .
in addition it was known that there was a still weaker isospin breaking ( possibly qed ) which by itself preserved a different su(2 ) invariance ( so - called u - spin ) . here ,
an analogy for neutrinos of the first two steps was studied in a perturbative framework . in @xcite and @xcite
the possibility that the neutrinos are not strictly degnerate at the unperturbed level was contemplated .
however in this paper we have examined a strictly degenerate first stage ( setting to zero the parameters called respectively @xmath155 and @xmath156 in those two papers ) . at the @xmath14 invariant level
the neutrino mixing matrix is actually arbitrary up to a rotation in a 2-dimensional subspace .
this problem can be settled ( since degenerate perturbation theory is involved ) by specifying the transformation property of the perturbation to be added . although there is widespread agreement that the first perturbation should preserve the @xmath38 subgroup which interchanges the second and third neutrinos , we presented in section ii , for completeness and interest , the mixing matrices for the other two possibilities also . in sections iii and iv we carried out the perturbation analysis for any choice of a majorana - type phase , @xmath0 which plays an important role in this model .
if @xmath0 is considered fixed there are three parameters in the model ( in @xcite @xmath0 was considered fixed at the value @xmath54 . )
these three parameters can be taken as @xmath121 , @xmath122 and @xmath157 defined above .
the quantities @xmath121 and @xmath122 were found in terms of the neutrino squared mass differences for each choice of neutrino spectrum type , ie normal or inverted hierarchy .
the value of @xmath1 depends on the presently unknown absolute value of any neutrino mass .
the magnitudes of @xmath121 and @xmath122 are similar ( though not exactly equal ) but differ in sign .
thus the perturbation corrections which involve ( @xmath138 + @xmath122 ) are very small .
clearly ( see the first of eqs.([deltanus ] ) ) this is due to the small solar neutrino mass difference .
this situation occurs for the correction to the mixing parameter @xmath158 in addition to @xmath110 and @xmath111 , the masses of the first two neutrinos . the perturbation dependence on ( @xmath138 -@xmath159 is not suppressed however .
this occurs for the mass , @xmath32 of the third neutrino .
this result was used to make a sketch of the region in the @xmath0-@xmath32 plane for which the perturbation approach given seems numerically reasonable .
the explicit role of the higgs sector , which is believed to be at the heart of the matter , was not discussed in the present paper . however , this as well as some further technical details were discussed in @xcite .
further treatment of this aspect is interesting for future work as is a detailed investigation of the weakest perturbation , the analog of the u - spin preserving perturbation in the strong interaction .
this could be used to further study other consequences @xcite of possibly non - zero @xmath160 .
we are happy to thank amir fariborz , salah nasri and francesco sannino for helpful discussions and encouragment . the work of j. schechter and m.n .
shahid was supported in part by the us doe under contract no .
de - fg-02 - 85er 40231 ; they would also like to thank the cp@xmath161-origins group at the university of southern denmark for their warm hospitality and partial support .
s. pakvasa and h. sugawara , phys .
b * 73 * , 61 ( 1978 ) ; * 82 * , 105 ( 1979 ) ; e. derman and h.s.tsao , phys . rev .
d * 20 * , 1207 ( 1979 ) and y. yamanaka , h. sugawara and s. pakvasa phys .
d * 25 * , 1895 ( 1982 ) .
t. fukuyama and h. nishiura , hep - ph/9702253 ; r. n. mohapatra and s. nussinov , phys .
d 60 * , 013002 ( 1999 ) ; e. ma and m. raidal , phys .
lett . * 87 * , 011802 ( 2001 ) ; c. s. lam , hep - ph/0104116 ; t. kitabayashi and m. yasue , phys.rev . * d67 * 015006 ( 2003 ) ; w. grimus and l. lavoura , hep - ph/0305046 ; 0309050 ; y. koide , phys.rev .
* d69 * , 093001 ( 2004);y .
h. ahn , sin kyu kang , c. s. kim , jake lee , hep - ph/0602160 ; a. ghosal , hep - ph/0304090 ; w. grimus and l. lavoura , phys .
b * 572 * , 189 ( 2003 ) ; w. grimus and l. lavoura , j. phys .
g * 30 * , 73 ( 2004 ) .
w. grimus , a. s.joshipura , s. kaneko , l. lavoura , h. sawanaka , m. tanimoto , hep - ph/0408123 ; r. n. mohapatra , jhep , * 0410 * , 027 ( 2004 ) ; a. de gouvea , phys.rev .
* d69 * , 093007 ( 2004 ) ; r. n. mohapatra and w. rodejohann , phys . rev . *
d 72 * , 053001 ( 2005 ) ; t. kitabayashi and m. yasue , phys .
b 621 * , 133 ( 2005 ) ; r. n. mohapatra and s. nasri , phys .
d * 71 * , 033001 ( 2005);r .
n. mohapatra , s. nasri and h. b. yu , phys.lett .
b * 615 * , 231 ( 2005 ) .
see , for example , eq .
( 10 ) of s.s .
masood , s. nasri and j. schechter , phys .
d * 71 * , 093005 ( 2005 ) .
this reference also discusses a more symmetrical parameterization which may be convenient for treating neutrinoless double beta decay in general . | we study the effects of the perturbation which violates the permutation symmetry of three majorana neutrinos but preserves the well known ( 23 ) interchange symmetry .
this is done in the presence of an arbitrary majorana phase @xmath0 which serves to insure the degeneracy of the three neutrinos at the unperturbed level . |
the nonlinear schrdinger ( nls ) equation , as one of the universal equations that describe the evolution of slowly varying packets of quasi - monochromatic waves in weakly nonlinear dispersive media , has been very successful in many applications such as nonlinear optics and water waves @xcite .
the nls equation is integrable and can be solved by the inverse scattering transform @xcite . however , in the regime of ultra - short pulses where the width of optical pulse is in the order of femtosecond ( @xmath1 s ) , the nls equation becomes less accurate @xcite .
description of ultra - short processes requires a modification of going beyond the standard slow varying envelope approximation ( svea ) .
recently , schfer and wayne derived a short pulse ( sp ) equation @xmath2 in attempting to describe the propagation of ultra - short optical pulses in nonlinear media @xcite . here , @xmath3 is a real - valued function , representing the magnitude of the electric field , the subscripts @xmath4 and @xmath5 denote partial differentiation .
it has been shown that the sp equation performs better than nls under this case @xcite .
apart from the context of nonlinear optics , the sp equation has also been derived as an integrable differential equation associated with pseudospherical surfaces @xcite .
the sp equation has been shown to be completely integrable @xcite .
the loop soliton solutions as well as smooth soliton solutions of the sp equation were found in @xcite .
the connection between the sp equation and the sine - gordon equation through the hodograph transformation was clarified , and then the @xmath0-soliton solutions including multi - loop and multi - breather ones were given in matsuno_spe , matsuno_spereview by using hirota s bilniear method @xcite .
an integrable discretization of the sp equation was constructed by the authors in @xcite , and its geometric interpretation was given in @xcite .
a major simplification made in the derivation of the short pulse equation is to assume that the polarization is preserved during its propagating inside an optical fiber .
however , this is not always the case in practice .
for example , we have to take into account the effects of polarization or anisotropy in birefringent fibers @xcite .
therefore , an extension to a two - component version of the short pulse equation is needed in order to describe the propagation of ultra - short pulse in birefringent fibers .
in fact , several integrable coupled short pulse have been proposed in the literature pkb_cspe , hoissen_cspe , matsuno_cspe , feng_cspe , zengyao_cspe , comspe .
the bi - hamiltonian structures for several coupled short pulse equations were obtained in @xcite . in the present paper , we are concerned with the integrable semi - discretization of a multi - component short pulse ( mcsp ) equation @xmath6 where the coefficients @xmath7 are arbitrary constants with symmetry @xmath8 .
( [ cspe ] ) was proposed by matsuno @xcite through hirota s bilinear method , meanwhile , multi - soliton solution was given as well .
the remainder of the present paper is organized as follows . in section 2
, the mcsp equation is briefly reviewed .
we provide its @xmath0-soliton solution in an alternative pfaffian form and prove it by the pfaffian technique . in section 3 , by using a bcklund transformation of the bilinear equations and defining a discrete hodograph transformation , we construct a semi - discrete analogue of the mcsp equation .
meanwhile , @xmath0-soliton solution in terms of pfaffian is provided and proved . in section 4 ,
we investigate in detail the one- and two - soliton solutions to the semi - discrete complex short pulse equation , which can be reduced from the mcsp equation .
the paper is concluded by several remarks in section 5 .
it was shown by matsuno in @xcite that the sp equation ( [ spe ] ) is derived from bilinear equations @xmath9
\displaystyle d_{s}d_{y}\tilde{f}^{\prime } \cdot \tilde{f}^{\prime } = \frac{1% } { 2}\left ( { \tilde{f^{\prime } } } ^{2}-\tilde{f}^{2}\right ) \,,%
\end{array}% \right .
\label{spe_bilinear1}\]]through transformations @xmath10here @xmath11 is called hirota @xmath11-operator defined by @xmath12recently , in view of the fact that the sp equation ( [ spe ] ) can also be derived from another set of bilinear equations @xmath13
\displaystyle d_{s}^{2}f\cdot f=\frac{1}{2}g^{2}\,,% \end{array}% \right .
\label{spe_bilinear2}\]]through transformations @xmath14matsuno @xcite constructed a multi - component generalization of the short pulse equation ( [ spe ] ) based on a multi - component generalization of bilinear equations ( [ spe_bilinear2 ] ) , which reads @xmath15 \displaystyle d_{s}^{2}f\cdot f=\frac{1}{2}\sum_{1\leq j < k\leq n}c_{jk}g_{j}g_{k}\,.% \end{array}% \right . \label{csp_bilinear}\ ] ] the set of bilinear equations ( [ spe_bilinear1 ] ) is actually obtained from a 2-reduction of the kp - toda hierarchy , which basically delivers only two tau - functions out of a sequence of the tau - functions .
furthermore , when these two tau - functions are made complex conjugate to each other , the bilinear equations ( [ spe_bilinear1 ] ) is converted into the sine - gordon equation @xmath16 via a transformation @xmath17 , which is further converted into the sp equation ( [ spe ] ) by a hodograph transformation . in @xcite , hirota and
one of the authors have shown that both the bilinear equations ( [ spe_bilinear1 ] ) and ( [ spe_bilinear2 ] ) derive the sine - gordon equation .
furthermore , the relations between tau - functions , which read @xmath18 were also presented . as mentioned previously , eqs .
( [ spe_bilinear1 ] ) originate from 2-reduction of single component kp - toda hierarchy , whereas , eqs .
( [ spe_bilinear2 ] ) come from @xmath19-reduction of two - component kp - toda hierarchy .
since they both belong to @xmath20 of the euclidean lie algebra @xcite , it is natural that they both derive the sp equation .
however , the latter can be easily extended to @xmath21-reduction of multi - component kp - toda hierarchy , which gives rise to the multi - component generalization of the short pulse equation . in what follows
, we will briefly review how the bilinear equations ( csp_bilinear ) determines a multi - component generalization of the sp equation([spe ] ) . dividing both sides by @xmath22 ,
the bilinear equations ( [ csp_bilinear ] ) can be cast into @xmath23 \displaystyle\left ( \ln f\right ) _
{ ss}=\frac{1}{4}\sum_{1\leq j
< k\leq n}c_{jk}\frac{g_{j}g_{k}}{f^{2}}\,.% \end{array}% \right . \label{csp_bl2}\]]introducing a hodograph transformation @xmath24and a dependent variable transformation @xmath25we then have @xmath26which implies @xmath27 by letting @xmath28 . notice that the first equation in ( [ csp_bl2 ] ) can be rewritten as @xmath29 or @xmath30 which is converted into @xmath31 by the conversion relation ( [ cov - relation ] ) . obviously , eq . ( [ cpe1 ] ) is nothing but the mcsp ( [ cspe ] ) .
next , we give an alternative representation of the @xmath0-solution to the mcsp equation ( [ cspe ] ) in the form of pfaffians . to this end , let us define a class of set @xmath32 , @xmath33 , which satisfies the following condition , @xmath34then we define the elements of the pfaffians ( others not mentioned below are zeros ) @xmath35@xmath36@xmath37 @xmath38 here @xmath39 , @xmath40 , @xmath41 and @xmath42 is an integer . by defining the elements of the pfaffians
, we can give the pfaffian solutions satisfying bilinear equations ( [ csp_bilinear ] ) .
the bilinear equations ( [ csp_bilinear ] ) have the following pfaffian solution @xmath43 where @xmath44 and the elements of the pfaffians are given in eqs .
( [ cspe_pf1])([cspe_pf4 ] ) . since @xmath45 @xmath46 @xmath47 @xmath48 where @xmath49 for integers @xmath42 and @xmath50
, we then have @xmath51 @xmath52 @xmath53 @xmath54 here @xmath55 is abbreviated by @xmath56 , so as other similar pfaffians .
furthermore , it can be shown @xmath57 \\ & & = \sum_{j=1}^{2n } ( -1)^{j } \left [ \left ( { \partial_y } \mathrm{pf } ( d_0 , a_j ) \right ) \mathrm{pf } ( \beta_i , \cdots , \hat{a}_j , \cdots ) + \mathrm{pf } ( d_0 , a_j ) { \partial_y } \mathrm{pf } ( \beta_i , \cdots , \hat{a}_j , \cdots ) % \right ] \\ & & = \sum_{j=1}^{2n } ( -1)^{j } \left [ \mathrm{pf } ( d_1 , a_j ) \mathrm{pf } ( \beta_i , \cdots , \hat{a}_j , \cdots ) + \mathrm{pf } ( d_0 , a_j ) \mathrm{pf } ( \beta_i , d_0 , d_1 , \cdots , \hat{a}_j , \cdots ) \right ] \\ & & = \mathrm{pf } ( d_1 , \beta_i , \cdots)+ \mathrm{pf } ( d_0 , \beta_i , d_0 , d_1 , \cdots ) \\ & & = \mathrm{pf } ( d_1 , \beta_i , \cdots)\,.\end{aligned}\ ] ] here @xmath58 means that the index @xmath59 is omitted .
similarly , we can show @xmath60 @xmath61 \\ & & = \sum_{j=1}^{2n } ( -1)^{j } \left [ \left ( { \partial_y } \mathrm{pf } ( d_{-1 } , a_j ) \right ) \mathrm{pf } ( \beta_i , \cdots , \hat{a}_j , \cdots ) + \mathrm{pf } ( d_{-1 } , a_j ) { \partial_y } \mathrm{pf } ( \beta_i , \cdots , \hat{a}% _ j , \cdots ) \right ] \\ & & = \sum_{j=1}^{2n } ( -1)^{j } \left [ \mathrm{pf } ( d_0 , a_j ) \mathrm{pf } ( \beta_i , \cdots , \hat{a}_j , \cdots ) + \mathrm{pf } ( d_{-1 } , a_j ) \mathrm{pf } ( \beta_i , d_0 , d_1 , \cdots , \hat{a}_j , \cdots ) \right ] \\ & & = \mathrm{pf } ( d_0 , \beta_i,\cdots)+ \mathrm{pf } ( d_{-1 } , \beta_i , d_0 , d_1 , \cdots)\,.\end{aligned}\ ] ] an algebraic identity of pfaffian @xcite @xmath62 implies @xmath63 which is actually the first bilinear equation . the second bilinear equation is proved in a similar way as the one used by iwao and hirota @xcite in connection with a different system .
we start from the r.h.s of the bilinear equation .
@xmath64 next , the expansion of the vanishing pfaffian @xmath65 on @xmath66 yields @xmath67 which subsequently leads to @xmath68 similarly , we can show @xmath69 substituting eqs .
( [ csp1_proof2])([csp1_proof2 ] ) into eq .
( csp1_proof1 ) , we arrive at @xmath70 now we work on the l.h.s . of the second bilinear equation @xmath71 \nonumber \\ & & \quad \times \mathrm{pf } ( d_0 , \cdots , \hat{a}_i , \cdots ) \mathrm{pf } ( d_0 , \cdots , \hat{a}_j , \cdots ) \nonumber \\ & & = \sum_{i , j=1}^{2n } ( -1)^{i+j+1 } \left[p_i^{-2 } + p_i^{-1}p_j^{-1 } \right ] \mathrm{pf } ( a_i , a_j ) \mathrm{pf } ( d_0 , \cdots , \hat{a}_i , \cdots ) \mathrm{% pf } ( d_0 , \cdots , \hat{a}_j , \cdots ) \nonumber\end{aligned}\ ] ] the summation over the second term within the bracket vanishes due to the fact that @xmath72 therefore , @xmath73 \nonumber \label{csp1_proof5}\end{aligned}\ ] ] further , we note that the following identity can be substituted into the term within bracket @xmath74 which is obtained from the expansion of the following vanishing pfaffian @xmath75 on @xmath76 .
consequently , we have @xmath77\ , , \nonumber \\ & & = -\mathrm{pf } ( \cdots ) \mathrm{pf } ( d_{-2 } , d_0 , \cdots)+ \sum_{i=1}^{2n } p_i^{-2 } \mathrm{pf } ( d_0 , \cdots , \hat{a}_i , \cdots ) \mathrm{pf } ( d_0 , \cdots , \hat{b}_i , \cdots)\ , , \nonumber \\ & & = -\frac{\partial^2 f } { \partial s^2 } f+ \frac 14 \sum_{1 \le \mu < \nu \le n } c_{\mu \nu } g_{\mu } g_{\nu } \,,\end{aligned}\ ] ] which can be rewritten as @xmath78 the above equation is nothing but the second bilinear equation
. thus , the proof is complete .
in this section , we attempt to construct an integrable semi - discretization of the mcsp equation ( [ cspe ] ) .
firstly , we propose a semi - discrete analogue of bilinear equations ( [ csp_bilinear ] ) @xmath79 \displaystyle d_{s}^{2}f_{k}\cdot f_{k}=\frac{1}{2}\sum_{1\leq i < j\leq
n}c_{ij}g_{k}^{(i)}g_{k}^{(j)}\,.% \end{array}% \right . \label{eq : sdbl1}\ ] ] by introducing a dependent variable transformation @xmath80and a discrete version of the hodograph transformation @xmath81the second bilinear equation in ( [ eq : sdbl1 ] ) is rewritten as @xmath82from the discrete hodograph transformation , we can define an nonuniform mesh @xmath83it then immediately follows @xmath84from eq .
( [ sd_bl2 ] ) .
next , dividing both sides by @xmath85 , the first bilinear equation in ( [ eq : sdbl1 ] ) can be calculated out by @xmath86or @xmath87which is recast into @xmath88with the use of eqs .
( [ eq : v - trf ] ) and ( [ delta ] ) , we finally arrive at @xmath89 eqs .
( [ sd1-csp ] ) , ( [ sd2-csp ] ) constitute the semi - discrete analogue of the mcsp equation .
we summarize the results by the following theorem .
the bilinear equations ( [ eq : sdbl1 ] ) yield a semi - discrete multi - component short pulse equation @xmath90 \displaystyle \frac{d \delta_k}{ds } = -\frac{1}{2}\sum_{1\leq i < j\leq n}c_{ij } \left(u_{k+1}^{(i)}u_{k+1}^{(j)}-u_{k}^{(i)}u_{k}^{(j)}\right)\ , .
\end{array}% \right .
\label{eq : sd - msp}\ ] ] through dependent variable transformation @xmath91and discrete hodograph transformation @xmath92 where @xmath93 . to assure its integrability
, we provide its multi - soliton solution in terms of pfaffians by the following theorem .
the elements of the pfaffians are defined as follows : @xmath94 @xmath95 @xmath96 @xmath97 @xmath98 where @xmath99 here @xmath100 , @xmath40 and @xmath101 are arbitrary integers .
other pfaffian elements not mentioned above are all zeros .
note that @xmath102 has the following property @xmath103 which is used in the proof of the theorem . the bilinear equations ( [ eq : sdbl1 ] ) have the following pfaffian solution @xmath104where @xmath105 , the elements of pfaffians are defined in eqs .
( sd - pf1)([sd - pf5 ] ) .
since @xmath106 @xmath107 @xmath108 we have @xmath109 @xmath110 @xmath111 furthermore , we can verify @xmath112 @xmath113 @xmath114 therefore , an algebraic identity of pfaffian @xmath115together with above pfaffian relations gives @xmath116which is nothing but the first bilinear equation .
the second bilinear equation can be proved in a similar way as in the continuous case .
bilinear equations ( [ eq : sdbl1 ] ) can be viewed as a bcklund transformation of bilinear equations ( [ csp_bilinear ] ) , which yield the mcsp equation . in other words , if @xmath117 and @xmath118 satisfy ( [ csp_bilinear ] ) , so do @xmath119 and @xmath120 . based on ( [ eq : sdbl1 ] )
, we propose a semi - discrete analogue of the mcsp equation .
the integrability of the semi - discrete mcsp equation is guaranteed by the existence of @xmath0-soliton solution .
finally , let us show that in the continuous limit , @xmath121 ( @xmath122 ) , the proposed semi - discrete multi - component short pulse equation recovers the continuous one ( [ cpe1 ] ) .
the dependent variable @xmath123 is regarded as a function of @xmath5 and @xmath4 , where @xmath5 is the space coordinate of the @xmath124-th lattice point and @xmath4 is the time , defined by @xmath125 where @xmath126 . in the continuous limit , @xmath121 ( @xmath127 ) , we have @xmath128 @xmath129 hence @xmath130 where the origin of space coordinate @xmath131 is taken so that @xmath132 cancels @xmath133 .
thus the first semi - discrete multi - component sp equation converges to @xmath134 which is exactly the mcsp equation ( [ cspe ] ) .
since the two - component short pulse equation is of particular importance for applications in nonlinear optics , we provide a detailed study for this two - component system , together with its semi - discrete analogue in this section . for the continuous case of @xmath135
, we can take @xmath136 without loss of generality and arrive at the following two - component system @xcite @xmath137 where @xmath138 .
furthermore , if we assume @xmath123 is a complex - values function and impose a complex conjugate condition @xmath139 , where @xmath140 means the complex conjugate of @xmath123 .
( [ 2cspea1 ] ) leads to a * complex short pulse equation * studied in @xcite @xmath141 since the complex short pulse equation ( [ complex_sp ] ) is a special case of two - component system ( [ 2cspea1])([2cspea2 ] ) , its @xmath0-soliton solution can be obtained from the @xmath0-soliton solution of system ( 2cspea1)([2cspea2 ] ) by requiring @xmath142 , @xmath143 .
these requirements can be achieved by putting @xmath144 and @xmath145 , where @xmath146 , @xmath147 . in particular , the tau - functions for one - soliton solution ( @xmath148 )
are found to be @xmath149 @xmath150 let @xmath151 , and we assume @xmath152 without loss of generality , then the one - soliton solution can be expressed in the following parametric form @xmath153 @xmath154 where @xmath155 eq . ( [ csp1solitona ] ) represents an envelope soliton of amplitude @xmath156 and phase @xmath157 . the details analysis concerning its property
was carried out in @xcite . in summarythree types
can be classified . * when @xmath158
, it is a * smooth soliton * solution , which is similar to the envelope soliton solution for the nonlinear schrdinger equation .
* when @xmath159 , it is a * loop soliton * solution , which admits multi - valued property .
* when @xmath160 , it is a * cuspon soliton * , this is a case which divides the single - valued and multi - valued solution . based on the results in previous section , we have an integrable semi - discrete analogue of two - component system ( [ 2cspea1])([2cspea2 ] ) @xmath161which admit the following @xmath0-soliton solution in parametric form @xmath162 and hodograph transformation @xmath163 with @xmath164 @xmath165 where the elements of pfaffians are ( others not mentioned are all zeros ) @xmath166 @xmath167 @xmath168 by imposing complex conjugate conditions @xmath144 , @xmath145 ( @xmath146 , @xmath169 ) , it then follows @xmath170 , @xmath171 , thus @xmath172 , which leads to a semi - discrete analogue of the complex short equation ( [ complex_sp ] ) @xmath173its @xmath0-soliton solution immediately follows from the @xmath0-soliton solution of eqs .
( [ eq : twocomnsol1])([eq : twocomnsol2 ] ) under complex conjugate conditions mentioned above . in
what follows , we list the one- and two - soliton solutions .
* one - soliton solution : * the tau - functions for one - soliton solution to eq .
( [ sd_2comsp ] ) are @xmath174 with @xmath175 where @xmath176 , @xmath177 .
similar to the continuous case , if @xmath178 , we then arrive at the one - soliton solution of semi - discrete complex short pulse equation ( [ sd_2comsp ] ) @xmath179 @xmath180 where @xmath181 in fig . 1 ( a)(c )
, we illustrate the envelope soliton for @xmath182 , @xmath183 , @xmath184 , which correspond to the smooth , cuspon and loop solition , respectively in the continuous case . ,
( b ) cuspon soliton with @xmath185 ,
( c ) loop soliotn with @xmath186.,title="fig : " ] , ( b ) cuspon soliton with @xmath185 , ( c ) loop soliotn with @xmath186.,title="fig : " ] -0.3to + 0.3 , ( b ) cuspon soliton with @xmath185 , ( c ) loop soliotn with @xmath186 . ] -0.3to + 0.3 * two - soliton : * the tau - functions for two - soliton solutions to semi - discrete complex short pulse equation ( [ sd_2comsp ] ) can be obtained by @xmath187 @xmath188 where @xmath189 and @xmath190 and @xmath191 . in fig .
2 ( a)(d ) , we show the process of interaction between a smooth envelop soliton and a cuspon envelop soliton with @xmath192 , @xmath193 , @xmath194 , @xmath195 . , dashed line : @xmath196 .
( a ) @xmath197 ; ( b ) @xmath198 ; ( c ) @xmath199 ; ( d ) @xmath200.,title="fig : " ] , dashed line : @xmath196 . ( a ) @xmath197 ; ( b ) @xmath198 ; ( c ) @xmath199 ; ( d ) @xmath200.,title="fig : " ] -0.3to + 0.25 , dashed line : @xmath196 .
( a ) @xmath197 ; ( b ) @xmath198 ; ( c ) @xmath199 ; ( d ) @xmath200.,title="fig : " ] , dashed line : @xmath196 .
( a ) @xmath197 ; ( b ) @xmath198 ; ( c ) @xmath199 ; ( d ) @xmath200.,title="fig : " ] -0.3to + 0.25
we have derived an integrable semi - discrete analogue of the multi - component short pulse equation proposed by matsuno @xcite based on a bcklund transform and hirota s bilinear method .
we find its @xmath0-soliton solution in terms of pfaffians and prove it .
moreover , a complex short pulse equation , which possess smooth , cuspon or loop type envelop soliton , is proposed and its semi - discrete analogue is constructed as well .
we conclude the present paper by the following remarks .
* the @xmath0-solution for multi - component short pulse equation given in the present paper agrees with the one given by matsuno in @xcite .
this solution is a benchmark for the study of soliton interactions . *
similar to our previous results @xcite , the semi - discrete multi - component short pulse equation proposed here can be served as an integrable numerical scheme , the so - called self - adaptive moving mesh method , for the numerical simulation of multi - component short pulse equation , as well as complex and coupled complex short pulse equation .
however , it is obviously beyond the scope of the present paper , we would like to report our results in this aspect in a forthcoming paper .
* the integrable fully discretization of the multi - component short pulse equation is a further topic deserve to study .
this work is partially supported by the national natural science foundation of china ( no . 11428102 ) . | in the present paper , we mainly study the integrable semi - discretization of a multi - component short pulse equation .
firstly , we briefly review the bilinear equations for a multi - component short pulse equation proposed by matsuno ( j. math . phys . * 52 * 123705 ) and reaffirm its @xmath0-soliton solution in terms of pfaffians . then by using a bcklund transformation of the bilinear equations and defining a discrete hodograph ( reciprocal ) transformation ,
an integrable semi - discrete multi - component short pulse equation is constructed . meanwhile , its @xmath0-soliton solution in terms of pfaffians is also proved . |
the so - called acceleration theorem for wave - packet motion in periodic potentials , formulated already in 1928 by bloch , @xcite has proven to be of outstanding value to solid - state physics for understanding the dynamics of bloch electrons within a semiclassical picture .
@xcite in its most - often used variant , this theorem states that if we consider an electronic wave packet in a spatially periodic lattice , which is centered in @xmath0 space around some wave vector @xmath1 , and if an external electric field @xmath2 is applied under single - band conditions , then this center wave vector evolves in time according to @xmath3 , with @xmath4 being the electronic charge .
perhaps its best - known application is the explanation of bloch oscillations of particles exposed to a homogeneous , constant force , @xcite which we recapitulate here in the simplest guise : take a particle in a one - dimensional tight - binding energy band @xmath5 , where @xmath6 is the band width and @xmath7 denotes the lattice period .
assume that the particle s wave packet is centered around @xmath8 initially and subjected to a homogeneous force of strength @xmath9 .
then the acceleration theorem , now taking the form @xmath10 tells us @xmath11 , so that the packet moves through @xmath0 space at a constant rate .
@xcite according to another classic work by jones and zener , @xcite the particle s group velocity @xmath12 in real space is determined , quite generally , by the derivative of @xmath13 with respect to @xmath0 when evaluated at the moving center @xmath14 , @xmath15 in our case , this relation immediately gives @xmath16 implying that the particle s response to the constant force is an oscillating motion with the bloch frequency @xcite @xmath17 .
this elementary example , to which we will come back later in sec .
[ sec : s_4 ] , strikingly illustrates the power of this type of approach .
but an obvious restriction stems from the necessity to remain within the scope of the single - band approximation ; the above acceleration theorem ( [ eq : oat ] ) is put out of action when several bloch bands are substantially coupled by the external force .
nonetheless , in the present work we demonstrate that there exists a generalization of the acceleration theorem which can be applied even under conditions of strong interband transitions .
specifically , we consider situations in which a bloch particle is subjected to a strong oscillating force which possibly induces pronounced transitions between the unperturbed energy bands . by abandoning the customary crystal - momentum representation @xcite and introducing an alternative floquet representation instead , we show that the effect of an additional force then is well captured by another acceleration theorem which closely mimics the spirit of the original .
we obtain two major results : the floquet analog ( [ eq : gat ] ) of bloch s acceleration theorem ( [ eq : oat ] ) , and the floquet analog ( [ eq : fgv ] ) of the jones - zener expression ( [ eq : ogv ] ) for the group velocity .
these findings are particularly useful for control applications , when a strong oscillating field `` dresses '' the lattice and thus significantly alters its band structure , while a second , comparatively weak homogeneous force is employed to effectuate controlled population transfer .
we first outline the formal mathematical arguments in secs .
[ sec : s_2 ] and [ sec : s_3 ] , and then we give two applications of topical interest , discussing `` super '' bloch oscillations in sec .
[ sec : s_4 ] and coherently controlled interband population transfer in sec .
[ sec : s_5 ] .
although we restrict ourselves here for notational simplicity to one - dimensional lattices , our results can be carried over to general , higher - dimensional settings .
we consider a particle of mass @xmath18 moving in a one - dimensional lattice potential @xmath19 with spatial period @xmath7 under the influence of a homogeneous , time - dependent force @xmath20 , as described by the hamiltonian @xmath21 subjecting the particle s wave function @xmath22 to the unitary transformation @xmath23 the new function @xmath24 obeys the schrdinger equation @xmath25 with the transformed hamiltonian @xmath26 now let us further assume that the force @xmath20 is periodic in time with period @xmath27 , such that its one - cycle integral either vanishes or equals an integer multiple of @xmath28 times the reciprocal lattice wave number @xmath29 : @xmath30 for example , this is accomplished by a monochromatic oscillating force with an additional static bias , @xmath31 provided the latter satisfies the condition @xmath32 .
then the floquet theorem guarantees that the time - dependent schrdinger equation ( [ eq : tse ] ) admits a complete set of spatiotemporal bloch waves , @xcite that is , of solutions of the form @xmath33 with spatially _ and _ temporally periodic functions @xmath34 as usual , @xmath35 is the band index and @xmath0 a wave number ; @xmath36 thus is the quasienergy dispersion relation for the @xmath35th band .
if @xmath37 in eq .
( [ eq : res ] ) , the existence of these solutions is obvious , because then @xmath38 , so that the wave functions ( [ eq : stb ] ) generalize the customary bloch waves @xcite for particles in spatially periodic lattice potentials by also accounting for the temporal periodicity of the driving force . when @xmath39 , so that @xmath40 itself is not periodic in time , spatiotemporal bloch waves ( [ eq : stb ] ) emerge nonetheless because @xmath0 is projected to the first quasimomentum brillouin zone , as first discussed by zak .
@xcite in any case , the quasienergies @xmath36 may depend in a complicated manner on the parameters of the driving force , and the wave functions @xmath41 pertaining to a single quasienergy band may be nontrivial mixtures of several unperturbed energy bands . for later use , we observe that their spatial parts @xmath42 obey the quasienergy eigenvalue equation @xmath43 as follows immediately when plugging the solutions ( [ eq : stb ] ) into the schrdinger equation ( [ eq : tse ] ) . throughout ,
we adopt the standard normalization @xmath44 an arbitrary wave packet @xmath24 may now be expanded with respect to these spatiotemporal bloch waves , and written in the form @xmath45 with @xmath46 denoting the fundamental brillouin zone .
the expansion coefficients @xmath47 depend on the way the system has been prepared and on the way the driving force has been turned on , whereas the basis functions @xmath48 and their quasienergies @xmath36 are given by the eigenvalue equation ( [ eq : qee ] ) and obviously are independent of such details .
clearly , one has @xmath49 so that the populations @xmath50 remain constant in time .
this expansion ( [ eq : flr ] ) , referred to as the floquet representation of the wave packet , is formally reminiscent of its customary crystal - momentum representation , that is , of an expansion with respect to the bloch states of the unperturbed potential @xmath51 which underlies the standard acceleration theorem .
@xcite there are , however , substantial differences which become most clear when considering a wave packet occupying a single quasienergy band , @xmath52 here and in the following , we omit the band index @xmath35 for ease of notation . now
this wave packet ( [ eq : sbp ] ) may describe , for instance , the dynamics in a situation where two unperturbed energy bands are resonantly coupled by the driving force @xmath20 ; consequently , in a crystal - momentum representation one would have to account for rabi - type oscillations between these two bands by coefficients which quantify the oscillating band populations . in the floquet respresentation ,
on the other hand , the rabi oscillations are already incorporated into the basis states ( [ eq : stb ] ) , so that one merely encounters single quasienergy band dynamics , with the remaining time evolution of @xmath53 simply given by eq .
( [ eq : dgt ] ) .
thus , although the external force effectuates transitions between the unperturbed bloch bands , there are no inter - quasienergy band transitions ; @xmath54 remains constant in time .
second , even in a situation where @xmath20 does not couple different energy bands , the wave packet s center @xmath14 evolves according to the standard acceleration theorem @xmath55 in the crystal - momentum representation , whereas in the floquet representation the moment @xmath56 obviously stays constant in time . in short ,
an expansion of the wave packet with respect to the spatiotemporal bloch waves ( [ eq : stb ] ) implies constant coefficients , and hence constant occupation probabilities , if the external force @xmath20 adheres to the specification ( [ eq : res ] ) .
this formal shift of the dynamics from the occupation numbers to the basis states which is implied by the floquet representation now allows for a clear and physically transparent description of the additional effects which emerge when the external force does _ not _ obey eq .
( [ eq : res ] ) ; these effects are captured by the generalized acceleration theorem exposed in the following .
we take a wave packet occupying a single quasienergy band and stipulate that in addition to the possibly strong driving force @xmath20 there is a second homogeneous force @xmath57 which we denote as the _ probe force _ ; this is assumed to be sufficiently weak so that it does not introduce transitions among different quasienergy bands . to be precise
, the total hamiltonian now reads @xmath58 where the time - periodic force @xmath20 is resonant in the sense of eq .
( [ eq : res ] ) and thus creates a basis of spatiotemporal bloch waves ( [ eq : stb ] ) , whereas the probe force @xmath57 also is spatially homogeneous , but not necessarily periodic in time . after performing the unitary transformation ( [ eq : unt ] )
, we obtain the hamiltonian in the form @xmath59 with @xmath40 given by eq .
( [ eq : uph ] ) .
moreover , we start from an initial wave packet of the form ( [ eq : sbp ] ) .
because of the additional probe force @xmath57 , the time evolution of @xmath53 is no longer given by eq .
( [ eq : dgt ] ) ; the aim now is to find an effective hamiltonian @xmath60 which governs the resulting dynamics of @xmath53 , under the proposition that this remains restricted to the single , initially occupied quasienergy band .
exploiting the normalization ( [ eq : nrm ] ) , we have @xmath61 this gives @xmath62 \varphi_k \right)^ * \psi \nonumber\\ & & - \sqrt{\frac{a}{2\pi } } { f_{\rm p}}\int \ !
{ { \rm d}}x \ , \varphi_k^ * x \psi \ ; , \end{aligned}\ ] ] having suppressed the arguments @xmath63 and @xmath64 for better legibility ; all integrals here are taken over the entire lattice . in the first term on the right - hand side of this equation we exploit the quasienergy eigenvalue equation , eq .
( [ eq : qee ] ) , yielding @xmath65 . for rewriting the second term we use @xmath66 which is obtained by taking the derivative of the complex conjugate to eq .
( [ eq : dfp ] ) with respect to @xmath0 , and leads to @xmath67 for making the final step , we have resubstituted the expression ( [ eq : sbp ] ) for @xmath68 and have made use of the identity @xmath69 with the scalar product @xmath70 being given by an integral over a single lattice period .
note that @xmath71 as an immediate consequence of eq .
( [ eq : nrm ] ) , which implies @xmath72 so that @xmath73 is purely imaginary .
collecting all the pieces , we obtain the desired evolution equation @xmath74 with the effective hamiltonian for the floquet representation , @xmath75 from this expression we deduce the generalized acceleration theorem , that is , the acceleration theorem for the floquet representation : since the moment ( [ eq : fra ] ) obeys the the equation @xmath76 \rangle\ ] ] and the commutator appearing here on the right - hand side is easily evaluated , @xmath77 = { f_{\rm p}}$ ] , we are directly led to @xmath78 this is the central result of the present work ; its analogy to the standard acceleration theorem ( [ eq : oat ] ) for the crystal - momentum representation is evident .
observe that there is an intuitively clear reason for the appearance of the term proportional to @xmath73 in the effective hamiltonian ( [ eq : efh ] ) : the twofold periodic parts @xmath79 of the spatiotemporal bloch waves are obtained by solving the eigenvalue equation ( [ eq : qee ] ) .
this is done for each wave number @xmath0 separately , so that one is free to bestow upon each eigensolution an arbitrary phase factor @xmath80 . on the other hand
, the evolution equation ( [ eq : eeg ] ) for the wave function @xmath53 in the floquet representation naturally establishes a `` connection '' between those different eigensolutions @xcite and therefore requires information about the gauge function @xmath81 ; this is provided by the expression @xmath73 . note
further that when multiplying eq .
( [ eq : eeg ] ) by @xmath82 and subtracting the complex conjugate of the resulting equation , this piece drops out , and one is left with @xmath83 thus , @xmath84 does not depend on @xmath0 and @xmath64 separately , but rather on the combination @xmath85 , so that the distribution @xmath53 moves through the floquet @xmath0 space without change of shape , again in precise analogy to the classic behavior .
@xcite but we reemphasize that this seemingly simple dynamics might be unrecognizable in the usual crystal - momentum representation , because the system might undergo violent transitions between different energy bands when monitored in a basis of time - independent bloch waves . as the introductory example has shown , the standard ( crystal - momentum ) acceleration theorem develops its main power in combination with the jones - zener expression ( [ eq : ogv ] ) for the wave packet s group velocity in real space , and the question naturally arises whether there exists a similar connection in the floquet representation . obviously , one can establish a relation corresponding to eq .
( [ eq : ogv ] ) by applying a stationary - phase argument to the expansion ( [ eq : flr ] ) , but here we follow an alternative line of reasoning which may be found particularly enlightening . considering a well - localized wave packet @xmath22 in the original frame of reference
to which the hamiltonian operators ( [ eq : hor ] ) and ( [ eq : hot ] ) pertain , that packet s group velocity is given by @xmath86 on the other hand , exploiting the operator identity @xmath87 the eigenvalue equation ( [ eq : qee ] ) transforms into the even more basic eigenvalue equation @xmath88 for the periodic core pieces @xmath89 of the spatiotemporal bloch waves ( [ eq : stb ] ) , invoking the parametrically @xmath0-dependent operator @xmath90 this eigenvalue problem can efficiently be implemented for numerical calculations .
@xcite it also manifestly contains the origin of the condition ( [ eq : res ] ) imposed on the oscillating force @xmath20 , since @xmath0 is reduced to fall within @xmath91 .
most importantly , this eigenvalue problem ( [ eq : epu ] ) poses itself in an extended hilbert space made up of functions @xmath89 which are periodic in both space and time , in accordance with eq .
( [ eq : dpu ] ) . consequently
, `` time '' has to be regarded as a coordinate in this extended hilbert space and therefore needs to be integrated over when forming a scalar product , just like any spatial coordinate .
thus , the natural scalar product in this extended hilbert space is given by @xcite @xmath92 with @xmath93 denoting the standard scalar product in the original , physical hilbert space , as already employed in eqs .
( [ eq : def ] ) and ( [ eq : nru ] ) .
it follows that the quasienergies @xmath36 can be written as diagonal elements of the matrix of the quasienergy operator , @xmath94 inviting us to make use of an analog of the hellmann - feynman theorem : @xcite @xmath95 in the final step we have undone the shift ( [ eq : udo ] ) ; the wave functions @xmath96 then denote the functions which are obtained from the spatiotemporal bloch waves ( [ eq : stb ] ) by inverting the transformation ( [ eq : unt ] ) .
comparison of eqs .
( [ eq : vgd ] ) and ( [ eq : vge ] ) , keeping in mind the definition ( [ eq : scp ] ) , now yields the desired relation : supposing that @xmath97 were made up from a single spatiotemporal bloch wave labeled by @xmath35 and @xmath98 , say , one would obtain the formal identity @xmath99 but this is not what we want , because an individual spatiotemporal bloch wave is uniformly extended over the lattice and thus does not correspond to a `` group '' which propagates in space .
rather , we require a wave packet ( [ eq : sbp ] ) which is reasonably well centered in the floquet @xmath0 space , with a center @xmath100 given by eq .
( [ eq : fra ] ) . then we have @xmath101 to good accuracy , so that the cycle - averaged group velocity of the floquet wave packet is given by the derivative of its quasienergy dispersion relation , evaluated at its center @xmath100 . again , this floquet relation ( [ eq : fgv ] ) closely mimics its historic crystal - momentum antecessor , given by eq .
( [ eq : ogv ] ) .
in contrast to the equation of motion ( [ eq : gat ] ) for @xmath100 itself , which holds exactly within a single quasienergy band setting , this relation ( [ eq : fgv ] ) is an approximation which holds the better , the narrower the packet s floquet @xmath0 space distribution .
although it seems self - evident , it might be worthwhile to stress that the argument required to evaluate the derivative ( [ eq : fgv ] ) is floquet @xmath100 , not crystal momentum @xmath102 .
the phenomenon termed `` super '' bloch oscillations @xcite arises when a bloch particle is subjected to both a static ( dc ) and an oscillating ( ac ) force , such that an integer multiple of the ac frequency is only slightly detuned from the bloch frequency associated with the dc component of the force .
@xcite although the effect itself appears almost trivial from the mathematical point of view , we nonetheless dwell on this at some length , because it provides a particularly instructive example for juxtaposing the familiar crystal - momentum representation to the floquet representation introduced in sec .
[ sec : s_2 ] and for demonstrating in detail how they match . to be definite ,
we consider the total force to be of the form @xmath103 \ ; , \label{eq : tto}\ ] ] where @xmath104 denotes the heaviside function , so that both the dc and the ac component of the force are turned on instantaneously and simultaneously at @xmath105 ; that moment @xmath105 thus determines the relative phase between the bloch oscillations caused by the dc component and the driving oscillations of the ac component .
the basic assumptions now are that _
( i ) _ we are given an initial wave packet which occupies a single energy band , being centered around @xmath106 at the moment @xmath107 in the crystal - momentum representation , and that _ ( ii ) _ interband transitions remain negligible for @xmath108 , despite the action of the force @xmath20 .
we then encounter single - band dynamics which are fully covered by the `` old '' acceleration theorem @xmath109 , giving @xmath110 \nonumber \end{aligned}\ ] ] for @xmath108 . as an archetypal example we now take a tight - binding cosine energy dispersion relation for the band considered , @xmath111 parametrized as in the introduction .
utilizing eq .
( [ eq : ogv ] ) , one then finds the packet s group velocity : @xmath112 this expression describes super bloch oscillations if we assume further that the dc component of the force is almost resonant in the sense of eq .
( [ eq : res ] ) .
we therefore decompose this component according to @xmath113 where @xmath114 with some nonzero integer @xmath115 as previously in eq .
( [ eq : int ] ) , so that @xmath116 equals the bloch frequency @xmath117 , while @xmath118 is quite small compared to @xmath119 .
we then have @xmath120 with frequency detuning @xmath121 , so that the group velocity ( [ eq : igv ] ) takes the form @xmath122 having introduced the scaled driving amplitude @xmath123 and a constant phase @xmath124 which accounts for the initial conditions . because @xmath125 according to our specifications , the contribution @xmath126 to the argument of @xmath127 does not vary appreciably during one single cycle @xmath128 of the ac component .
thus , when averaging the instantaneous group velocity over one such cycle , this `` slow '' time dependence may be ignored , meaning that @xmath126 may be considered as constant when taking the average .
@xcite invoking the jacobi - anger indentity in the guise @xmath129 where @xmath130 denote the bessel functions of the first kind , one immediately obtains @xmath131 according to the above reasoning , here the `` fast '' time dependence is integrated out , but the slow dependence on @xmath126 remains .
@xcite integrating , this yields the cycle - averaged drift motion of the packet , that is , its position @xmath132 without the fast ac quiver , @xmath133 with a suitably chosen origin of the @xmath63 axis .
this result finally clarifies what is `` super '' with these dynamics : because the residual force @xmath118 is quite small , the amplitude of this oscillation ( [ eq : sbo ] ) can be fairly large ; indeed , in a corresponding experiment with weakly interacting bose - einstein condensates in driven optical lattices haller _ et al .
_ have observed giant center - of - mass oscillations with displacements across hundreds of lattice sites .
@xcite as far as the phenomenon itself is concerned there is nothing more to add ; because one requires single bloch - band dynamics right from the outset , a floquet treatment is not necessary .
nevertheless the floquet approach is of its own intrinsic value even here , since it provides a somewhat different view which , in contrast to the above crystal - momentum calculation , is capable of some generalization .
the floquet analysis starts from the spatiotemporal bloch waves and their quasienergies . in a single - band setting with an external homogeneous force ,
these are exceptionally easy to obtain : writing the bloch waves of the undriven lattice in the form @xmath134 where @xmath135 denotes a wannier function localized around the @xmath136th lattice site , @xcite the so - called houston functions @xcite @xmath137 are solutions to the time - dependent schrdinger equation in the original frame , for arbitrary @xmath20 , provided the `` moving wave numbers '' @xmath138 are given by @xmath139 always assuming the viability of the single - band approximation .
@xcite taking a force of the particular form ( [ eq : int ] ) with exactly resonant @xmath119 obeying @xmath32 , we have @xmath140 this implies that both the exponentials @xmath141 and @xmath142 are @xmath27 periodic in time , with @xmath128 , whereas the integral over @xmath142 is not , because the fourier expansion of @xmath142 contains a zero mode , so that its integral contains a linearly growing contribution .
but this observation reveals that the `` accelerated bloch waves '' ( [ eq : hou ] ) with resonant time - periodic forcing ( [ eq : int ] ) are precisely the required spatiotemporal bloch waves in the original frame , with their quasienergies being determined by the zero mode : @xmath143 the remarkable fact that the quasienergy bands collapse , _ i.e. _ , become flat when @xmath144 is such that @xmath145 , indicates that an oscillating force can effectively shut down the tunneling contact between neighboring wells ; this `` coherent destruction of tunneling '' is a generic feature of driven single - band systems .
@xcite a bit of reflection then shows that the core pieces @xmath79 of the spatiotemporal bloch waves , that is , the solutions to the eigenvalue equation ( [ eq : epu ] ) , are given by @xmath146 \right ) \ ; .
\end{aligned}\ ] ] although this has not been particularly emphasized , the above construction makes sure that any spatiotemporal bloch wave ( [ eq : hou ] ) is labeled by the same wave number @xmath0 as the ordinary bloch wave to which it reduces when the external force vanishes .
@xcite otherwise , there is nothing particular about the choice @xmath147 for the lower bound of integration in eq .
( [ eq : qkt ] ) for @xmath138 : in contrast to eq .
( [ eq : tto ] ) , where @xmath107 has been singled out as the moment when the force is turned on , and which thus designates an initial - value problem for a particular wave packet , the solution of the eigenvalue problem ( [ eq : epu ] ) for the entire spatiotemporal bloch basis requires a force @xmath20 which is perfectly periodic in time ; the resulting expression for @xmath138 thus holds for both @xmath148 and @xmath149 . also note that it would be meaningless to include some additional constant phase into the argument of the ac component of the force ( [ eq : int ] ) : because this expression holds for all times @xmath64 , such a phase would merely amount to a shift of the origin of the time coordinate and thus is as irrelevant for the calculation of the quasienergy dispersion relation as would be a shift of the origin of the spatial coordinate system for the calculation of a crystal s energy band structure .
knowing the quasienergy dispersion relation ( [ eq : qed ] ) , the machinery established in sec . [ sec : s_3 ] can be set to work : according to eq .
( [ eq : fgv ] ) , the cycle - averaged group velocity of a floquet wave packet ( [ eq : sbp ] ) is given by @xmath150 if we now turn back to the specific forcing ( [ eq : tto ] ) , and thus consider exactly the same initial - value problem as in the previous crystal - momentum exercise , we can make operational use of the decomposition ( [ eq : dec ] ) of the dc force : its resonant part @xmath119 has already been incorporated into the spatiotemporal bloch waves ( [ eq : hou ] ) , which means that it has already been accounted for in `` dressing '' the lattice and changing its original energy dispersion @xmath13 to the quasienergy dispersion @xmath151 .
therefore , it is only the small residual part @xmath118 which enters into the equation of motion for @xmath100 , that is , into the generalized acceleration theorem ( [ eq : gat ] ) ; this part @xmath118 thus constitutes a particular , time - independent example of a probe force @xmath57 as considered in sec .
[ sec : s_3 ] .
we now have @xmath152 giving @xmath153 all that remains to be done now is to express the initial floquet center @xmath154 in terms of the initial wave packet s center @xmath106 , which had been specified in the crystal - momentum representation .
but this is an easy task , comparing the original bloch waves ( [ eq : blo ] ) to their spatiotemporal descendents ( [ eq : hou ] ) : at the moment @xmath105 when the force ( [ eq : tto ] ) is turned on , @xmath106 coincides with @xmath155 for one particular @xmath0 ; this evidently is the desired @xmath154 .
the equality identifying @xmath154 thus is @xmath156 which , written out in full detail , reads @xmath157 using this to eliminate @xmath154 from eq .
( [ eq : fkt ] ) , we arrive at @xmath158 with precisely the same phase @xmath159 as already defined in eq .
( [ eq : phi ] ) . inserting this argument ( [ eq : arg ] ) into the cycle - averaged group velocity ( [ eq : cag ] ) , and comparing with the previous expression ( [ eq : cac ] ) ,
one confirms that the result of the floquet analysis fully coincides with that of the more customary crystal - momentum calculation .
the necessity to painstakingly distinguish between crystal momentum @xmath102 and floquet @xmath100 at all stages may appear a bit mind - boggling ; if this is not done with sufficient care , one might overlook a contribution to @xmath159 . @xcite
but if respected properly , the mathematical structure of the floquet picture unerringly leads to the correct answer .
if one strips the above reasoning to the bare essentials , that is , if one starts from the quasienergy dispersion relation ( [ eq : qed ] ) , takes its derivative to obtain the formal expression ( [ eq : cag ] ) for the cycle - averaged group velocity , and then inserts the solution to the equation of motion ( [ eq : par ] ) dictated by the generalized acceleration theorem in order to compute the group velocity of the wave packet actually considered , one sees that this procedure exactly parallels the explanation of the usual bloch oscillations , as reviewed in the introduction .
thus , super bloch oscillations may be seen as ordinary bloch oscillations arising in response to a weak probe force @xmath118 , but occurring in a spatiotemporal lattice , as created by dressing the original lattice through application of the strong force ( [ eq : int ] ) .
one might finally wish to get away from the particular , instantaneous onset of the forcing assumed in eq .
( [ eq : tto ] ) : the dc and the ac component might not be switched on simultaneously , or not abruptly , possibly involving two different turn - on functions for the two components . in any case , at some moment @xmath105 the final amplitudes will have been reached , so that the previous analysis goes through unaltered for @xmath108 , if one only interprets @xmath106 correctly : this would no longer indicate the crystal - momentum wave number around which the initial wave packet had been prepared , but rather that to which the latter had been shifted during the turn - on phase . expressed differently , the phase @xmath159 in eqs .
( [ eq : cac ] ) and ( [ eq : sbo ] ) depends significantly on the precise turn - on protocol : not surprisingly , the way the external force has been turned on in the past crucially affects the coherent wave - packet motion after the turn - on is over .
aside from its aesthetic value , the floquet picture offers at least one further benefit : bloch oscillations in dressed lattices may also occur under conditions such that the quasienergy bands are mixtures of several unperturbed energy bands , disabling a crystal - momentum treatment . a floquet analysis , on the other hand , would merely require one to replace the single - band quasienergies ( [ eq : qed ] ) by the actual ones and then again invoke the generalized acceleration theorem ( [ eq : gat ] ) , similar to the examples worked out in the next section .
a field of major current interest in which the floquet picture may find possibly groundbreaking applications concerns ultracold atoms , or weakly interacting bose - einstein condensates , in time - periodically driven optical lattices .
@xcite as opposed to ordinary crystalline matter exposed to high - power laser fields , such systems offer the advantage that one can apply even nonperturbatively strong driving forces without inducing unwanted inhomogeneities , as caused by polarization effects or domain formation .
@xcite the issue at stake here is not merely redoing well - known condensed - matter physics in another setting , and thus selling old wine in new skins , but rather finding genuinely new ways of coherently controlling mesoscopic matter waves , such that target states are created which have not been accessible before , and are manipulated according to some prescribed protocol . here
we point out that the generalized acceleration theorem ( [ eq : gat ] ) may be a valuable tool in this quest .
a standard one - dimensional ( 1d ) optical lattice is described by a cosine potential @xmath160 where @xmath161 is the wave number of the two counterpropagating laser beams generating the lattice .
@xcite its depth @xmath162 is measured in multiples of the single - photon recoil energy @xmath163 for orientation , if one traps @xmath164rb atoms in a lattice with @xmath161 corresponding to the wavelength @xmath165 nm , as in a recent experiment by zenesini _
et al . _ , @xcite one finds @xmath166 ev ; typical optical lattices are a few recoil energies deep .
figure [ fig : f_1 ] shows quasienergy spectra for such a 1d cosine lattice ( [ eq : ola ] ) with depth @xmath167 under pure ac forcing , that is , for @xmath168 not containing a dc component , with driving frequency @xmath169 . under the laboratory conditions specified above ( @xmath164rb at @xmath165 nm ) , this corresponds to @xmath170 khz .
figure [ fig : f_1](a ) results when the scaled driving amplitude ( [ eq : sda ] ) is set to zero ; this subfigure therefore is obtained by projecting the lowest three unperturbed energy bands to the fundamental quasienergy brillouin zone , which extends from @xmath171 to @xmath172 on the ordinate .
figure [ fig : f_1](b ) displays the quasienergy band structure for the moderate driving strength @xmath173 ; here avoided crossings show up which generally indicate multiphotonlike resonances . @xcite figure [ fig :
f_1](c ) then reveals pronounced ac stark shifts ( that is , shifts of the quasienergies against the zone - projected original energies ) for @xmath174 , corresponding to truly strong forcing .
we now turn from the quasienergy spectrum to an exemplary initial - value problem : at @xmath147 we prepare an initial wave packet ( [ eq : sbp ] ) in the lowest bloch band @xmath175 with a gaussian momentum distribution , @xmath176 centered around @xmath177 with width @xmath178 , and subject it to a pulse , @xmath179 starting at @xmath147 and ending at @xmath180 , endowed with a smooth , squared - sine envelope function : @xmath181 we again set @xmath169 , as in fig . [
fig : f_1 ] ; adjust the pulse length to 50 cycles ; @xmath182 , and fix the maximum driving amplitude @xmath183 such that @xmath184 , corresponding to the conditions reached in fig .
[ fig : f_1](c ) .
we then monitor the resulting wave - packet dynamics both in the basis of the unperturbed energy bands and in the bases provided by the instantaneous spatiotemporal bloch waves , that is , in the family of floquet bases which are obtained when the driving amplitude is kept fixed at any value @xmath185 reached during the pulse .
figure [ fig : f_2 ] displays the results : the jagged lines in the main frame show the occupation probabilities of the lowest three unperturbed bloch bands @xmath186 , and 3 ; in the middle of the pulse the band @xmath187 contains even more population than the band @xmath188 . on the other hand
, the horizontal line at the top depicts the occupation of the instantaneous floquet band emerging from the lowest bloch band : this floquet band contains practically _ all _ the population during the entire pulse , which means that the wave function adjusts itself adiabatically to the changing morphology of its quasienergy band , @xcite as previously sketched in fig .
[ fig : f_1 ] , when the driving amplitude @xmath189 is first increased and then decreased back to zero . to quantify the precise degree of adiabatic following , the inset in fig .
[ fig : f_2 ] shows the variation of the floquet band population on a much finer scale .
observe that the final adiabaticity defect is on the order of merely @xmath190 , even though the driving amplitude reaches its fairly high maximum strength within no more than 25 cycles .
with respect to the concepts developed in sec .
[ sec : s_2 ] , fig .
[ fig : f_2 ] strikingly demonstrates the advantages of the floquet picture over the traditional crystal - momentum representation for the situation considered .
if there were an additional probe force , its effect would have to be tediously disentangled from the fast oscillations of the bloch band populations .
when the same dynamics are seen from the floquet viewpoint , essentially `` nothing '' happens , because practically all inter - bloch - band transitions are already accounted for by continuously adapting the floquet basis , so that the action of a probe force would stand out most clearly .
although , of course , the crystal - momentum representation is mathematically equivalent to the floquet picture , there is no question which one is preferable here .
note also that fig .
[ fig : f_2 ] answers one further pertinent question : how do we prepare a wave packet which occupies merely a single quasienergy band , although it is undergoing rapid transitions between several bloch bands at the same time ?
the recipe for achieving this is simple : start with a wave packet occupying a single bloch band and switch on the driving force smoothly thereby enabling adiabatic following .
at this point an important issue needs to be stressed : the concept of adiabatic following , or parallel transport in a differential - geometric language , usually is applied to energy eigenstates ; @xcite in the context of optical lattices this has been exploited , e.g. , by fratalocchi and assanto for studying nonlinear adiabatic evolution and emission of coherent bloch waves .
@xcite in contrast , here we consider adiabatic following of explicitly time - dependent _ quasi_energy eigenstates , that is , of solutions to the quasienergy eigenvalue equation ( [ eq : qee ] ) ; this is what allows us to separate the fast , oscillating time dependence of the driving force from the slow , parametric time dependence of its envelope .
having learned these lessons , we now set the generalized acceleration theorem ( [ eq : gat ] ) to work .
suppose that we are prompted to empty the ground - state energy band .
starting again from an initial wave packet ( [ eq : ini ] ) , we then may proceed as follows : first we smoothly turn on an ac force which dresses the lattice , creating avoided quasienergy crossings initially not `` seen '' by the adiabatically following packet . for instance , we may wish to utilize the avoided crossings showing up in fig .
[ fig : f_1](b ) . to this end
, we again take an ac force with frequency @xmath169 and fix its scaled driving amplitude at the plateau value @xmath173 .
this dressing force is switched on during 25 cycles with half a squared - sine envelope , maintained at maximum amplitude for 50 further cycles , and switched off again for another 25 cycles , as sketched in fig .
[ fig : f_3](a ) .
if this were all we did , the wave packet would simply undergo adiabatic evolution and finally restore its initial condition , as previously observed in fig .
[ fig : f_2 ] . instead ,
once the maximum dressing amplitude has been reached , we now apply an additional weak probe force @xmath57 in order to exploit eq .
( [ eq : gat ] ) for moving the packet away from the brillouin zone center , driving it over the avoided crossings that have opened up in fig .
[ fig : f_1](b ) .
this probe force is implemented in the form of two smooth , squared - sine shaped dc pulses , one acting during the plateau of the dressing pulse , the other acting with reversed sign after the dressing pulse is over , as drawn in fig .
[ fig : f_3](a ) .
the maximum strength of the probe force here is only 2.5% of that of the dressing force ; for better visibility , the probe force is magnified in fig .
[ fig : f_3](a ) by a factor of 10 . it is now almost obvious how to describe the response of the wave packet within the floquet picture : the initial state ( [ eq : ini ] ) first is adiabatically shifted into a single quasienergy - band packet during the turn - on of the dressing force .
in contrast to a crystal - momentum representation , all dressing - induced fast oscillations are taken out of the dynamics of @xmath191 in the floquet representation , as shown in fig .
[ fig : f_3](b ) .
when the first probe pulse acts at constant dressing amplitude , it forces the wave packet over the avoided crossing seen in fig .
[ fig : f_1](b ) , so that the packet undergoes zener - type transitions to `` higher '' quasienergy bands , @xcite splitting into individual subpackets associated with the different quasienergy bands involved .
when the dressing force is switched off , each of these subpackets moves adiabatically on its own quasienergy surface , finally reaching the continuously connected bloch bands .
the second , reversed probe pulse , applied after the dressing pulse is over , then acts in accordance with bloch s original acceleration theorem ( [ eq : oat ] ) , shifting the various subpackets back to the brillouin zone center . in the scenario displayed in fig .
[ fig : f_3 ] , the lowest band is almost entirely depopulated by the probe - induced zener transitions , so that only a marginal fraction of the initial packet returns , as depicted in fig .
[ fig : f_3](c ) .
thus , the main part of the initial packet has been placed in higher bloch bands , as intended .
we have also checked by explicit calculation that without the comparatively weak probe pulses the returning wave packet would be almost identical to the initial one .
the above example of our `` dressing and probing '' strategy immediately lends itself to a host of further modifications and extensions . to give but one further instance , if the probe pulse is still weaker , such that the wave packet does not pass over the avoided - crossing regime , but rather stops there , the zener transitions are incomplete , so that a signifcant part of the initial state is recovered when the process is over .
this is elaborated in fig .
[ fig : f_4 ] with the same dressing force as above , but now the maximum strength of the probe force amounts to only 1.7% of that of the dressing force .
the final subpacket still occupying the lowest bloch band then is no longer centered around @xmath192 , implying that this subpacket will move over the lattice . in a sense ,
the left wing of the initial wave packet has been cut out , so that fig .
[ fig : f_4 ] may be regarded as a particular paradigm of `` wave - packet surgery . ''
summarizing our line of reasoning , we have introduced in sec . [
sec : s_2 ] a representation of wave packets of quantum particles in spatially periodic lattices subjected to homogeneous , time - periodic forcing which is based on an expansion with respect to spatiotemporal bloch waves and reduces to the standard crystal - momentum representation when the forcing is turned off .
it embodies forcing - induced oscillations into the basis , so that only the actually relevant dynamics remain to be dealt with . within this floquet representation one
encounters many features already familiar from solid - state physics in time - independent lattice potentials , but here their scope is different . as a prominent example , the generalized acceleration theorem derived in sec .
[ sec : s_3 ] takes the same form as its historic antecessor formulated by bloch , @xcite but applies to single quasienergy band dynamics , which can be drastically different from single energy band behavior .
there are further features which can be carried over from the crystal - momentum representation to the floquet picture and acquire a modified meaning there , such as the expression for the group velocity of a wave packet or zener transitions among different bands .
the super bloch oscillations considered in sec . [ sec : s_4 ] provide a mainly pedagogical example which can be worked out in full detail analytically . here
the floquet picture can not exert its full strength , because one assumes _ a priori _ that the driving force does not induce transitions from the initially occupied energy band to other ones , so that the historic acceleration theorem remains capable of describing the entire dynamics .
the floquet approach leads to exactly the same result , but implies a different viewpoint , separating the dc component of the force into one part which is resonant with the ac component , and together with the latter dresses the lattice , creating a quasienergy band ; the remaining residual part of the dc force then probes this new quasienergy band , rather than the original unperturbed energy band .
this theme of `` dressing and probing '' also prompts far - reaching strategies for achieving coherently controlled interband population transfer and even more .
two basic examples for this have been discussed in sec .
[ sec : s_5 ] , but the possibilities obviously extend much farther . utilizing the generalized acceleration theorem , an initial wave packet may by split coherently into two components at an avoided quasienergy band crossing in a dressed lattice , and the lattice may then be redressed ( that is , exposed to an ac force with different parameters ) such that another quasienergy band structure is generated , possibly involving avoided crossings which affect only one of the daughter wave packets created in the first step , but not the other .
moreover , daughter wave packets can be made to move , possibly into different directions , and to interfere with other wavelets having been manipulated separately before in distant parts of the lattice .
this vision apparently will be hard to realize with traditional solids , but it has come into immediate reach in current laboratory experiments with weakly interacting bose - einstein condensates in driven optical lattices . seen against this background , the generalized acceleration theorem almost provides a blueprint for a wave - packet processor . | a representation is put forward for wave functions of quantum particles in periodic lattice potentials subjected to homogeneous time - periodic forcing , based on an expansion with respect to bloch - like states which embody both the spatial and the temporal periodicity .
it is shown that there exists a generalization of bloch s famous acceleration theorem which grows out of this representation , and captures the effect of a weak probe force applied in addition to a strong dressing force . taken together , these elements point at a `` dressing and probing '' strategy for coherent wave - packet manipulation , which could be implemented in present experiments with optical lattices . |
phase transition phenomena in quantum field theories are typically of nonperturbative nature and thus naive perturbation theory based on an expansion in the coupling constant can not be employed .
this is clearly the case of phase changes at high temperatures , where perturbation theory becomes unreliable because powers of the coupling constant become surmounted by powers of the temperature @xcite .
problems with perturbation theory also happen in phenomena occurring close to critical points , because large fluctuations can emerge in the system due to infrared divergences , thus requiring nonperturbative methods as well in their studies .
this is the case of studies involving second order phase transitions and also in weak first order phase transitions @xcite .
typical examples where these problems can manifest are in studies of symmetry changing phenomena in a hot and dense medium , a subject of interest in quantum chromodynamics ( qcd ) in the context of heavy - ion collision experiments , and also in studies of the early universe .
consequently , there is a great deal of interest in investigating thermal field theories describing matter under extreme conditions @xcite .
familiar nonperturbative methods that have been used to study symmetry changing phenomena at finite temperatures are resummationlike techniques , such as the daisy and superdaisy schemes @xcite , composite operator methods @xcite , and field propagator dressing methods @xcite .
other methods used include expansions in parameters not related to a coupling constant , like the @xmath1 expansion and the @xmath2-expansion @xcite .
in addition , there are numerical methods , the most notably ones are those based on lattice monte carlo simulations @xcite .
each method has its own advantages and disadvantages .
for instance , in numerical methods there may be issues related to numerical precision , lattice spacing , and lattice sizes . in addition
, there is the notorious problem of simulating fermions on the lattice at finite chemical potentials @xcite . in any nonperturbative method based on an expansion in some parameter
one has to face the problem of higher order terms becoming increasingly cumbersome , so stalling further analysis .
this is usually the case when carrying out calculations beyond leading order in the @xmath1 expansion .
careless use of a nonperturbative method can also lead to problems like the lack of self - consistency or overcounting of effects .
known examples of such problems are the earlier resummation works dealing with daisy and superdaisy schemes , that at some point were giving wrong results , e.g. predicting a first order transition @xcite for the @xmath3 theory , an unexpected result since the model belongs to the universality class of the ising model , which is second order .
these methods also predicted a strong first order phase transition in the electroweak standard model , a result proved to be misleading @xcite .
let us recall that the breakdown of perturbation theory at high temperatures and its poor convergence properties have been dealt with many different methods .
examples are the use of self - consistent approximations @xcite , hard - thermal - loop ( htl ) resummation @xcite , perturbative expansions in the coupling constant with resummation implemented with the use of a variational mass parameter , also known as screened perturbation theory ( spt ) @xcite , and the use of two - particle irreducible ( 2pi ) effective actions @xcite .
the 2pi method , in particular , leads to a much better convergence of thermodynamic quantities ( like the pressure ) as compared to some of the other methods @xcite .
related to the 2pi method is the @xmath4-derivable technique , which has been used to study the thermodynamics of scalar and gauge theories @xcite .
one difficulty with the 2pi effective actions is that the renormalization procedure is nontrivial @xcite .
in addition , there seems that the @xmath4-derivable technique breaks down for a coupling beyond some value @xcite . in general , it is desirable that any analytical nonperturbative method obey two basic requirements .
first , it should be self - consistent , and second , it should produce useful results already at lowest orders without the need for going to higher orders .
that is , it should produce results that quickly converge at some order where calculations are still feasible analytically or semianalytically . though some of the cited methods may satisfy one , or to some extent both of these requirements , in the present paper we are particularly interested in the one known as the linear @xmath0 expansion ( lde ) @xcite , a nonperturbative method that has been used successfully in different contexts related to thermal field theories @xcite and in many other theories for a long , but
far from complete list of references see refs . @xcite . in the lde ,
a linear interpolation on the original model lagrangian density is performed in terms of a fictitious expansion parameter @xmath0 , which is used only for bookkeeping purposes and set at the end equal to one .
the standard application of the lde to a theory described by a lagrangian density @xmath5 starts with an interpolation defined by @xmath6\ ; , \label{opt}\end{aligned}\ ] ] where @xmath7 is the lagrangian density of a solvable theory , which is modified by the introduction of an arbitrary mass parameter ( or parameters ) @xmath8 .
the lagrangian density @xmath9 interpolates between the solvable @xmath10 ( when @xmath11 ) and the original @xmath5 ( when @xmath12 ) . the procedure defined by eq .
( [ opt ] ) leads to modified feynman vertices , that become multiplied by @xmath0 , and modified propagators , that now depend on @xmath8 .
all quantities evaluated at any finite order in the lde will then depend explicitly on @xmath8 , unless one could perform a calculation to all orders .
up to this stage the results remain strictly perturbative and very similar to the ones obtained via an ordinary perturbative calculation .
it is through the freedom in fixing @xmath8 that nonperturbative results can be generated in this method .
since @xmath8 does not belong to the original theory , one may fix it requiring that a physical quantity @xmath13 , calculated perturbatively to order-@xmath14 , be evaluated at the value where it is less sensitive to this parameter .
this criterion , known as the principle of minimal sensitivity ( pms ) , translates into the variational relation @xcite @xmath15 the optimum value @xmath16 which satisfies eq .
( [ pms ] ) is a function of the original parameters of the theory . in particular
, @xmath16 is a nontrivial function of the couplings and because of this nonperturbative results are generated .
another optimization procedure used is known as the fastest apparent convergence ( fac ) criterion @xcite .
it requires from the @xmath17-th coefficient of the perturbative expansion @xmath18 that @xmath19\bigr|_{\delta=1 } = 0\ ; , \label{fac}\ ] ] which is just equivalent to taking the @xmath17-th coefficient ( at @xmath20 ) in eq . ( [ fac0 ] ) equal to zero .
one should note that it is not at all guaranteed that the condition in eq .
( [ pms ] ) has a nontrivial solution . in cases where this may happen , the second criterion , eq .
( [ fac ] ) , may be more appropriate .
one example where the condition given by eq .
( [ pms ] ) fails to produce a nontrivial solution was in the problem studied by the authors in ref .
@xcite , who applied the lde to compute the effective potential in superspace . there
, the authors found that while the pms condition was unable to give a nonperturbative solution to the effective potential , the fac criterion worked perfectly well .
of course , in many situations both optimization criteria may work and in this case one may ask whether they lead to equivalent results .
previous studies indicated that this is indeed so , but a full comparison of results obtained with both optimization criteria is still lacking . another issue associated with
the lde is its convergence .
rigorous lde convergence proofs have been obtained for the problem of the quantum anharmonic oscillator , at zero temperature , considered in ref .
@xcite , while its partition function at finite temperatures was considered in @xcite . for quantum field theories , ref .
@xcite has proved convergence for a particular perturbative series in an asymptotically - free , renormalizable model at zero temperature . for a critical @xmath3 @xmath21 theory in three dimensions the issue of convergence
was studied in @xcite employing both pms and fac optimization criteria . finally , regarding the possible solutions that can emerge from the optimization criteria ( pms or fac ) , we must use a definite approach in selecting the optimum root @xmath16 from either eq .
( [ pms ] ) or ( [ fac ] ) .
the problem of dealing with the many possible solutions for @xmath22 was treated in details in the first two papers cited in ref .
@xcite , where the convergence of the lde was also studied in details .
typically , the higher the order in @xmath0 , the more solutions can appear .
as shown in those references , all solutions at each given order in @xmath0 can be classified into families .
the optimum value for @xmath8 is chosen as follows : the trivial solutions for @xmath22 , e.g. @xmath23 and those that are not dependent on the coupling constant ( and thus can not lead to nonperturbative results ) are not considered .
in addition to these , at first order there is only one nontrivial solution ( first family ) , consistent with all our approximations , ( like the high - temperature approximation , used later in our calculations ) .
this family is then followed in the next orders and used in all our calculations .
as proved in earlier references with the lde method , this is a consistent and unambiguous way for choosing the optimum value for @xmath8 .
it is important to stress that in the method of the lde the selection and evaluation of feynman diagrams proceed in the same fashion as in ordinary perturbation theory , including the renormalization procedure @xcite .
the results obtained are free from infrared divergences , even at the critical point and in its neighborhood , thus making it a particularly suitable method to study phase transition phenomena in quantum field theories .
it is important to recall here that there are similarities between the lde and the spt methods .
in particular , the implementation of the latter can be put in a form similar to the lde by means of a modified loop expansion @xcite , named optimized perturbation theory ( opt ) in this reference .
but there are also some major differences between these methods .
for instance , in the lde no assumption is made _ a priori _ for the parameter @xmath8 , while in the spt / opt it is assumed that such a mass term is already of some order in the coupling constant .
the implication of this is that the order counting of loop expansion has to be readjusted accordingly . in the present paper
we study the application of the lde to the @xmath24 theory
. we will study the applicability of the pms and fac optimization criteria for the symmetric and broken phases of the theory , and compare results obtained with both methods .
in addition , in the present work we choose to optimize the free energy , instead of the self - energy like in many other works employing the lde , particularly refs .
there are several reasons for doing so @xcite , but an important one is that in some situations it might happen that the optimization of the self - energy does not lead to nontrivial solutions , while optimization of the free energy with pms or fac are seen to lead to nontrivial solutions already at first order in @xmath0 .
the critical temperature @xmath25 , the pressure @xmath26 , and the background dependent free energy @xmath27 are obtained here in an explicit calculation up to order @xmath28 . calculations at this order require a calculation of vacuum terms up to three loops .
since the thermodynamics of this model has been extensively studied before in the literature with a number of methods , our calculation here will be useful to benchmark the application of the lde and its two main optimization procedures against those previous applications .
in addition , we compare our results with those obtained with standard perturbation theory .
besides correctly reproducing the expected second order phase transition pattern for the model , our results at order @xmath28 are shown to be sufficient to obtain the thermodynamics of the model , in the sense that the results at @xmath29 are not much different from the ones at @xmath30 . the results point towards a quickly convergent lde , as already indicated in previous studies with different models under different conditions @xcite .
this work is organized as follows . in the next section
we introduce the interpolation procedure for the model . in sec .
iii we compute the free energy in the symmetric and broken phases to @xmath29 . in sec .
iv we present the results obtained from the optimization procedures .
the pressure is evaluated and contrasted order by order with the one obtained within perturbation theory .
the critical temperature , the temperature dependent vacuum expectation value of the scalar field and the free energy are determined to @xmath29 .
our conclusions are presented in sec .
the interpolation defined by eq .
( [ opt ] ) when applied for the standard @xmath31 model gives @xmath32 where @xmath33 and @xmath34 is the part of the lagrangian density carrying the renormalization terms needed to render the model finite .
details about this renormalization procedure in the lde and the explicit form for @xmath34 are given e.g. in ref .
@xcite for the case of background field dependent contributions ( broken symmetry phase ) , while the field independent contributions ( symmetric phase ) were given in ref .
@xcite within the context of the spt , so we will not repeat those same renormalization details here .
one should also note that the only new " terms introduced by the @xmath0-expansion interpolation are quadratic terms and so the renormalizability of the original theory is not changed .
this means that the renormalization of the theory can be carried out in an analogous way as in ordinary perturbation theory @xcite . specifically , the interpolation procedure given by eq .
( [ interp ] ) introduces a new ( quadratic ) interaction term , with feynman rule @xmath35 .
in addition to this modification , the original bare propagator , @xmath36 now becomes @xmath37 while the original quartic vertex is changed from @xmath38 to @xmath39 . in the next section
we will show the results for the finite temperature free energy density using the interpolated model with the lde at @xmath40 .
we will consider the cases of @xmath41 and @xmath42 in eq .
( [ l0 ] ) , corresponding to the symmetric and broken phases , respectively .
we perform the standard derivation of the free energy @xcite up to @xmath40 . with the constant field introduced through the usual shift of the scalar field , @xmath43 ,
the lagrangian density is rewritten as @xmath44=\mathcal{l}_2[\phi(x),\varphi ] + \mathcal{l}_i[\phi(x),\varphi]\;,\end{aligned}\ ] ] where @xmath45 is the part of the lagrangian quadratic in the fields , @xmath46 = \frac{1}{2 } \left ( \partial _ { \mu } \phi\right ) ^{2}-\frac{1}{2}\omega ^{2}\phi^{2}\ ; , \label{l2}\end{aligned}\ ] ] while @xmath47 is @xmath48 = -\frac { \delta \lambda } { 6}\varphi \phi^{3}-\frac{\delta \lambda } { 4!}\phi^{4}\;,\end{aligned}\ ] ] where in eq .
( [ l2 ] ) @xmath49 is given by @xmath50 note that in all loop contributions the propagators will carry a mass term as given by eq .
( [ omega ] ) .
these terms are then expanded in @xmath0 to the desired order , thus generating the insertions of @xmath51 that appear as a consequence of the quadratic vertex introduced in eq .
( [ interp ] ) .
the free energy is @xmath52&=&f_0(\varphi ) + f_{\rm 1-loop } ( \varphi)\nonumber\\ & + & \frac{1}{{\cal v } } i \ln \bigg{\langle}\exp\left\ { i\int d^4x\ , \mathcal{l}_i[\phi(x),\varphi]\right\ } \bigg{\rangle}\ ; , \label{free energy}\end{aligned}\ ] ] where @xmath53 is the tree - level classical potential and @xmath54 is the one - loop contribution to the free energy ( @xmath55 is the space volume ) given by @xmath56 } \;.\end{aligned}\ ] ] higher loops are given by the last term in eq .
( [ free energy ] ) , with the average @xmath57 meaning @xmath58 } } { \int \mathcal{d}\phi \ ; e^ { i\int d^{4}x\,\mathcal{l}_{2}\left [ \phi\left ( x\right ) , \varphi \right ] } } \;.\ ] ] as said above , the scalar field propagators in the diagrams are obtained from @xmath59 $ ] , and the vertices are determined from @xmath60 $ ] , with both as given at the end of sec .
our calculations are performed , as usual , in the imaginary - time formalism @xcite .
thus , the scalar boson field has euclidean four - momentum @xmath61 , with @xmath62 , where @xmath63 are the discrete matsubara bosonic frequencies @xmath64 , with @xmath65 , and @xmath66 .
loop diagrams involve sums over the matsubara frequencies and integrals over the space momentum @xmath67 .
all space momentum integrals are performed in arbitrary dimension @xmath68 and renormalization is performed in the modified minimal subtraction scheme ( @xmath69 ) . the measure used in the sum - integrals
is then defined as @xmath70 where @xmath71 is an arbitrary momentum scale in dimensional regularization .
the factor @xmath72 is introduced so that , after minimal subtraction of the poles in @xmath2 due to ultraviolet divergences , @xmath71 coincides with the renormalization scale in the @xmath69 scheme . from eq .
( [ free energy ] ) , the free energy is expressed up to @xmath73 by expanding all appropriate terms in @xmath0 .
considering the vacuum contributions to the free energy , this means that terms up to three - loops must be included .
all bare ( unrenormalized ) contributions are shown in fig . [ loopdiagrams ] . , given by ( a ) vacuum diagrams and ( b ) background field ( external legs ) .
the black dots indicate a @xmath74 insertion . ]
the renormalization procedure for the symmetric phase was performed in detail in refs .
the counterterms for the vacuum diagrams are given in ref .
@xcite , while those for the field dependent diagrams are given in ref .
we also note that the divergences in the broken phase can be removed by the same counterterms determined for the symmetric phase @xcite , so the renormalization for the broken phase does not require extra effort .
the renormalization proceeds just as in standard perturbation theory and as shown in detail in ref .
@xcite , only temperature independent counterterms are required and the temperature dependent divergent terms cancel out exactly .
all diagrams of counterterms contributing to @xmath75 $ ] up to @xmath76 are shown in fig .
[ ctdiagrams ] . : ( a ) vacuum contribution , ( b ) background field contribution . as in fig .
1 , the black dot indicates a @xmath74 insertion .
the circle - cross denotes either insertion of a mass counterterm or of a vertex counterterm . ] the circle - cross in fig .
[ ctdiagrams ] denotes either a mass counterterm vertex @xmath77 , or a vertex counterterm @xmath78 , given respectively by @xcite @xmath79 - \delta^2 \frac{\lambda^2}{(32 \pi^2)^2}\left(\frac{-2}{\epsilon^2}+ \frac{1}{\epsilon}\right)\left(m^2 + \eta^2\right)\ ; , \label{deltam } \\ & & \delta \lambda = -\delta^2 \frac{3\lambda^2}{32 \pi^2 \epsilon}\;. \label{deltalambda}\end{aligned}\ ] ] the final expression for the renormalized free energy @xmath75 $ ] , including all terms shown in figs .
[ loopdiagrams ] and [ ctdiagrams ] becomes @xmath80 = f_{\rm{vacuum } } + f_{\varphi } \label{freeenergy}\end{aligned}\ ] ] where @xmath81 denotes the vacuum contributions , @xmath82 \mathcal{m}^{4 } -\frac{1 } { 2\left ( 4\pi ^{2}\right ) } j_{0}\left ( \beta \mathcal{m } \right ) t^{4 } \nonumber\\ & + & \delta \frac{\lambda } { 8\left ( 4\pi \right ) ^{4}}\left [ \left ( \ln \left ( \frac{\mu ^{2}}{\mathcal{m}^{2}}\right ) + 1\right ) \mathcal{m}^{2}-j_{1 } \left ( \beta \mathcal{m } \right ) t^{2}\right ] ^{2 } \nonumber \\ & + & \delta \frac{\eta ^{2}}{2\left ( 4\pi \right ) ^{2}}\left [ \left ( \ln \left ( \frac{\mu ^{2}}{\mathcal{m } ^{2}}\right ) + 1\right ) \mathcal{m } ^{2}-j_{1}\left ( \beta \mathcal{m } \right ) t^{2}\right ] \nonumber \\ & -&\delta ^{2}\frac{\eta ^{4}}{4\left ( 4\pi \right ) ^{2}}\left [ \ln \left ( \frac{\mu ^{2}}{\mathcal{m } ^{2}}\right ) + j_{2}\left ( \beta \mathcal{m } \right ) \right ] \nonumber \\ & -&\delta ^{2}\frac{\lambda } { 4\left ( 4\pi \right ) ^{4}}\eta ^{2 } \left [ \ln\left ( \frac{\mu ^{2}}{\mathcal{m } ^{2}}\right ) + j_{2 } \left ( \beta \mathcal{m } \right ) \right ] \left [ \left ( \ln \left ( \frac{\mu ^{2}}{\mathcal{m } ^{2}}\right ) + 1\right ) \mathcal{m}^{2}-j_{1}\left ( \beta \mathcal{m } \right ) t^{2}\right ] \nonumber \\ & -&\delta ^{2}\frac{\lambda ^{2}}{48\left ( 4\pi \right ) ^{6}}\left\ { \left [ 5\ln ^{3}\left ( \frac{\mathcal{m}^{2}}{\mu ^{2}}\right ) + 17\ln ^{2}\left ( \frac{\mathcal{m } ^{2}}{\mu
^{2}}\right ) + \frac{41}{2}\ln \left ( \frac{\mathcal{m }
^{2}}{\mu^{2}}\right ) -23-\frac{23}{12\pi ^{2}}\right .
\right .
\nonumber \\ & -&\left .
\psi ^{\prime \prime } \left ( 1\right ) + c_{0 } + 3\left ( \ln \left ( \frac{\mathcal{m } ^{2}}{\mu
^{2}}\right ) + 1\right)^{2}j_{2}\left ( \beta\mathcal{m } \right ) \right ] \mathcal{m}^{4 } -\left [ 12\ln ^{2}\left ( \frac{\mathcal{m } ^{2}}{\mu ^{2}}\right ) + 28\ln \left ( \frac{\omega ^{2}}{\mu ^{2}}\right ) \right.\right . \nonumber \\ & -&\left . \left .
12-\pi ^{2}-4c_{1}+6\left ( \ln \left ( \frac{\mathcal{m}^{2}}{\mu^{2}}\right ) + 1\right ) j_{2 } \left ( \beta \mathcal{m } \right ) \right ] j_{1 } \left ( \beta \mathcal{m } \right ) \omega^{2}t^{2}\right
. \nonumber \\ & + & \left . \left
[ 3\left ( 3\ln \left ( \frac{\mathcal{m}^{2}}{\mu^{2}}\right ) + 4\right ) j_{1}^{2}\left ( \beta \mathcal{m } \right ) + 3j_{1}^{2}\left ( \beta \mathcal{m } \right ) j_{2}\left ( \beta \mathcal{m } \right ) + 6k_{2 } + 4k_{3}\right ] t^{4}\right\ } \ ; , \label{fphi0}\end{aligned}\ ] ] and @xmath83 denotes the background field dependent contributions , @xmath84 -\delta ^{2}\frac{\lambda \eta ^{2}}{32\pi ^{2}}\left [ \ln \left ( \frac { \mathcal{m}^{2}}{\mu ^{2}}\right ) -j_{2}\left ( \beta \mathcal{m}\right ) \right ] \right . \nonumber \\ & -&\delta ^{2}\frac{\lambda ^{2}\mathcal{m}^{2}}{2\left ( 32\pi ^{2}\right ) ^{2}}\left [ \left ( \ln \left ( \frac{\mathcal{m}^{2}}{\mu ^{2}}\right ) \right ) ^{2}+\frac{\pi ^{2}}{6}\right ] -\delta ^{2}\frac{3\lambda ^{2 } \mathcal{m}^{2}}{2\left ( 32\pi ^{2}\right ) ^{2}}\left [ \left ( \ln \left ( \frac{\mathcal{m}^{2}}{\mu ^{2}}\right ) -1\right ) ^{2}+1+\frac{\pi ^{2}}{6 } \right ] \nonumber \\ & + & \delta ^{2}\frac{\lambda ^{2}}{1024\pi ^{4}}\left [ \mathcal{m}^{2}\left ( 1+\frac{\pi ^{2}}{6}\right ) + 4\mathcal{m}^{2}\ln \left ( \frac{\mu } { \mathcal { m}}\right ) \left [ 1+j_{2}\left ( \beta \mathcal{m}\right ) \right ] + j_{2}\left ( \beta \mathcal{m}\right ) \mathcal{m}^{2}\right . \nonumber \\ & + & \left . 8\mathcal{m}^{2}\ln ^{2}\left ( \frac{\mu } { \mathcal{m}}\right ) -4\ln \left ( \frac{\mu } { \mathcal{m}}\right ) j_{1}\left ( \beta \mathcal{m } \right ) t^{2}-j_{2}\left ( \beta \mathcal{m}\right ) j_{1}\left ( \beta \mathcal{m}\right ) t^{2}\right ] \nonumber \\ & + & \delta ^{2}\frac{\lambda ^{2}t^{2}}{24\left ( 4\pi \right ) ^{2}}\left [ \ln \left ( \frac{\mathcal{m}^{2}}{t^{2}}\right ) + 5.3025\right ] + \delta ^{2}\frac { \lambda ^{2}\mathcal{m}^{2}}{256\pi ^{4}}\left [ \frac{\pi ^{2}}{24}-3\ln \left ( \frac{\mathcal{m}}{\mu } \right ) \right . \nonumber \\ & + & \left . \left .
2\ln ^{2}\left ( \frac{\mathcal{m}}{\mu } \right ) + 1.164032 \right ] \right\ } \frac{\varphi ^{2}}{2}-\delta ^{2}\frac{3\lambda ^{2 } } { 32\pi ^{2}}\left\ { \log \left ( \frac{\mathcal{m}^{2}}{\mu ^{2}}\right ) -j_{2}\left ( \beta \mathcal{m}\right ) \right\ } \frac{\varphi ^{4}}{4!}\;. \label{fphineq0}\end{aligned}\ ] ] with @xmath85 given by @xmath86 \varphi^2 + \delta \frac{\lambda}{24 } \varphi^4\;. \label{f0}\end{aligned}\ ] ] in eqs .
( [ fphi0 ] ) and ( [ fphineq0 ] ) , @xmath87 , and the constant terms appearing in eq .
( [ fphi0 ] ) are defined as follows : @xmath88 , where @xmath89 is the zeta function , @xmath90 and @xmath91 , while @xmath92 and @xmath93 are three - dimensional integrals that can be evaluated numerically @xcite . in the high - temperature limit , @xmath94 ,
they are given by @xcite @xmath95 - 372.65 \beta\mathcal{m}\left [ \ln \left(\beta\mathcal{m } \right)+ 1.4658\right]\;,\end{aligned}\ ] ] and @xmath96\;.\ ] ] in eqs .
( [ fphi0 ] ) and ( [ fphineq0 ] ) , we have also defined the temperature dependent integrals @xmath97 ( @xmath98 ) as follows , @xmath99 which can be expressed as a series expansion as follows @xcite @xmath100 @xmath101 + \frac { 4\pi ^{2}}{3 } \nonumber \\ & -&16\sum_{n=1}^\infty\left ( \frac{\left ( -1\right)^{n}\left ( 2n-1\right ) ! ! \zeta \left ( 2n+1\right ) a^{\left ( 2n+2\right ) } } { 4n!2^{n+1}\left ( n+1\right ) \left ( 2\pi \right)^{2n } } \right)\ ; , \label{j1}\end{aligned}\ ] ] and @xmath102 \;. \label{j2}\end{aligned}\ ] ] equations ( [ j0])-([j2 ] ) are all convergent in the high - temperature limit as can be easily checked by considering a few terms in the sums in these equations .
we should note that when optimizing the free energy , since @xmath103 , @xmath104 and @xmath105 are dependent on @xmath8 , it is important to check the stability of the results when truncating the sums in eqs .
( [ j0])-([j2 ] ) .
this is particularly critical for parameter values such that @xmath106 is not much smaller than 1 , a situation that requires a fairly large number of terms in the sums . in all results shown in the next section
we have used enough terms in eqs .
( [ j0])-([j2 ] ) so to obtain stable results for all parameter and temperature values used .
we now turn to the application of the optimization procedures in the lde and show the results obtained by implementing the pms , eq .
( [ pms ] ) , and fac , eq .
( [ fac ] ) .
as we explained in the introduction , the optimization criteria are applied directly to the free energy .
the results obtained with each optimization criterion are contrasted with each other and with those available from other methods .
this will then allow us to gauge the performance of each optimization procedure regarding both reliability and convergence .
we initially restrict our calculations to the symmetric phase ( with positive mass term in the classical potential ) and evaluate the pressure , @xmath107 . in fig .
[ fig : p_pert ] we show our results for the pressure using the usual perturbation theory in @xmath108 up to @xmath109 and where we have restricted to the case of high temperatures ( @xmath110 ) . in this figure
the behavior of the pressure is shown as a function of the renormalized coupling constant , @xmath111 , and @xmath112 is a reference energy scale chosen as @xmath113 , where @xmath114 is the renormalized mass .
this is similar as done in ref .
@xcite using the 2pi method .
note that in ref . @xcite the authors define the quartic coupling differently from us . in their case , @xmath115 , and their results
are plotted as a function of the renormalized coupling @xmath116 .
we choose here the same scale as in ref .
@xcite so to facilitate the comparison with their results for the pressure .
it is clear in fig .
[ fig : p_pert ] the typical alternating behavior of the perturbative calculation , which indicates its very poor convergence . up to @xmath109 .
the parameters used are @xmath117 for the temperature and @xmath118 for the renormalization scale . ]
next , we use the result for the free energy evaluated up to @xmath76 , given by eq .
( [ fphi0 ] ) .
note that in the symmetric phase the pressure depends only on vacuum terms , since the free energy is evaluated at the vacuum expectation value for the field , @xmath119 . by optimizing the free energy using the pms criterion , eq .
( [ pms ] ) , we determine the root @xmath22 , which is then substituted back into the expression for the free energy , with the criterion used for choosing the optimum root as discussed below eq .
( [ fac ] ) .
this naturally brings nonlinear @xmath108 contributions and generates nonperturbative results . the pressure obtained in this case
is shown in fig .
[ fig : p_pert+lde ] , where we show the results obtained up to orders @xmath0 and @xmath28 . in the same figure
we also show the perturbative results of fig .
[ fig : p_pert ] for comparison .
it becomes evident here the convergence of the results with the lde - pms , with both @xmath30 and @xmath29 results not differing too much , in contrast to the perturbative ( in @xmath108 ) results . and @xmath28 using the pms optimization criterion - the perturbative results of fig .
[ fig : p_pert ] are also shown .
the parameters used are the same as those in fig .
[ fig : p_pert ] . ] in fig .
[ fig : p_lde ] we show again the results for the pressure , but now using the fac optimization criterion , eq .
( [ fac ] ) .
we once again see the excellent convergence for the pressure when contrasted to the perturbative results . and @xmath28 using the fac optimization criterion . ] in fig .
[ fig : p_fac+pms_d2 ] we plot side by side our results for the pressure at order @xmath28 using the pms and fac optimization criteria .
it is seen as an excellent agreement between the two optimization criteria and it shows the equivalence of these two optimization procedures . using the pms and fac optimization criteria . ] a side by side comparison of the order @xmath28 result for the pressure ( from either the pms or fac ) with the 2pi two - loop result of ref .
@xcite ( second panel of theirs fig .
1 ) shows an excellent agreement between the two results .
since operationally the lde is much simpler to be implemented than the 2pi calculation and also when compared with other methods , like those based on the renormalization group and schwinger - dyson equations , this may be a great advantage of the lde .
many previous applications of the lde to a large variety of problems ( cited previously ) also confirm the strength of the method .
its strength comes basically from the fact that its implementation is similar to that of standard perturbation theory .
the important and fundamental difference with standard perturbation theory resides in the optimization procedure that fixes an initial , _ a priori _ , arbitrary parameter , @xmath8 .
it is then interesting to investigate what kind of role the optimum @xmath8 represents in the lde after optimization .
this is partially clarified in the plot shown in fig .
[ fig : eta2_lde_d1 ] , where we show the optimum @xmath8 as a function of the renormalized coupling constant .
it shows that by increasing the order in @xmath0 , @xmath16 becomes closer and closer to the thermal mass @xmath120 , here computed at one - loop order for simplicity . in general
, we can extrapolate this expectation and say that the expected optimum value for @xmath8 should be the ( quantum and ) thermal mass ( quadratic in the field ) corrections , as would be derived from a true gap equation .
this is in fact confirmed by the many applications of the lde to the gross - neveu model @xcite , in which case exact results are known ( in the large-@xmath121 approximation ) and can then be readily compared with the results obtained from the lde method applied to that model . with respect to the renormalized coupling constant and evaluated at order @xmath0 and @xmath28 using the fac optimization criterion .
the parameters used are the same as in the previous figures . ]
let us now turn to the symmetry broken case ( with negative square mass term in the classical potential ) .
for this case we found that only the fac optimization criterion leads to nontrivial solutions for @xmath8 .
the fac criterion is applied to the free energy and the resulting nonlinear equation is solved simultaneously with the equation giving the minimum condition for the field ( thermal ) expectation value , @xmath122 , given by the condition @xmath123 } { d\ , \varphi } \right\vert _ { \varphi = \nu(t ) } = 0\;.\end{aligned}\ ] ] as it is well known , the phase transition in the pure scalar theory is second order @xcite , as required by universality reasons .
our results for the free energy using the fac criterion indicate a second order phase transition .
this is shown in fig .
[ fig : veff ] , where the free energy for @xmath124 is plotted for different temperature values .
the critical temperature obtained here is @xmath125 , consistent with the perturbative prediction @xcite and other nonperturbative calculations @xcite .
another quantity that indicates that the transition is a continuous one is the temperature evolution of the minimum of free energy , @xmath122 .
this is shown in fig .
[ fig : nut ] for @xmath126 and @xmath127 . evaluated at order @xmath28 and
using the fac optimization criterion , for three different temperatures : @xmath128 , @xmath129 and @xmath130 , where @xmath131 ( in units of the renormalization scale @xmath71 ) . here
we have set @xmath132 . ] at order @xmath28 obtained with the fac optimization criterion .
here @xmath133 is the tree - level minimum of the bare free energy . ]
[ fig : nut ] finally , in fig .
[ fig : mt2 ] we show the temperature dependence of the thermal mass , @xmath120 , as derived from the free energy ,
@xmath134}{d^2 \varphi}\bigr|_{\varphi=0}\;. \label{mt2}\ ] ] we once again can notice a continuous and smooth transition .
we note that one can determine the critical temperature by looking at which value of @xmath135 @xmath136 changes sign and check whether this gives the same result for @xmath25 as obtained from @xmath137 ( as in fig .
[ fig : nut ] ) . evaluated with the fac criterion for two values of the coupling constant @xmath108 .
we use in this plot @xmath132 . ]
one of the motivations for using the lde to study the thermodynamics of the scalar field theory at high temperatures , as done in this work , was its ease of implementation and renormalization , which is no different from those of the standard perturbation theory .
one recalls that similar studies in the context of the 2pi and related methods typically face difficulties in the renormalization procedure , making their applicability a nontrivial task .
in addition , the lde method , differently from other methods , like spt ( or opt - optimized perturbation theory ) , makes no assumption on the introduced mass parameter @xmath8 , thus we do not need to adjust the order of the loop expansion accordingly . by using the lde , we have studied the thermodynamics of the @xmath3 scalar field theory in the symmetric and broken phases .
the lde is used with two popular optimization procedures , known as pms and fac .
there are two major differences with the work we have done here and previous ones , like e.g. in refs .
first , while in general the pms procedure is the favorite optimization criterion in the literature related to the lde and similar methods , we here have shown that the fac procedure leads to numerically indistinguishable results from the ones obtained with the pms .
in addition , while there may be cases where the pms procedure leads to trivial results only , the same may not be the case for the fac ( here we have shown this to be the case in the broken phase ) . in this sense
they can be used in a complementary way , when pms fails , one can try fac , or vice - versa . secondly , unlike in refs .
@xcite , where the quantity optimized is the self - energy , here we choose to optimize the free energy .
one advantage of this is that , while there is no solution for the lde at first order when optimizing the self - energy , we do find solutions when optimizing the free energy already at first order in @xmath0 .
furthermore , as shown in ref .
@xcite , the optimization of the free energy can be shown to immediately lead to the solution of the gap equation ( here verified numerically through the results for the optimum @xmath8 ) , while in optimizing the self - energy further constraints must be employed , as for example renormalization group equations . in the numerical studies performed in the present work
, we have shown that the optimum @xmath8 carries both temperature and coupling constant contributions .
thus , the lde with optimization of the free energy implements automatically a nonperturbative resummation of the thermal corrections , in conformity with analytical results produced when this method was used to study the gross - neveu model in ref .
@xcite , from which exact solutions are available and a close comparison with the lde results is possible . by studying the behavior of the pressure and contrasting the results obtained with perturbation theory and the 2pi method , we have shown that the lde leads to convergent results already at lowest order in the lde expansion parameter @xmath0 , with the first and second order results changing only slightly and producing results consistent with the 2pi nonperturbative method . in addition ,
as already mentioned above , we have shown that both optimization procedures , fac and pms , lead to equivalent results for the pressure .
another important result of our work is that the lde is shown to be adequate for studying phase transitions at high temperatures .
in particular , when applied to the phase transition in the @xmath3 , the lde predicts the correct order of the phase transition , which is second order , in agreement with general results of statistical mechanics . besides this ,
since the lde method automatically produces an infrared cutoff , the results are shown to be valid and applicable _ below _ , _ above _ , and _ at _ the critical temperature @xmath25 , showing that the lde circumvents the usual problem seen in perturbation theory , namely , the appearance of infrared divergences close to critical points .
we would like to thank u. reinosa for helpful discussions regarding their 2pi results and the renormalization issues in the method .
we would like to thank f. gardim for discussions on related matters .
this work was partially supported by cnpq , fapesp , and faperj ( brazilian agencies ) .
99 m. le bellac , _ thermal field theory _
( cambridge university press , cambridge , 1996 ) . m. gleiser and r. o. ramos , phys .
b300 * , 271 ( 1993 ) . j .-
blaizot , e. iancu , and a. k. rebhan , aip conf .
739 , 63 ( 2004 ) .
d. h. rischke , prog .
phys.*52 * , 197 ( 2004 ) .
u. kraemmer and a. k. rebhan , rept .
prog . phys .
* 67 * , 351 ( 2004 ) .
j. o. andersen and m. strickland , ann .
* 317 * , 281 ( 2005 ) .
j. r. espinosa , m. quirs and f. zwirner , phys .
b291 * , 115 ( 1992 ) .
j. arafune , k. ogata and j. sato , prog .
. phys . * 99 * , 119 ( 1998 ) .
g. amelino - camelia and s .- y .
pi , phys .
d * 47 * , 2356 ( 1993 ) .
n. banerjee and s. mallik , phys .
d * 43 * , 3368 ( 1991 ) .
r. r. parwani , phys .
d * 45 * , 4695 ( 1992 ) ; erratum , phys .
d * 48 * , 5965 ( 1993 ) .
j. zinn - justin , _ quantum field theory and critical phenomena _
( oxford university press , 1996 ) .
s. muroya , a. nakamura , c. nonaka and t. takaishi , prog .
* 110 * , 615 ( 2003 ) .
m. e. carrington , phys .
d * 45 * , 2933 ( 1992 ) .
p. arnold and o. espinosa , phys .
d * 47 * , 3546 ( 1993 ) .
b. gruter , r. alkofer , a. maas and j. wambach , eur .
j. c * 42 * , 109 ( 2005 ) .
e. braaten and r. d. pisarski , nucl . phys . *
b337 * , 569 ( 1990 ) ; j. frenkel and j. c. taylor , nucl . phys .
* b334 * , 199 ( 1990 ) ; j. c. taylor and s. m. h. wong , nucl . phys .
* b346 * , 115 ( 1990 ) . j. o. andersen , e. braaten , and m. strickland , phys . rev
. lett . * 83 * , 2139 ( 1999 ) ; phys .
d * 61 * , 014017 ( 1999 ) ; phys .
d * 61 * , 074016 ( 2000 ) ; j. o. andersen , e. braaten , e. petitgirard , and m. strickland , phys .
d * 66 * , 085016 ( 2002 ) ; j. o. andersen , e. petitgirard , and m. strickland , phys .
d * 70 * , 045001 ( 2004 ) .
f. karsch , a. patks , and p. petreczky , phys .
b * 401*,69 ( 1997 ) ; j. o. andersen and m. strickland , phys .
d * 64 * , 105012 ( 2001 ) ; phys .
d * 71 * , 025011 ( 2005 ) ; j. o. andersen and l. kyllingstad , hep - ph/0805.4478 .
andersen , e. braaten and m. strickland , phys .
d * 63 * , 105008 ( 2001 ) .
j. m. cornwall , r. jackiw and e. tomboulis , phys .
d * 10 * ( 1974 ) 2428 .
j. berges , sz .
borsanyi , u. reinosa and j. serreau , phys .
d * 71 * , 105004 ( 2005 ) . j .-
blaizot , e. iancu and a. rebhan , phys .
b * 470 * , 181 ( 1999 ) . j .-
blaizot , e. iancu and a. rebhan , phys .
lett . * 83 * , 2906 ( 1999 ) . j .-
blaizot , e. iancu and a. rebhan , phys .
d * 63 * , 065003 ( 2001 ) .
a. peshier , phys .
d * 63 * , 105004 ( 2001 ) .
h. van hees and j. knoll , phys .
d * 65 * , 025010 ( 2001 ) ; d * 65 * , 105005 ( 2002);j .- p .
blaizot , e. iancu and u. reinosa , phys .
b * 568 * , 160 ( 2003 ) ; nucl .
* a736 * , 149 ( 2004 ) ; j. berges , s. borsanyi , u. reinosa , and j. serreau , ann .
* 320 * , 344 ( 2005 ) ; g. fejos , a. patkos , and zs .
szep , nucl . phys . *
a803 * , 115 ( 2008 ) ; a. arrizabalaga and u. reinosa , nucl .
a * 785 * , 234 ( 2007 ) ; eur .
j. a * 31 * 754 ( 2007 ) .
e. braaten and e. petitgirard , phys .
d * 65 * , 085039 ( 2002 ) ; a. okopinska , phys .
d * 35 * , 1835 ( 1987 ) ; m. moshe and a. duncan , phys .
b * 215 * , 352 ( 1988 ) .
kneur , m. b. pinto and r. o. ramos , phys .
d * 74 * , 125020 ( 2006 ) ; braz .
* 37 * , 258 ( 2007 ) ; int.j.mod.phys . *
e16 * , 2798 ( 2007 ) ; j .-
kneur , m. b. pinto , r. o. ramos and e. staudt , phys .
d * 76 * , 045020 ( 2007 ) ; phys .
b * 657 * , 136 ( 2007 ) ; int .
* e16 * , 2802 ( 2007 ) .
m. b. pinto and r. o. ramos , phys .
d * 60 * , 105005 ( 1999 ) .
m. b. pinto and r. o. ramos , phys .
d * 61 * , 125016 ( 2000 ) .
v. i. yukalov , moscow univ .
phys . bull . *
31 * , 10 ( 1976 ) ; theor . math . phys .
* 28 * , 652 ( 1976 ) ; r. seznec and j. zinn - justin , j. math . phys . * 20 * , 1398 ( 1979 ) ; j. c. le guillou and j. zinn - justin , ann . phys . * 147 * , 57 ( 1983 ) ; v. i. yukalov , moscow univ .
phys . bull .
* 31 * , 10 ( 1976 ) ; w. e. caswell , ann .
( n.y ) * 123 * , 153 ( 1979 ) ; i. g. halliday and p. suranyi , phys . lett .
b * 85 * , 421 ( 1979 ) ; j. killinbeck , j. phys .
* a14 * , 1005 ( 1981 ) ; r. p. feynman and h. kleinert , phys .
a * 34 * , 5080 ( 1986 ) ; h. f. jones and m. moshe , phys .
b * b234 * , 492 ( 1990 ) ; a. neveu , nucl . phys .
* b18 * , 242 ( 1991 ) ; v. yukalov , j. math .
phys * 32 * , 1235 ( 1991 ) ; c. m. bender , k. a. milton and m. moshe , phys . rev .
d * 45 * , 1248 ( 1992 ) ; s. gandhi and m. b. pinto , phys .
d * 46 * , 2570 ( 1992 ) ; h. yamada , z. phys .
* c59 * , 67 ( 1993 ) ; k. g. klimenko , z. phys . *
c60 * , 677 ( 1993 ) ; a.n .
sissakian , i. l. solovtsov and o. p. solovtsova , phys .
b * 321 * , 381 ( 1994 ) ; h. kleinert , phys . rev .
d * 57 * , 2264 ( 1998 ) ; phys .
b * 434 * , 74 ( 1998 ) ; for a review , see h. kleinert and v. schulte - frohlinde , _ critical properties of @xmath138-theories _ , chap .
19 ( world scientific , singapure 2001 ) ; m. b. pinto , r. o. ramos and p. j. sena , physica * a342 * , 570 ( 2004 ) . g. krein , d.p .
menezes and m.b .
pinto , phys .
b * 370 * , 5 ( 1996 ) ; g. krein , r. s. m. de carvalho , d.p .
menezes , m. nielsen and m. b. pinto , eur .
j. a * 1 * , 45 ( 1998 ) ; g. krein , d.p .
menezes and m.b .
pinto , braz . j. phys .
* 28 * , 66 ( 1998 ) .
p. m. stevenson , phys .
d * 23 * , 2916 ( 1981 ) .
abdalla , j.a .
helayel - neto , daniel l. nedel and carlos r. senise , jr , phys .
d * 77 * , 125020 ( 2008 ) .
buckley , a. duncan and h.f .
jones , phys .
d * 47 * , 2554 ( 1993 ) ; c. m. bender , a. duncan and h.f .
jones , phys .
d * 49 * , 4219 ( 1994 ) ; c. arvanitis , h. f. jones and c.s .
parker , phys .
d * 52 * , 3704 ( 1995 ) ; h. kleinert and w. janke , phys . lett . *
a206 * , 283 ( 1995 ) ; r. guida , k. konishi and h. suzuki , ann .
( n.y . ) * 241 * ( 1995 ) 152 ; _ ibid . _ * 249 * , 109 ( 1996 ) ; b. bellet , p. garcia and a. neveu , int .
j. of mod .
a11 * , 5587 ( 1996 ) ; _ ibid .
_ * a11 * , 5607 ( 1996 ) .
a. duncan and h. f. jones , phys .
d * 47 * , 2560 ( 1993 ) .
kneur and d. reynaud , eur .
j. * c24 * , 323 ( 2002 ) ; phys .
d * 66 * , 085020 ( 2002 ) .
kneur , m. b. pinto and r. o. ramos , phys .
. lett . * 89 * , 210403 ( 2002 ) ; phys .
a * 68 * , 043615 ( 2003 ) ; e. braaten and e. radescu , phys .
lett . * 89 * , 271602 ( 2002 ) ; phys .
a * 66 * , 063601 ( 2002 ) .
s. chiku and t. hatsuda , phys .
d * 58 * , 076001 ( 1998 ) .
l. dolan and r. jackiw , phys .
d * 9 * , 3320 ( 1974 ) ; s. weinberg , ibid .
* 9 * , 3357 ( 1974 ) ; a. linde , rep .
* 42 * , 389 ( 1979 ) .
lee , nucl . phys . *
b9 * , 649 ( 1969 ) ; b.w .
lee , chiral dynamics,(gordon and breach , new york , 1972 ) t. kugo , prog .
. phys . * 57 * , 593 ( 1977 ) . | the thermodynamics of a scalar field with a quartic interaction is studied within the linear @xmath0 expansion ( lde ) method .
using the imaginary - time formalism the free energy is evaluated up to second order in the lde .
the method generates nonperturbative results that are then used to obtain thermodynamic quantities like the pressure .
the phase transition pattern of the model is fully studied , from the broken to the symmetry restored phase .
the results are compared with those obtained with other nonperturbative methods and also with ordinary perturbation theory .
the results coming from the two main optimization procedures used in conjunction with the lde method , the principle of minimal sensitivity ( pms ) and the fastest apparent convergence ( fac ) are also compared with each other and studied in which cases they are applicable or not .
the optimization procedures are applied directly to the free energy . |
neutrinos are the only massless fermions predicted by the standard model ( sm ) .
this seems to be a reasonable assumption as none of the experiments designed to measure the neutrino mass in laboratory experiments have found any positive evidence for a non - zero neutrino mass . at present
the existing limits from laboratory searches are @xcite : @xmath0 the square of the electron neutrino mass is measured in tritium beta decay experiments by fitting the end point distribution . in several of these experiments there has been found a negative mass squared which is concluded to be due to unknown effects which cause the accumulation of events near the endpoint .
this makes the limit above still far from certain .
the muon neutrino mass limit is derived from the measurement of the muon neutrino momenta on the decay @xmath1 , while the tau neutrino mass limit given above is based on kinematics of @xmath2 decays . for a detail discussion on the @xmath2 neutrino mass limit see @xcite . however , the confidence on the masslessness of the neutrino is now under question due to the important results of underground experiments , starting by the geochemical experiments of davis and collaborators till the more recent gallex , sage , kamiokande and superkamiokande experiments @xcite .
altogether they provide solid evidence for the existence of anomalies in the solar and the atmospheric neutrino fluxes .
particularly relevant has been the recent confirmation by the superkamiokande collaboration @xcite of the atmospheric neutrino zenith - angle - dependent deficit which strongly indicates towards the existence of @xmath3 conversion . together with these results
there is also the indication for neutrino oscillations in the @xmath4 channel by the lsnd experiment @xcite .
if one tries to include all these requirements in a single framework , we finds three mass scales involved in neutrino oscillations .
the simplest way to reconcile these requirements invokes the existence of a light sterile neutrino i.e. one whose interaction with standard model particles is much weaker than the sm weak interaction so it does not affect the invisible z decay width , precisely measured at lep @xcite . to this
we may add the possible role of neutrinos in the dark matter problem and structure formation @xcite .
at the moment , evidence for a solar neutrino deficit comes from four experiments @xcite , homestake , kamiokande , gallex and sage experiments .
the most recent data on the rates can be summarized as : @xmath5 the different experiments are sensitive to different parts of the energy spectrum of solar neutrinos and putting all these results together seems to indicate that the solution to the problem is not astrophysical but must concern the neutrino properties . moreover , non - standard astrophysical solutions are strongly constrained by helioseismology studies @xcite . within the standard solar model approach , the theoretical predictions clearly lie far from the best - fit solution what leads us to conclude that new particle physics is the only way to account for the data . the standard explanation for this deficit would be the oscillation of @xmath6 to another neutrino species either active or sterile .
different analyses have been performed to find the allowed mass differences and mixing angles in the two - flavour approximation @xcite .
the last result from refs .
@xcite indicate that for oscillations into active neutrinos there are three possible solutions for the parameters : + * vacuum ( also called `` just so '' ) oscillations with @xmath7@xmath8 ev@xmath9 and @xmath10@xmath11 * non - adiabatic - matter - enhanced oscillations via the msw mechanism @xcite with @xmath12@xmath13 ev@xmath9 and @xmath14@xmath15 , and * large mixing via the msw mechanism with @xmath16@xmath17 ev@xmath9 and @xmath18@xmath11 . in fig .
[ msw ] i show the allowed two - flavour regions obtained in an updated msw global fit analysis of the solar neutrino data for the case of active neutrino conversions .
the analysis uses the model from @xcite but with an arbitrary @xmath19 ( from the reaction @xmath20 ) neutrino flux @xcite .
[ vac98 ] shows the regions of just - so oscillation parameters obtained in a recent global fit of the data .
it has been pointed out that the expected seasonal effects in this scenario ( due to the variation of the earth - sun distance ) could be used to further constrain the parameters @xcite , and also to help discriminating it from the msw transition . for oscillations into an sterile neutrino
there are differences partly due to the fact that now the survival probability depends both on the electron and neutron density in the sun but mainly due to the lack of neutral current contribution to the kamiokande experiment .
this last effect requires a larger @xmath6 survival probability . as a result
the vacuum oscillation solution is very marginal and the large mixing msw solution is ruled out .
the small mixing solution is still valid @xcite .
the large mixing solution for oscillations into sterile neutrinos is also in conflict with the constraints from big bang nucleosynthesis ( bbn ) @xcite .
the presence of additional weakly interacting light particles , such as a light sterile neutrino , is constrained by bbn since the @xmath21 would enter into equilibrium with the active neutrinos in the early universe via neutrino oscillations .
however the derivation of the bbn bounds may be subject to large systematical uncertainties .
for example , it has been argued in @xcite that present observations of primordial helium and deuterium abundances can allow up to @xmath22 neutrino species if the baryon to photon ratio is small .
the presence of a relic lepton number asymmetry in the early universe may also relax this constraint @xcite .
atmospheric showers are initiated when primary cosmic rays hit the earth s atmosphere .
secondary mesons produced in this collision , mostly pions and kaons , decay and give rise to electron and muon neutrino and anti - neutrinos fluxes @xcite .
there has been a long - standing anomaly between the predicted and observed @xmath3 @xmath23 ratio of the atmospheric neutrino fluxes @xcite . although the absolute individual @xmath3 or @xmath6 fluxes are only known to within @xmath24 accuracy , different authors agree that the @xmath3 @xmath23 ratio is accurate up to a @xmath25 precision . in this
resides our confidence on the atmospheric neutrino anomaly ( ana ) , now strengthened by the high statistics sample collected at the super - kamiokande experiment @xcite .
the most important feature of the atmospheric neutrino 535-day data sample reported by the sk collaboration at neutrino 98 @xcite is that it exhibits a _
zenith - angle - dependent _ deficit of muon neutrinos which is inconsistent with expectations based on calculations of the atmospheric neutrino fluxes .
this experiment has marked a turning point in the significance of the ana .
-0.5 cm in fig .
[ ang_mu ] i show the measured zenith angle distribution of electron - like and muon - like sub - gev and multi - gev events , as well as the one predicted in the absence of oscillation .
i also give the expected distribution in various neutrino oscillation schemes .
the thick - solid histogram is the theoretically expected distribution in the absence of oscillation , while the predictions for the best - fit points of the various oscillation channels is indicated as follows : for @xmath26 ( solid line ) , @xmath27 ( dashed line ) and @xmath28 ( dotted line ) . the error displayed in the experimental points is only statistical . in the theoretical analysis
it has been used the latest improved calculations of the atmospheric neutrino fluxes as a function of zenith angle , including the muon polarization effect and took into account a variable neutrino production point @xcite .
clearly the data are not reproduced by the no - oscillation hypothesis , adding substantially to our confidence that the atmospheric neutrino anomaly is real .
the most likely solution of the ana involves neutrino oscillations @xcite . in principle
we can invoke various neutrino oscillation channels , involving the conversion of @xmath3 into either @xmath6 or @xmath29 ( active - active transitions ) or the oscillation of @xmath3 into a sterile neutrino @xmath21 ( active - sterile transitions ) @xcite . in fig .
[ mutausk4 ] i show the allowed neutrino oscillation parameters obtained in a recent global fit of the sub - gev and multi - gev ( vertex - contained ) atmospheric neutrino data @xcite including the recent data reported at neutrino 98 , as well as all other experiments combined at 90 ( thick solid line ) and 99 % cl ( thin solid line ) for each oscillation channel considered . -0.5 cm the two lower panels in fig .
[ mutausk4 ] differ in the sign of the @xmath30 which was assumed in the analysis of the matter effects in the earth for the @xmath26 oscillations .
it was found that @xmath31 oscillations give a slightly better fit than @xmath32 oscillations . at present
the atmospheric neutrino data can not distinguish between the @xmath33 and @xmath34 channels .
notice that in all channels where matter effects play a role the range of acceptable @xmath35 is shifted towards larger values , when compared with the @xmath36 case .
this follows from looking at the relation between mixing _ in vacuo _ and in matter .
in fact , away from the resonance region , independently of the sign of the matter potential , there is a suppression of the mixing inside the earth . as a result
, there is a lower cut in the allowed @xmath35 value , and it lies higher than what is obtained in the data fit for the @xmath36 channel .
i also display in fig .
[ mutausk4 ] the sensitivity of present accelerator and reactor experiments , as well as that expected at future long - baseline ( lbl ) experiments in each channel .
the first point to note is that the chooz reactor @xcite data already excludes the region indicated for the @xmath37 channel when all experiments are combined at 90% cl . from the upper - left panel in fig .
[ mutausk4 ] one sees that the regions of @xmath28 oscillation parameters obtained from the atmospheric neutrino data analysis can not be fully tested by the lbl experiments , as presently designed .
one might expect that , due to the upward shift of the @xmath35 indicated by the fit for the sterile case ( due to the effects of matter in the earth ) it would be possible to completely cover the corresponding region of oscillation parameters .
although this is the case for the minos disappearance test , in general most of the lbl experiments can not completely probe the region of oscillation parameters allowed by the @xmath38 atmospheric neutrino analysis .
this is so irrespective of the sign of @xmath35 assumed .
for a discussion of the various potential tests that can be performed at the future lbl experiments in order to unravel the presence of oscillations into sterile channels see ref .
@xcite .
los alamos meson physics facility ( lsnd ) has searched for @xmath39 oscillations with @xmath40 from @xmath41 decay at rest @xcite .
the @xmath42 s are detected in the quasi elastic process @xmath43 in correlation with a monochromatic photon of @xmath44 mev arising from the neutron capture reaction @xmath45 . in ref .
@xcite they report a total of 22 events with @xmath46 energy between 36 and @xmath47 mev while @xmath48 background events are expected .
they fit the full @xmath46 event sample in the energy range @xmath49 mev by a @xmath50 method and the result yields @xmath51 beam - related events . subtracting the estimated neutrino background with a correlated gamma of @xmath52 events results into an excess of @xmath53 events .
the interpretation of this anomaly in terms of @xmath54 oscillations leads to an oscillation probability of ( @xmath55)% . using a likelihood method they obtain a consistent result of ( @xmath56)% . in the two - family formalism this result leads to the oscillation parameters shown in fig .
the shaded regions are the 90 % and 99 % likelihood regions from lsnd . also shown
are the limits from bnl776 , karmen1 , bugey , ccfr , and nomad .
-0.5 cm .2 cm .1 cm there is increasing evidence that more than @xmath57 of the mass in the universe is dark and non - baryonic .
neutrinos , if massive , constitute a source for dark matter .
stable neutrinos can fill the universe of hot dark matter if their masses add up to a maximum of about 30 ev .
however , scenarios with only hot dark matter run into trouble in the explanation of the formation of structures on small scales of the universe .
the research on the nature of the cosmological dark matter and the origin of galaxies and large scale structure in the universe within the standard theoretical framework of gravitational collapse of fluctuations as the origin of structure in the expanding universe has undergone tremendous progress recently .
indeed the observations of cosmic background temperature anisotropies on large scales performed by the cobe satellite @xcite combined with cluster - cluster correlation data e.g. from iras @xcite can not be reconciled with the simplest cold dark matter ( cdm ) model . currently , the best scenario for a zero cosmological constant to explain the data considers a mixture of cold plus hot dark matter @xcite .
this translates into an upper limit on neutrino masses : @xmath58 this mass scale is similar to that indicated by the hints reported by the lsnd experiment @xcite .
very recent data on type - i supernovae at high redshifts @xcite has provided evidence at more than 99 % cl for an accelerating expansion of the universe .
they measure the light curve of the supernovae which gives the absolute luminosity . in this way they are able to determine the distance as a function of the redshift , and from there to measure the deceleration parameter @xmath59 .
they find @xmath60 .
@xmath61 gives a measurement of the different contributions to the energy density in the universe coming from matter and from the presence of a cosmological constant . in a flat universe
both contributions must verify @xmath62=1 .
in other words , the data from supernovae searches indicate a non - zero cosmological constant , or equivalently , @xmath63 .
actually the results indicate , for a flat universe , @xmath64 at 99 % cl . should these results be confirmed the amount of dark matter of the universe would be considerably reduced and in consequence the corresponding limit on the stable neutrino mass will become tighter .
future sky maps of the cosmic microwave background radiation ( cmbr ) with high precision at the upcoming map and planck missions should bring more light into the nature of the dark matter and the possible role of neutrinos .
naive two - family counting shows that it is very difficult to fit all experimental information even in the three neutrino scenario , even without invoking the lsnd data .
one has to choose between throwing away part of the data and considering a larger scheme .
the solar neutrino deficit could be due to @xmath65 oscillations and the atmospheric neutrino deficit to @xmath66 oscillations with the appropriate mass differences , for example with a mass hierarchy @xmath67 .
however , fitting this together with the present laboratory limits leaves no room for hot dark @xcite .
the only possible way out is to require that all three neutrinos are almost degenerate .
this requires a certain degree of fine - tuning in order to explain the neutrinoless double beta decay data .
notice that this scenario is also inconsistent with the oscillation parameters observed by lsnd .
one could have @xmath68 oscillations for the atmospheric neutrino deficit with almost degenerate @xmath3 and @xmath29 with masses @xmath69 few ev and @xmath70 , but leaving out the explanation for the solar neutrino deficit . or @xmath71 and @xmath72 few ev to explain the atmospheric data but leaving unexplained both solar neutrino deficit and dark matter . also , it is possible to explain the solar neutrino deficit with @xmath73 with almost degenerate @xmath6 and @xmath74 with masses @xmath75 few ev and @xmath76 , but leaving the atmospheric neutrino deficit unexplained . also @xmath77 and @xmath78 ev would explain the lsnd data if confirmed but leaves both atmospheric and dark matter without explanation .
the minimal " scheme to explain @xmath79 data without fine - tuning seems to be a four - neutrino framework @xmath80 where @xmath21 is a sterile neutrino .
the simplest way to open the possibility of incorporating the lsnd scale to the solar and atmospheric neutrino scales is to invoke a sterile neutrino , i.e. one whose interaction with standard model particles is much weaker than the sm weak interaction so it does not affect the invisible z decay width , precisely measured at lep .
the sterile neutrino must also be light enough in order to participate in the oscillations involving the three active neutrinos @xcite .
after imposing the present constrains from the negative searches at accelerator @xcite and reactor @xcite neutrino oscillation experiments one is left with two possible mass patterns as described in fig .
[ masses ] which i will call scenario i and ii .
in scenario i there are two lighter neutrinos at the solar neutrino mass scale and two maximally mixed almost degenerate ev - mass neutrinos split by the atmospheric neutrino scale . in scenario ii the two lighter neutrinos are maximally mixed and split by the atmospheric neutrino scale while the two heavier neutrinos are almost degenerate separated by the solar neutrino mass difference . in both scenarios solar neutrino data together with reactor neutrino constrains , imply that the electron neutrino must be maximally projected over one of the states belonging to the pair split by the solar neutrino scale : the lighter ( heavier ) pair for scenario i ( ii ) . on the other hand , atmospheric neutrino data together with the bounds from accelerator neutrino oscillation experiments imply that the muon neutrino must be maximally projected over the pair split by the atmospheric neutrino mass difference : the heavier ( lighter ) pair for scenario i ( ii ) . in both scenarios
there are two possible assignments for the sterile and tau neutrinos which i denote by .a and .b depending on whether the tau neutrino is maximally projected over the pair responsible for the atmospheric neutrino oscillations and the sterile neutrino is responsible for the solar neutrino deficit ( @xmath81 and @xmath82 ) or viceversa ( @xmath83 and @xmath84 ) .
these four possibilities offer different signatures at future experiments : * in scenario ii the electron neutrino must be mainly composed of one of the heavier states with a mass characteristic of the lsnd mass difference and the dark matter @xmath85 and may be tested at future neutrinoless double-@xmath86 decay and tritium @xmath86 decay experiments . * as mentioned before in the atmospheric neutrino section , scenarios i.a and ii.a give a slightly better fit to the atmospheric neutrino anomaly .
future long baseline experiments can be sensitive to this oscillation and the most sensitive test would be a @xmath2 appearance experiment . *
scenarios i.b and ii.b where the atmospheric @xmath3 deficit is due to the oscillation into an sterile neutrino imply a higher value of @xmath87 as can be seen in fig .
[ mutausk4 ] due to the effect of propagation through the earth which suppresses the lower mass region . as a consequence
this scenario can be easier to test at future long baseline experiments .
however only a disappearance - type experiment is possible and , in general these tests can achieve lower sensitivity . * for solar neutrinos
the three regions discussed in subsection [ solar ] are valid for scenarios i.a and ii.a when the solar data is explained in terms of @xmath88 . for scenarios i.b and ii.b where @xmath89 are invoked to account for the solar neutrino deficit , there are differences mainly due to the lack of neutral current contribution to the kamiokande experiment .
this last effect requires a larger @xmath6 survival probability . as a result
the vacuum oscillation solution is very marginal and the large mixing msw solution is ruled out . * the neutral - to - charged current ratio is a very important observable in neutrino oscillation phenomenology , which is especially sensitive to the existence of singlet neutrinos .
this test can be carried out both at future solar and atmospheric neutrino experiments as well as long baseline experiments . at present
one may study the ratios of @xmath90-events and the events induced mainly by the charged currents @xcite .
superkamiokande has reported the result @xcite @xmath91 the expected values are 1 .
( 0.75 ) for scenarios i.a and ii.a ( i.b and ii.b ) .
the result above is consistent with both scenarios with a slight preference for the former . * in scenarios i.b and ii.b the sterile neutrino is largely mixed with one active neutrino .
this gives a larger contribution to the effective degrees of freedom at the time of bbn .
should the bbn constrains become more precise , these scenarios may be ruled out .
the impressive re - confirmation of an angle - dependent atmospheric neutrino deficit by superkamiokande leaves little room for doubt that the atmospheric neutrino anomaly is an strong evidence for neutrino masses and mixings .
likewise , it has become more and more difficult to avoid neutrino oscillations as an explanation for the solar neutrino puzzle .
also the lsnd evidence for @xmath42-@xmath40-oscillations still remains a viable hypothesis although more restricted by the exclusion limit of the karmen experiment .
these three results can be interpreted in terms of neutrino oscillations but with the need of three different mass scales .
thus if the lsnd result stands the test of time , this would be a puzzling indication for the existence of a light sterile neutrino .
the two scenarios to reconcile these observations invoke either @xmath88 oscillations to explain the solar data , with @xmath92 oscillations accounting for the atmospheric deficit , or viceversa .
they have distinct implications at future tritium beta decay and neutrino - less double beta decay experiments as well as solar , atmospheric and long baseline neutrino experiments .
in particular the neutral - to - charged current ratio is an important observable to discriminate among the different scenarios as it is sensitive to the existence of singlet neutrinos .
this test can be carried out both at future solar and atmospheric neutrino experiments as well as long baseline experiments . *
acknowledgments * + i am grateful to a. pich and a. ruiz for the kind hospitality in santander . my thanks to john
bahcall and bill louis , for providing me with their postscript figures .
this work was supported by grants cicyt aen96 - 1718 , dgicyt pb95 - 1077 and dgicyt pb97 - 1261 , and by the eec under the tmr contract erbfmrx - ct96 - 0090 .
kamiokande collaboration , y. fukuda _ et al .
lett . * 77 * , 1683(1996 ) ; gallex collaboration : p. anselmann _
et al . _ ,
b*342 * , 440 ( 1995 ) and w. hampel _
et al . _ ,
b*388 * , 364 ( 1996 ) ; sage collaboration , v. gavrin _ et al .
_ , in neutrino _ 96 _ , proceedings of the 17th international conference on neutrino physics and astrophysics , helsinki , edited by k. huitu , k. enqvist and j. maalampi ( world scientific , singapore , 1997 ) , p. 14 ; superkamiokande collaboration , y. suzuki , proceedings neutrino 98 , of the xviii international conference on neutrino physics and astrophysics , takayama , japan , 49 june 1998 , y. suzuki , y. totsuka ( eds . ) . to be published in nucl
b ( proc . suppl . ) nusex collaboration , m. aglietta _
et al . _ ,
* 8 * , 611 ( 1989 ) ; frjus collaboration , ch .
et al . _ ,
* b227 * , 489 ( 1989 ) ; imb collaboration , d. casper _
et al . _ ,
66 * , 2561 ( 1991 ) ; r. becker - szendy _ et al .
. rev . * d46 * , 3720 ( 1992 ) ; kamiokande collaboration , h. s. hirata _
et al . _ ,
* b205 * , 416 ( 1988 ) and phys . lett . *
b280 * , 146 ( 1992 ) ; kamiokande collaboration , y. fukuda _ et al .
_ , phys . lett . *
b335 * , 237 ( 1994 ) ; soudan collaboration , w. w. m allison _ et al .
b391 * , 491 ( 1997 ) .
peltoniemi and j.w.f .
valle , nucl .
phys.*b406 * , 409 ( 1993 ) . d.o .
caldwell and r.n .
mohapatra , phys .
rev.*d48 * , 3259 ( 1993 ) .
gomez - cadenas and m.c .
gonzalez - garcia , z. phys .
* c71 * , 443 ( 1996 ) .
bilenky , c. giunti and w. grimus , eur .
j. * c1 * , 247 ( 1998 ) .
n. okada and o. yasuda , int .
j. mod . phys . * a12 * , 3669 ( 1997 ) .
caldwell , hep - ph/9804367 , to be publ . in proc .
cosmo97 . q.y .
liu and a.yu .
smirnov , hep - ph/9712493 .
v. barger , t.j .
weiler and k. whisnant , phys .
* b427 * , 97 ( 1998 ) ; v. barger , s. pakvasa , t.j . weiler and k. whisnant , phys .
* d58 * 093016 ( 1998 ) ; s. c. gibbons , r. n. mohapatra , s. nandi , and a. raychaudhuri , phys .
* b430 * , 296 ( 1998 ) .
primack , j. holtzman , a. klypin and d.o .
caldwell , phys .
lett . * 74 * , 2160 ( 1995 ) ; d. pogosyan and a.a .
starobinsky , astrophys .
j. * 447 * , 465 ( 1995 ) ; a. klypin , r. nolthenius and j. primack , astrophys . j. * 474 * , 533 ( 1997 ) ; r.w . strickland and d.n .
schramm , astrophys .
j. * 481 * , 571 ( 1997 ) ; see also , e. gawiser and j. silk , science , * 280 * , 1405 ( 1998 ) , and references therein .
l. krauss , e. gates and m. white , phys .
* b299 * , 94 ( 1993 ) ; p. i. krastev and s. t. petkov , phys.lett .
* b299 * , 99 ( 1993 ) ; p. i. krastev and s. t. smirnov , phys . lett . *
b338 * , 282 ( 1994 ) .
j. n. bahcall , p. krastev , a.yu .
smirnov , phys .
rev . * d58 * 096016 ( 1998 )
. j. n. bahcall , p. i. krastev , phys .
* b436 * 243 ( 1988 ) . n. hata and p. langacker , phys .
* d50 * , 632 ( 1994 ) ; s.m .
bilenky and c. giunti , phys . lett . *
b320 * , 323 ( 1994 ) ; j. t. peltoniemi hep - ph/9506228 .
x. shi , d.n .
schramm and b. d. fields , phys .
* d48 * , 2563 ( 1993 ) ; r. barbieri and a. dolgov , phys .
* b 237 * , 440 ( 1990 ) ; k. enquist , k. kainulainen and j. maalampi , phys . lett .
* b 249 * , 531 ( 1992 ) ; d. p. kirilova and m. chizov , nucl .
* b534 * 447 ( 1998 ) ; s.m .
bilenkii , c. giunti , w .
grimus , t. schwetz , hep - ph/9804421 .
j. g. learned , s. pakvasa , and t. j. weiler , phys .
* b207 * , 79 ( 1988 ) ; v. barger and k. whisnant , phys .
* b209 * , 365 ( 1988 ) ; k. hidaka , m. honda , and s. midorikawa , phys .
* 61 * , 1537 ( 1988 ) .
r. foot , r. r. volkas , o. yasuda phys . rev .
* d56 * 6653 ( 1997 ) ; phys .
* d58 * 013006 ( 1998 ) ; o. yasuda , phys .
* d58 * 091301 ( 1998 ) ; g. l. fogli , e. lisi , a. marrone , g. scioscia , hep - ph/9808205 ; e.kh .
akhmedov , a.dighe , p. lipari and a.yu .
smirnov , hep - ph/9808270 ; g.l .
fogli , e. lisi , phys .
* d52 * , 2775 ( 1995 ) ; g.l .
fogli , e. lisi , d. montanino , g. scioscia , phys . rev .
* d55 * , 4385 ( 1997 ) .
a. g. riess _ et al .
_ , astro - ph/9805201 ; s. perlmutter _
et al . _ ,
nature * 391 * ( 1998 ) 51 .
g. l. fogli , e. lisi and d. montanino , phys .
rev . * d49 * , 3626 ( 1994 ) ; astropart .
phys . * 4 * , 177 ( 1995 ) .
cdhsw collaboration , f. dydak _
et al . _ ,
b134 * , 281 ( 1984 ) ; e776 collaboration , l.borodvsky _ et al .
_ , phys .
* 68 * , 274 ( 1992 ) ; e531 collaboration , phys .
lett . * 57 * , 2898 ( 1986 ) ; karmen collaboration , k. eitel _
et al . _ hep - ex/9809007 ; chorus collaboration , e. eskut _
et al . _ ,
* b434 * ( 1998 ) ; nomad collaboration , j. altegoer _ et al . _ ,
* b431 * ( 1998 ) 219 . | i review the status of neutrino masses and mixings in the light of the solar and atmospheric neutrino data .
the result from the lsnd experiment and the possible role of neutrinos as hot dark matter are also included .
i also discuss the simplest schemes proposed to reconcile these data which include a light sterile neutrino in addition to the three standard ones .
implications for future experiments are commented . |
throughout the paper , we shall denote , for any enough differentiable function @xmath8 , its @xmath9-th derivative by @xmath10 or @xmath11 .
let @xmath0}$ ] be the linear brownian motion started at @xmath5 and @xmath12}$ ] be the linear brownian bridge within the time interval @xmath13 $ ] : @xmath12}=(b(t)|b(0)=b(1)=0)_{t\in [ 0,1]}$ ] .
these processes are gaussian processes with covariance functions @xmath14 for a given continuous function @xmath15 , the functions @xmath16 and @xmath17 respectively defined on @xmath13 $ ] by @xmath18 are the solutions of the respective boundary value problems on @xmath13 $ ] : @xmath19 observe that the differential equations are the same in both cases .
only the boundary conditions differ .
they are dirichlet - type boundary conditions for brownian bridge while they are dirichlet / neumann - type boundary conditions for brownian motion .
these well - known connections can be extended to the polyharmonic operator @xmath20 where @xmath3 is a positive integer .
this latter is associated with the @xmath2-fold integral of brownian motion @xmath21}$ ] : @xmath22$.}\ ] ] ( notice that all of the derivatives at time @xmath5 naturally vanish : @xmath23 . ) indeed , the following facts for instance are known ( see , e.g. , @xcite and @xcite ) : * the covariance fonction of the process @xmath21}$ ] coincide with the green function of the boundary value problem @xmath24 , \\ v(0)=v'(0)=\dots = v^{(n-1)}(0)=0 , \\ v^{(n)}(1)=v^{(n+1)}(1)=\dots = v^{(2n-1)}(1)=0 ; \end{cases}\ ] ] * the covariance fonction of the bridge @xmath25}$ ] coincide with the green function of the boundary value problem @xmath24 , \\ v(0)=v'(0)=\dots = v^{(n-1)}(0)=0 , \\ v^{(n-1)}(1)=v^{(n+1)}(1)=\dots = v^{(2n-1)}(1)=0 ; \end{cases}\ ] ] * the covariance fonction of the bridge @xmath26}$ ] coincide with the green function of the boundary value problem @xmath24 ,
\\ v(0)=v'(0)=\dots = v^{(n-1)}(0)=0 , \\
v(1)=v'(1)=\dots = v^{(n-1)}(1)=0 . \end{cases}\ ] ] observe that the differential equations and the boundary conditions at @xmath5 are the same in all cases .
only the boundary conditions at @xmath6 differ .
other boundary value problems can be found in @xcite and @xcite .
we refer the reader to @xcite for a pioneering work dealing with the connections between general gaussian processes and green functions ; see also @xcite .
we also refer to @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and the references therein for various properties , namely asymptotical study , of the iterated integrals of brownian motion as well as to @xcite , @xcite , @xcite and @xcite for interesting applications of these processes to statistics .
the aim of this work is to examine all the possible conditioned processes of @xmath21}$ ] involving different events at time @xmath6 : @xmath27}\ ] ] for a certain number @xmath28 of events , @xmath29 , and certain indices @xmath30 such that @xmath31 , and to make the connection with the boundary value problems : @xmath32 , \\ v(0)=v'(0)=\dots = v^{(n-1)}(0)=0 , \\
v^{(i_1)}(1)=v^{(i_2)}(1)=\dots = v^{(i_n)}(1)=0 \end{cases } \\[-7ex ] \noalign{\hfill(bvp)}\end{aligned}\ ] ] for certain indices @xmath33 such that @xmath34 .
actually , we shall see that this connection does not recover all the possible boundary value problems and we shall characterize those sets of indices for which such a connection exists .
the paper is organized as follows . in section [ sect - gaussian ] , we exhibit the relationships between general gaussian processes and green functions of certain boundary value problems . in section [ sect - iteratedibm ]
, we consider the iterated integrals of brownian motion . in section [ sect - bridges ] , we construct several bridges associated with the foregoing processes and depict explicitly their connections with the polyharmonic operator together with various boundary conditions .
one of the main results is theorem [ th - bvp - gene ] .
moreover , we exhibit several interesting properties of the bridges ( theorems [ th - drift ] and [ th - decompo - cov ] ) and solve the prediction problem ( theorems [ th - prediction ] ) . in section [ sect - ibm ] ,
we illustrate the previous results on the case @xmath35 related to integrated brownian motion
. finally , in section [ sect - general ] , we give a characterization for the green function of the boundary value problem ( bvp ) to be a covariance function . another one of the main results is theorem [ green - sym ] .
we consider a @xmath3-markov gaussian process @xmath36}$ ] evolving on the real line @xmath37 . by `` @xmath3-markov ''
, it is understood that the trajectory @xmath38 is @xmath3 times differentiable and the @xmath3-dimensional process @xmath39}$ ] is a markov process .
let us introduce the covariance function of @xmath36}$ ] : for @xmath40 $ ] , @xmath41 $ ] .
it is known ( see @xcite ) that the function @xmath42 admits the following representation : @xmath43 where @xmath44 , @xmath45 , are certain functions .
let @xmath46 be linear differential operators of order less than @xmath47 and let @xmath48 be a linear differential operator of order @xmath47 defined by @xmath49 where @xmath50 are continuous functions on @xmath13 $ ] .
more precisely , we have for any @xmath47 times differentiable function @xmath8 and any @xmath51 $ ] , @xmath52 [ th - gene ] assume that the functions @xmath44 , @xmath45 , are @xmath47 times differentiable and satisfy the following conditions , for a certain constant @xmath53 : @xmath54= \begin{cases } 0 & \mbox{if } \ , 0\le i\le p-2,\\ \kappa & \mbox{if } \
i = p-1 , \end{cases}\ ] ] @xmath55 then , for any continuous function @xmath15 on @xmath13 $ ] , the function @xmath56 defined on @xmath13 $ ] by @xmath57 solves the boundary value problem @xmath58 if the problem ( [ bvp - gene ] ) is determining , that is if it has a unique solution , then the covariance function @xmath42 is exactly the green function of the boundary value problem ( [ bvp - gene ] ) . in view of ( [ cov - gene ] ) ,
the function @xmath56 can be written as @xmath59\!.\ ] ] the derivative of @xmath56 is given by @xmath60\end{aligned}\ ] ] and its second order derivative , since @xmath61=0 $ ] , by @xmath62u(t)\right ] \\ & = \sum_{k=0}^{n-1}\left[\psi''_k(t ) \int_0^t \varphi_k(s)u(s)\,{\mathrm{d}}s + \varphi''_k(t ) \int_t^1 \psi_k(s)u(s)\,{\mathrm{d}}s\right]\!.\end{aligned}\ ] ] more generally , because of the assumptions ( [ condition - sum ] ) , we easily see that , for @xmath63 , @xmath64\end{aligned}\ ] ] and the @xmath47-th order derivative of @xmath56 is given by @xmath65u(t)\right ] \\ & = \sum_{k=0}^{n-1}\left[\psi^{(p)}_k(t ) \int_0^t \varphi_k(s)u(s)\,{\mathrm{d}}s + \varphi^{(p)}_k(t ) \int_t^1
\psi_k(s)u(s)\,{\mathrm{d}}s\right]+\kappa u(t).\end{aligned}\ ] ] actually , we have proved that , for @xmath63 , @xmath66 finally , due to ( [ condition - deriv ] ) , @xmath67+\kappa\alpha_p u(t ) = \kappa\alpha_p u(t).\end{aligned}\ ] ] concerning the boundary value conditions , referring to ( [ condition - deriv ] ) , we similarly have @xmath68 the proof of theorem [ th - gene ] is finished . in the two next sections , we construct processes connected to the equation @xmath69 subject to the boundary value conditions at @xmath5 : @xmath70 for @xmath71 and others at @xmath6 that will be discussed subsequently , where @xmath48 and @xmath72 are the differential operators ( @xmath48 being of order @xmath73 ) defined by @xmath74
let @xmath0}$ ] be the linear brownian motion limited to the time interval @xmath13 $ ] and started at @xmath5 .
we introduce the @xmath2-fold integral of brownian motion : for any @xmath51 $ ] , @xmath75 in particular , @xmath76 .
the trajectories of @xmath1}$ ] are @xmath3 times differentiable and we have @xmath77 for @xmath78 .
moreover , we have at time @xmath5 the equalities @xmath23 .
the process @xmath1}$ ] is a @xmath3-markov gaussian process since the @xmath3-dimensional process @xmath79}$ ] is markovian .
the covariance function of the gaussian process @xmath1}$ ] is given by @xmath80 in order to apply theorem [ th - gene ] , we decompose @xmath81 into the form ( [ cov - gene ] ) .
we have for , e.g. , @xmath82 , @xmath83 { \mathrm{d}}u \\ & = \sum_{k=0}^{n-1 } ( -1)^{n-1-k}\,\frac{t^k}{k ! } \int_0^s \frac{(s - u)^{n-1}}{(n-1)!}\ , \frac{u^{n-1-k}}{(n-1-k)!}\ , { \mathrm{d}}u \\ & = \sum_{k=0}^{n-1 } ( -1)^{n-1-k}\,\frac{s^{2n-1-k}}{(2n-1-k)!}\,\frac{t^k}{k!}.\end{aligned}\ ] ] we then obtain the following representation : @xmath84 with , for any @xmath45 , @xmath85 we state below a result of @xcite that we revisit here by using theorem [ th - gene ] .
[ th - bvp - ibm ] let @xmath15 be a fixed continuous function on @xmath13 $ ] .
the function @xmath56 defined on @xmath13 $ ] by @xmath86 is the solution of the boundary value problem @xmath87 , \\
v^{(i)}(0)=0 & \mbox{for } i\in\{0,1,\dots , n-1\ } , \\
v^{(i)}(1)=0 & \mbox{for } i\in\{n , n+1,\dots,2n-1\}. \end{cases}\ ] ] let us check that the conditions ( [ condition - sum ] ) and ( [ condition - deriv ] ) of theorem [ th - gene ] are fulfilled .
first , we have @xmath88}$ } } & = \sum_{k=0}^{n-1 } ( -1)^{n-1-k}\left[{1\hspace{-.27em}\mbox{\rm l}}_{\{k\ge i\ } } \,\frac{t^{2n-1-k}}{(2n-1-k)!}\,\frac{t^{k - i}}{(k - i ) ! } -{1\hspace{-.27em}\mbox{\rm
l}}_{\{k\le 2n-1-i\}}\,\frac{t^{2n-1-i - k}}{(2n-1-i - k)!}\,\frac{t^k}{k!}\right ] \\ & = ( -1)^{n-1}\,\frac{t^{2n-1-i}}{(2n-1-i ) ! } \left[\vphantom{\sum_n^n}\right.\!{1\hspace{-.27em}\mbox{\rm
l}}_{\{i\le n-1\}}\sum_{k = i}^{n-1 } ( -1)^{k}\binom{2n-1-i}{k - i } -\sum_{k=0}^{(2n-1-i)\wedge(n-1 ) } ( -1)^{k}\binom{2n-1-i}{k}\!\!\left
. \vphantom{\sum_n^n}\right ] \\ & = ( -1)^{n-1}\,\frac{t^{2n-1-i}}{(2n-1-i ) ! } \left[{1\hspace{-.27em}\mbox{\rm l}}_{\{i\le n-1\}}\left(\,\sum_{k=0}^{n-1-i } ( -1)^{i+k}\binom{2n-1-i}{k } -\sum_{k=0}^{n-1 } ( -1)^{k}\binom{2n-1-i}{k}\right)\!\!\right .
\\ & \hphantom{=\,}\left.-{1\hspace{-.27em}\mbox{\rm l}}_{\{i\ge n\}}\sum_{k=0}^{2n-1-i } ( -1)^{k}\binom{2n-1-i}{k}\!\right]\!.\end{aligned}\ ] ] performing the transformation @xmath89 in the first sum lying within the last equality , we get @xmath90 and then @xmath91 = ( -1)^n\delta_{i,2n-1}.\ ] ] on the other hand , setting @xmath92 we clearly see that , for any @xmath45 , @xmath93 consequently , by theorem [ th - gene ] , we see that the function @xmath56 solves the boundary value problem ( [ bvp - ibm ] ) .
the uniqueness part will follow from a more general argument stated in the proof of theorem [ th - bvp - gene ] .
in this section , we construct various bridges related to @xmath1}$ ] . more precisely , we take @xmath1}$ ] conditioned on the event that certain derivatives vanish at time @xmath6 .
let us recall that all the derivatives at time @xmath5 naturally vanish : @xmath23 .
for any @xmath94 , let @xmath95 be a subset of @xmath96 with @xmath31 and the convention that for @xmath97 , @xmath98 .
we see that for each @xmath94 , we can define @xmath99 subsets of indices @xmath100 , and the total number of sets @xmath100 is then @xmath101 .
set for any @xmath51 $ ] @xmath102 in this way , we define @xmath103 processes @xmath104}$ ] that we shall call `` bridges '' of @xmath1}$ ] . in particular , * for @xmath98 , we simply have @xmath104}=(x_n(t))_{t\in[0,1]}$ ] ; * for @xmath105 , the corresponding process is the @xmath2-fold integral of brownian bridge @xmath106}=\left(\int_0^t \frac{(t - s)^{n-1}}{(n-1)!}\ , { \mathrm{d}}\beta(s)\right)_{t\in[0,1]};\ ] ] * for @xmath107 , the corresponding process is the `` single '' bridge of @xmath2-fold integral of brownian brownian : @xmath108}=\left(\int_0^t \frac{(t - s)^{n-1}}{(n-1)!}\,{\mathrm{d}}b(s ) \,\bigg|
\int_0 ^ 1 \frac{(1-s)^{n-1}}{(n-1)!}\,{\mathrm{d}}b(s)=0\right)_{t\in[0,1]};\ ] ] * for @xmath109 , the corresponding process is @xmath110}.\ ] ] this is the natural bridge related to the @xmath3-dimensional markov process @xmath111}.$ ] in this section , we exhibit several interesting properties of the various processes @xmath112}$ ] .
one of the main goals is to relate these bridges to additional boundary value conditions at @xmath6 . for this , we introduce the following subset @xmath113 of @xmath114 : @xmath115\ ] ] with @xmath116 the cardinality of @xmath113 is @xmath3 .
actually , the set @xmath113 will be used further for enumerating the boundary value problems which can be related to the bridges labeled by @xmath100 .
conversely , @xmath113 yields @xmath100 through @xmath117 in the table below , we give some examples of sets @xmath113 and @xmath100 .
@xmath118 & & & & \\ \hline \end{array}\ ] ] below , we provide a representation of @xmath112}$ ] by means of @xmath1}$ ] subject to a random polynomial drift .
[ th - drift ] we have the distributional identity @xmath119}\stackrel{d}{=}\bigg(x_n(t)-\sum_{j\in j } p_j(t)x_j(1)\bigg)_{t\in[0,1]}\ ] ] where the functions @xmath120 , @xmath121 , are hermite interpolation polynomials on @xmath13 $ ] characterized by @xmath122 p_j^{(i)}(0)=0 & \mbox{for } i\in\{0,1,\dots , n-1\ } , \\[1ex ] p_j^{(i)}(1)=\delta_{j , n - i } & \mbox{for } i\in i. \end{cases}\ ] ] in the case where @xmath35 , we retrieve a result of @xcite .
moreover , the conditions ( [ condition - hermite ] ) characterize the polynomials @xmath120 , @xmath121 .
we prove this fact in lemma [ hermite ] in the appendix . by invoking classical arguments of gaussian processes theory
, we have the distributional identity @xmath119}\stackrel{d}{=}\bigg(x_n(t)-\sum_{j\in j } p_j(t)x_j(1)\bigg)_{t\in[0,1]}\ ] ] where the functions @xmath120 , @xmath121 , are such that @xmath123=0 $ ] for all @xmath124 .
we get the linear system @xmath125\,p_j(t)=\mathbb{e}[x_n(t)x_k(1)],\quad k\in j.\ ] ] we plainly have @xmath126=\int_0^{s\wedge t } \frac{(s - u)^{j-1}}{(j-1)!}\ , \frac{(t - u)^{k-1}}{(k-1)!}\,{\mathrm{d}}u.\ ] ] then , the system ( [ system ] ) writes @xmath127 the matrix of this system @xmath128 is regular as it can be seen by introducing the related quadratic form which is definite positive . indeed , for any real numbers @xmath129 , @xmath121 , we have @xmath130 and @xmath131 thus , the system ( [ system ] ) has a unique solution . as a result ,
the @xmath120 are linear combinations of the functions @xmath132 which are polynomials of degree less than @xmath133 .
hence , @xmath120 is a polynomial of degree at most @xmath134 .
we now compute the derivatives of @xmath120 at @xmath5 en @xmath6 .
we have @xmath135 for @xmath71 since the functions @xmath132 plainly enjoy this property . for checking that @xmath136 for @xmath137 , we successively compute * for @xmath71 , @xmath138=\mathbb{e}[x_{n - i}(t)x_k(1)];\ ] ] * for @xmath139 , @xmath140=\mathbb{e}[b(t)x_k(1 ) ] = \int_0^t \frac{(1-u)^{k-1}}{(k-1)!}\,{\mathrm{d}}u;\ ] ] * for @xmath141 , @xmath140=\frac{(1-t)^{k-1}}{(k-1)!};\ ] ] * for @xmath142 , @xmath140 = ( -1)^{i+n}\,\frac{(1-t)^{k+n - i-1}}{(k+n - i-1)!};\ ] ] * for @xmath143 , @xmath140=(-1)^{k-1};\ ] ] * for @xmath144 , @xmath140=0.\ ] ] consequently , at time @xmath145 , for @xmath146 , @xmath147\big|_{t=1}=(-1)^{k-1}\delta_{i , k+n-1}.\ ] ] now , by differentiating ( [ system ] ) , we get for @xmath71 , @xmath148 \
, p_j^{(i)}(1)=\mathbb{e}[x_{n - i}(1)x_k(1)],\quad k\in j.\ ] ] in particular , if @xmath149 , this can be rewritten as @xmath148\,p_j^{(i)}(1 ) = \sum_{j\in j } \delta_{j , n - i}\,\mathbb{e}[x_j(1)x_k(1)],\quad k\in j,\ ] ] which by identification yields @xmath136
. similarly , for @xmath150 , in view of ( [ deriv - cov ] ) , we have @xmath148\,p_j^{(i)}(1 ) = ( -1)^{k-1}\delta_{i ,
k+n-1},\quad k\in j.\ ] ] in particular , if @xmath151 , we have @xmath152 for @xmath124 , and then @xmath148\,p_j^{(i)}(1)=0,\quad k\in j,\ ] ] which by identification yields @xmath153 . the proof of theorem [ th - drift ] is finished
. let @xmath154 be the covariance function of @xmath104}$ ] : @xmath155 $ ] .
in the next theorem , we supply a representation of @xmath154 of the form ( [ cov - gene ] ) .
[ th - decompo - cov ] the covariance function of @xmath104}$ ] admits the following representation : for @xmath40 $ ] , @xmath156 with , for any @xmath45 , @xmath157 moreover , the functions @xmath158 , @xmath45 , are hermite interpolation polynomials such that @xmath159 \tilde{\psi}_k^{(i)}(0)=\delta_{i , k } & \mbox{for } i\in\{0,1,\dots , n-1\ } , \\[1ex ] \tilde{\psi}_k^{(i)}(1)=0 & \mbox{for } i\in i. \end{cases}\ ] ] we decompose @xmath160 into the difference @xmath161 we have @xmath162+\mathbb{e}[z(s)z(t)]-\mathbb{e}[x_n(s)z(t)]-\mathbb{e}[z(s)x_n(t ) ] \\ & = c_{_{x_n}}(s , t)+\sum_{j , k\in j } \mathbb{e}[x_j(1)x_k(1)]\,p_j(s)p_k(t ) \\ & \hphantom{=\,}-\sum_{j\in j } \mathbb{e}[x_n(t)x_j(1)]\,p_j(s ) -\sum_{j\in j } \mathbb{e}[x_n(s)x_j(1)]\,p_j(t).\end{aligned}\ ] ] by definition ( [ system ] ) of the @xmath120 s , we observe that @xmath163\,p_j(s)p_k(t ) -\sum_{k\in j } \mathbb{e}[x_n(s)x_k(1)]\,p_k(t)}$ } } & = \sum_{k\in j } \left(\vphantom{\sum_n^n}\right.\!\sum_{j\in j } \mathbb{e}[x_j(1)x_k(1)]\,p_j(s)-\mathbb{e}[x_n(s)x_k(1 ) ] \!\!\left.\vphantom{\sum_n^n}\right)\!p_k(t ) = 0.\end{aligned}\ ] ] then , we can simplify @xmath164 into @xmath165\,p_j(s).\ ] ] since the covariance functions @xmath154 and @xmath81 are symmetric , we also have @xmath166\,p_j(t).\ ] ] let us introduce the symmetric polynomial @xmath167\,p_j(t ) = \sum_{j\in j } \mathbb{e}[x_n(t)x_j(1)]\,p_j(s).\ ] ] it can be expressed by means of the functions @xmath44 s as follows : @xmath168 we can rewrite @xmath164 as @xmath169\!.\ ] ] and then , for @xmath40 $ ] , @xmath170 we immediately see that @xmath158 is a polynomial of degree less than @xmath171 such that @xmath172 for @xmath71 and , since @xmath136 , @xmath173 we deduce that @xmath174 \tilde{\psi}_k^{(i)}(1)=\psi_k^{(i)}(1)=0 & \mbox{if } i\in\{n , n+1,\dots,2n-1\}\backslash ( j+n-1 ) .
\end{cases}\ ] ] then @xmath175 for any @xmath137 .
this ends up the proof of theorem [ th - decompo - cov ] . in this section , we write out the natural boundary value problem which is associated with the process @xmath104}$ ] .
the following statement is the main connection between the different boundary value conditions associated with the operator @xmath20 and the different bridges of the process @xmath176}$ ] introduced in this work .
[ th - bvp - gene ] let @xmath15 be a fixed continuous function on @xmath13 $ ] .
the function @xmath56 defined on @xmath13 $ ] by @xmath177 is the solution of the boundary value problem @xmath178 , \\
v^{(i)}(0)=0 & \mbox{for } i\in\{0,1,\dots , n-1\ } , \\
v^{(i)}(1)=0 & \mbox{for } i\in i. \end{cases}\ ] ] @xmath179 _ first step .
_ recall that @xmath180 .
we decompose the function @xmath56 into the difference @xmath181 where , for @xmath51 $ ] , @xmath182 we know from theorem [ th - bvp - ibm ] that @xmath183 , @xmath184 for @xmath71 and @xmath185 for @xmath150 .
moreover , the function @xmath186 being a polynomial of degree less than @xmath171 , the function @xmath187 is also a polynomial of degree less than @xmath171 .
then @xmath188 , @xmath189 for @xmath71 and @xmath190 on the other hand , we learn from ( [ deriv - v ] ) that , for @xmath191 , @xmath192 and then , for @xmath71 , @xmath193\,u(s)\,{\mathrm{d}}s.\ ] ] we also have , for any @xmath191 , @xmath194\,p^{(i)}_j(1 ) = { 1\hspace{-.27em}\mbox{\rm l}}_{\{i\in ( n - j)\}}\mathbb{e}[x_n(s)x_{n - i}(1)].\ ] ] as a result , we see that @xmath195 this implies that for @xmath149 , @xmath196 and for @xmath151 , @xmath197 .
then @xmath196 for any @xmath137 , that is @xmath198 .
the function @xmath56 is a solution of ( [ bvp - bridge ] ) .
@xmath179 _ second step .
_ we now check the uniqueness of the solution of ( [ bvp - bridge ] ) .
let @xmath199 and @xmath200 be two solutions of @xmath69 . then the function @xmath201 satisfies @xmath202 , @xmath203 for @xmath71 and @xmath185 for @xmath137 .
we compute the following `` energy '' integral : @xmath204_0 ^ 1+(-1)^n\int_0 ^ 1 w(t)w^{(2n)}(t)\,{\mathrm{d}}t \\ & = ( -1)^{n+1}\sum_{i=0}^{n-1 } ( -1)^{i } w^{(i)}(1)w^{(2n-1-i)}(1).\end{aligned}\ ] ] we have constructed the set @xmath113 in order to have @xmath205 for any @xmath71 : when we pick an index @xmath71 , either @xmath137 or @xmath206 . indeed , * if @xmath207 , @xmath208 ; * if @xmath209 , by observing that @xmath210 , we have @xmath211 . since @xmath212 , we see that @xmath206 and then @xmath213 .
next , @xmath214 which entails @xmath215 , that is , @xmath216 is a polynomial of degree less than @xmath3 .
moreover , with the boundary value conditions at @xmath5 , we obtain @xmath217 or @xmath218 . the proof of theorem [ th - bvp - gene ] is finished .
[ remark - unique ] we have seen in the above proof that uniqueness is assured as soon as the boundary value conditions at @xmath6 satisfy @xmath205 for any @xmath71 .
these conditions are fulfilled when the set @xmath219 is such that @xmath220 , @xmath221 , @xmath222 , @xmath223 , @xmath224 .
this is equivalent to say that @xmath113 and @xmath225 make up a partition of @xmath114 , or @xmath226 in this manner , we get @xmath103 different boundary value problems which correspond to the @xmath103 different bridges we have constructed .
we shall see in section [ sect - general ] that the above identity concerning the differentiating set @xmath113 characterizes the possibility for the green function of the boundary value problem ( [ bvp - bridge ] ) to be the covariance of a gaussian process .
now , we tackle the problem of the prediction for the process @xmath104}$ ] .
[ th - prediction ] fix @xmath227 $ ] .
the shifted process @xmath228}$ ] admits the following representation : @xmath229}=\left(\tilde{y}_{t_0}(t)+\sum_{i=0}^{n-1 } q_{i , t_0}(t ) y^{(i)}(t_0 ) \right)_{t\in[0,1-t_0]}\ ] ] where @xmath230 the process @xmath231}$ ] is a copy of @xmath232}$ ] which is independent of @xmath233}$ ] .
the functions @xmath234 , @xmath121 , and @xmath235 , @xmath71 , are hermite interpolation polynomials on @xmath236 $ ] characterized by @xmath237 \tilde{p}^{(\iota)}_{j , t_0}(0)=0 & \mbox{for } \iota\in\{0,1,\dots , n-1\ } , \\[1ex ] \tilde{p}^{(\iota)}_{j , t_0}(1-t_0)=\delta_{\iota , n - j}&\mbox{for } \iota\in i , \end{cases}\ ] ] and @xmath238 q^{(\iota)}_{i , t_0}(0)=\delta_{\iota , i } & \mbox{for } \iota\in\{0,1,\dots , n-1\ } , \\[1ex ] q^{(\iota)}_{i , t_0}(1-t_0)=0 & \mbox{for } \iota\in i. \end{cases}\ ] ] actually , these functions can be expressed by means of the functions @xmath120 , @xmath121 , and @xmath239 , @xmath71 , as follows : @xmath240 in other words , the process @xmath241}$ ] is a bridge of length @xmath242 which is independent of @xmath243}$ ] , that is @xmath244}\stackrel{d}{= } \big(\tilde{x}_n(t ) \big| \tilde{x}_j(1-t_0)=0,j\in j\big)_{t\in[0,1-t_0]}.\ ] ] @xmath179 _ first step . _
fix @xmath227 $ ] .
we have the well - known decomposition , based on the classical prediction property of brownian motion stipulating that @xmath245}=\big(\tilde{b}_{t_0}(t)+b(t_0)\big)_{t\in[0,1-t_0]}$ ] where @xmath246}$ ] is a brownian motion independent of @xmath247}$ ] , @xmath248}=\left(\tilde{x}_n(t)+\sum_{k=0}^{n-1 } \frac{t^k}{k ! } \ , x_{n - k}(t_0)\right)_{t\in[0,1-t_0]}\ ] ] with @xmath249 . differentiating this equality @xmath250 times , @xmath251
, we obtain @xmath252 therefore , @xmath253 \nonumber\\ & = \!\left[\vphantom{\sum_n^n}\right.\!\tilde{x}_n(t)-\sum_{j\in j } p_j(t+t_0 ) \tilde{x}_j(1-t_0 ) \!\!\left.\vphantom{\sum_n^n}\right ] + \sum_{k=0}^{n-1 } \left[\vphantom{\sum_n^n}\right.\!\frac{t^k}{k ! } -\sum_{j\in j : j\ge n - k } \frac{(1-t_0)^{j+k - n}}{(j+k - n)!}\ , p_j(t+t_0 ) \!\!\left.\vphantom{\sum_n^n}\right ] x_{n - k}(t_0 ) .
\label{y - translated}\end{aligned}\ ] ] we are going to express the @xmath254 , @xmath255 , by means of the @xmath256 , @xmath121 .
we have , by differentiating ( [ y - translated ] ) @xmath9 times , for @xmath71 , @xmath257 \\ & \hphantom{=\ , } + \sum_{k=0}^{n-1 } \left[\vphantom{\sum_n^n}\right.\!\frac{t^{k - i}}{(k - i)!}{1\hspace{-.27em}\mbox{\rm l}}_{\{k\ge i\ } } -\sum_{j\in j : j\ge n - k } \frac{(1-t_0)^{j+k - n}}{(j+k - n)!}\ , p^{(i)}_j(t+t_0)\!\!\left.\vphantom{\sum_n^n}\right]\ !
x_{n - k}(t_0).\end{aligned}\ ] ] for @xmath258 , this yields @xmath259 x_{n - k}(t_0).\ ] ] set for @xmath260 @xmath261 and let us introduce the matrix @xmath262 together with its inverse matrix @xmath263 the equalities ( [ y - i - tzero ] ) for @xmath71 read as a linear system of @xmath2 equations and @xmath2 unknowns : @xmath264 which can be rewritten into a matrix form as @xmath265 the solution is given by @xmath266 and we see that @xmath254 , @xmath45 , is of the form @xmath267\tilde{x}_j(1-t_0).\ ] ] therefore , by plugging ( [ x - tzero ] ) into ( [ y - translated ] ) , we obtain @xmath268 \\ & \hphantom{=\,}\times\!\left[\vphantom{\sum_n^n}\right.\!\sum_{i=0}^{n-1 } b_{ki } y^{(i)}(t_0 ) + \sum_{j\in j } \left(\sum_{i=0}^{n-1 } b_{ki}p^{(i)}_j(t_0)\right)\tilde{x}_j(1-t_0 ) \!\!\left.\vphantom{\sum_n^n}\right ] \\ & = \tilde{x}_n(t)-\sum_{j\in j } \left[\vphantom{\sum_n^n}\right.\!p_j(t+t_0)- \sum_{0\le i , k\le n-1 } b_{ki } p^{(i)}_j(t_0 ) \\
& \hphantom{=\ , } \times\left(\vphantom{\sum_n^n}\right.\!\frac{t^{k}}{k ! }
-\sum_{\jmath\in j:\jmath\ge n - k } \frac{(1-t_0)^{\jmath+k - n}}{(\jmath+k - n)!}\ , p_{\jmath}(t+t_0 ) \!\!\left.\vphantom{\sum_n^n}\right)\!\!\left.\vphantom{\sum_n^n}\right]\!\tilde{x}_j(1-t_0 ) \\ & \hphantom{=\ , } + \sum_{0\le i , k\le n-1 } b_{ki}\!\left[\vphantom{\sum_n^n}\right.\!\frac{t^{k}}{k ! }
-\sum_{j\in j : j\ge n - k } \frac{(1-t_0)^{j+k - n}}{(j+k - n)!}\ , p_j(t+t_0 ) \!\!\left.\vphantom{\sum_n^n}\right]\
! y^{(i)}(t_0).\end{aligned}\ ] ] finally , @xmath269 can be written as @xmath270 where @xmath271\ ] ] and @xmath272 with @xmath273 \\ & = p_j(t+t_0)-\sum_{i=0}^{n-1 } p^{(i)}_j(t_0)q_{i , t_0}(t).\end{aligned}\ ] ] @xmath179 _ second step .
_ we easily see that the functions @xmath234 and @xmath235 are polynomials of degree less than @xmath171 .
let us compute now their derivatives at @xmath5 and @xmath274 .
first , concerning @xmath235 we have @xmath275\!.\ ] ] choosing @xmath258 and recalling the definition of @xmath276 and the fact that the matrices @xmath277 and @xmath278 are inverse , this gives for @xmath279 , @xmath280 = \sum_{k=0}^{n-1}a_{\iota k } b_{ki}=\delta_{\iota , i}.\ ] ] choosing @xmath281 , we have for @xmath282 @xmath283\!.\ ] ] by theorem [ th - drift ] , we know that @xmath284 for @xmath282 ; then @xmath285 observing that , if @xmath286 , the conditions @xmath287 and @xmath282 are equivalent , we simply have @xmath288 which immediately entails , by ( [ qitzero ] ) , @xmath289 next , concerning @xmath290 , we have @xmath291 choosing @xmath258 , this gives for @xmath279 , since @xmath292 , @xmath293 choosing @xmath281 , we have for @xmath282 , since @xmath294 and @xmath295 , @xmath296 the polynomials @xmath234 ( resp .
@xmath235 ) enjoy the same properties than the @xmath120 s ( resp . the @xmath239 s ) , regarding the successive derivatives , they can be deduced from these latter by a rescaling according as @xmath297 it is then easy to extract the identity in distribution below , by using the property of gaussian conditioning : @xmath298 } \stackrel{d}{= } \big(\tilde{x}_n(t ) \big| \tilde{x}_j(1-t_0)=0,j\in j\big)_{t\in[0,1-t_0]}.\ ] ] theorem [ th - prediction ] is established .
here , we have a look on the particular case where @xmath35 for which the corresponding process @xmath1}$ ] is nothing but integrated brownian motion ( the so - called langevin process ) : @xmath299 the underlying markov process is the so - called kolmogorov diffusion @xmath300}$ ] . all the associated conditioned processes that will be constructed are related to the equation @xmath301 with boundary value conditions at time @xmath5 : @xmath302 .
there are four such processes : * @xmath303}$ ] ( integrated brownian motion ) ; * @xmath304}$ ] ( integrated brownian bridge ) ; * @xmath305}$ ] ( bridge of integrated brownian motion ) ; * @xmath306}$ ] ( another bridge of integrated brownian motion ) . on the other hand ,
when adding two boundary value conditions at time @xmath6 to the foregoing equation , we find six boundary value problems : @xmath307 , @xmath308 , @xmath309 , @xmath310 , @xmath311 , @xmath312 . actually , only four of them can be related to some gaussian processes the above listed processes in the sense of our work whereas two others can not be . for each process
, we provide the covariance function , the representation by means of integrated brownian motion subject to a random polynomial drift , the related boundary value conditions at @xmath6 and the decomposition related to the prediction problem . since the computations are straightforward , we shall omit them and we only report here the results . for an account on integrated brownian motion in relation with the present work , we refer the reader to , e.g. , @xcite and references therein .
the process corresponding to the set @xmath98 is nothing but integrated brownian motion : @xmath313}=\left(\int_0^t b(s)\,{\mathrm{d}}s\right)_{t\in [ 0,1]}.\ ] ] the covariance function is explicitly given by @xmath314 ^ 2\,[3(s\vee t)-s\wedge t].\ ] ] this process is related to the boundary value conditions at @xmath6 ( @xmath315 ) : @xmath312 .
the prediction property can be stated as follows : @xmath316}=\big(\tilde{x}_2(t)+x_2(t_0)+tx_1(t_0)\big)_{t\in [ 0,1-t_0]}.\ ] ] the process corresponding to the set @xmath105 is integrated brownian bridge : @xmath317}=\left(\int_0^t b(s)\,{\mathrm{d}}s\,\big|b(1)=0\right)_{t\in [ 0,1 ] } = \left(\int_0^t \beta(s)\,{\mathrm{d}}s\right)_{t\in [ 0,1 ] } = ( x_2(t)|x_1(1)=0)_{t\in [ 0,1]}.\ ] ] this process can be represented as @xmath317}=\left(x_2(t)-\frac12\,t^2 x_1(1)\right)_{t\in [ 0,1]}.\ ] ] the covariance function is explicitly given by @xmath314 ^ 2\,[3(s\vee t)-s\wedge t]-\frac14\,s^2t^2.\ ] ] the process @xmath112}$ ] is related to the boundary value conditions at @xmath6 ( @xmath318 ) : @xmath311 .
the prediction property says that @xmath319}=\left(\tilde{y}_{t_0}(t)+y(t_0 ) + \left(t-\frac{t^2}{2(1-t_0)}\right)y'(t_0)\right)_{t\in [ 0,1-t_0]}.\ ] ] the process corresponding to the set @xmath320 is the bridge of integrated brownian motion : @xmath317}=\left(\int_0^t b(s)\,{\mathrm{d}}s\,\bigg|\int_0 ^ 1 b(s)\,{\mathrm{d}}s=0\right)_{t\in [ 0,1 ] } = ( x_2(t)|x_2(1)=0)_{t\in [ 0,1]}.\ ] ] the bridge is understood as the process is pinned at its extremities : @xmath321 .
this process can be represented as @xmath317}=\left(x_2(t)-\frac12\,t^2(3-t ) x_2(1)\right)_{t\in [ 0,1]}.\ ] ] the covariance function is explicitly given by @xmath314 ^ 2\,[3(s\vee t)-s\wedge t]-\frac{1}{12}\,s^2t^2(3-s)(3-t).\ ] ] the process @xmath112}$ ] is related to the boundary value conditions at @xmath6 ( @xmath322 ) : @xmath308 .
the prediction property says that @xmath323}}$ } } = \left(\tilde{y}_{t_0}(t)+\frac{t^3 - 3(1-t_0)t^2 + 2(1-t_0)^3}{2(1-t_0)^3}\,y(t_0 ) + \frac{t^3 - 3(1-t_0)t^2 + 2(1-t_0)^2t}{2(1-t_0)^2}\,y'(t_0)\right)_{t\in [ 0,1-t_0]}.\end{aligned}\ ] ] the process corresponding to the set @xmath324 is another bridge of integrated brownian motion ( actually of the two - dimensional kolmogorov diffusion ) : @xmath317}=\left(\int_0^t b(s)\,{\mathrm{d}}s\,\bigg|\int_0 ^ 1 b(s)\,{\mathrm{d}}s = b(1)=0\right)_{t\in [ 0,1 ] } = ( x_2(t)|x_2(1)=x_1(1)=0)_{t\in [ 0,1]}.\ ] ] the bridge here is understood as the process is pinned at its extremities together with its derivatives : @xmath325 .
this process can be represented as @xmath317}=\left(x_2(t)-t^2(t-1 ) x_1(1)-t^2(3 - 2 t ) x_2(1)\right)_{t\in [ 0,1]}.\ ] ] the covariance function is explicitly given by @xmath314 ^ 2\,[1-s\vee t]^2\,[3(s\vee t)-s\wedge t-2st].\ ] ] the process @xmath112}$ ] is related to the boundary value conditions at @xmath6 ( @xmath326 ) : @xmath307 .
the prediction property says that @xmath319}=\left(\tilde{y}_{t_0}(t)+\frac{t^2(t+t_0 - 1)}{(1-t_0)^3}\,y(t_0 ) + \frac{t^2(3 - 3t_0 - 2t)}{(1-t_0)^2}\,y'(t_0)\right)_{t\in [ 0,1-t_0]}.\ ] ] @xmath179 the solution of the problem associated with the boundary value conditions @xmath309 ( which corresponds to the set @xmath327 ) is given by @xmath328,\ ] ] where @xmath329 ^ 2\,[3(s\vee t)-s\wedge t]+\frac16\,s^2t^2(s-3).\ ] ] @xmath179 the solution of the problem associated with the boundary value conditions @xmath310 ( which corresponds to the set @xmath330 ) , is given by @xmath331,\ ] ] where @xmath332 ^ 2\,[3(s\vee t)-s\wedge t]+\frac16\,s^2t^2(t-3).\ ] ] we can observe the relationships @xmath333 and @xmath334 .
the green functions @xmath335 and @xmath336 are not symmetric , so they can not be viewed as the covariance functions of any gaussian process . in the next section , we give an explanation of these observations .
in this last part , we address the problem of relating the general boundary value problem @xmath337 , \\ v(0)=v'(0)=\dots = v^{(n-1)}(0)=0 , \\ v^{(i_1)}(1)=v^{(i_2)}(1)=\dots = v^{(i_n)}(1)=0 , \end{cases}\ ] ] for any indices @xmath33 such that @xmath34 , to some possible gaussian process . set @xmath338 .
we have noticed in theorem [ th - bvp - gene ] and remark [ remark - unique ] that , when @xmath113 satisfies the relationship @xmath339 , the system ( [ bvp - gene - bis ] ) admits a unique solution .
we proved this fact by computing an energy integral .
actually , this fact holds for any set of indices @xmath113 ; see lemma [ hermite ] .
our aim is to characterize the set of indices @xmath113 for which the green function of ( [ bvp - gene - bis ] ) can be viewed as the covariance function of a gaussian process .
a necessary condition for a function of two variables to be the covariance function of a gaussian process is that it must be symmetric .
so , we shall characterize the set of indices @xmath113 for which the green function of ( [ bvp - gene - bis ] ) is symmetric and we shall see that in this case this function is a covariance function .
we first write out a representation for the green function of ( [ bvp - gene - bis ] ) .
[ th - bvp - gene - bis ] the boundary value problem ( [ bvp - gene - bis ] ) has a unique solution .
the corresponding green function admits the following representation , for @xmath40 $ ] : @xmath340 where the @xmath341 , @xmath282 , are hermite interpolation polynomials satisfying @xmath342 r_{{\scriptscriptstyle i},\iota}^{(i)}(0)=0 & \mbox{for } i\in\{0,1,\dots , n-1\ } , \\[.5ex ] r_{{\scriptscriptstyle i},\iota}^{(i)}(1)=\delta_{\iota , i } & \mbox{for } i\in i. \end{cases}\ ] ] the conditions ( [ condition - hermite - bis ] ) characterize the polynomials @xmath341 , @xmath282 .
we prove this fact in lemma [ hermite ] in the appendix .
let us introduce the functions @xmath199 and @xmath200 defined , for any @xmath51 $ ] , by @xmath343 we plainly have @xmath344 and @xmath345 .
therefore , the function @xmath56 solves the system ( [ bvp - gene - bis ] ) if and only if the function @xmath200 satisfies @xmath346 , & \\[1ex ] v_2^{(i)}(0)=0 & \mbox{for } i\in\{0,1,\dots , n-1\ } , \\ \displaystyle v_2^{(i)}(1)=(-1)^{n-1}\int_0 ^ 1 \frac{(1-s)^{2n-1-i}}{(2n-1-i ) ! } \,u(s)\,{\mathrm{d}}s & \mbox{for } i\in i. \end{cases}\ ] ] referring to lemma [ hermite ] , the conditions ( [ bvp - v2 ] ) mean that @xmath200 is a hermite interpolation polynomial which can be expressed as a linear combination of the @xmath341 , @xmath282 , defined in theorem [ th - bvp - gene - bis ] as follows : @xmath347 \!u(s)\,{\mathrm{d}}s.\ ] ] consequently , the boundary value problem ( [ bvp - gene - bis ] ) admits a unique solution @xmath56 which writes @xmath348,\ ] ] where @xmath349 is defined in theorem [ th - bvp - gene - bis ] . the proof is finished .
we now state two intermediate results which will be used in the proof of theorem [ green - sym ] .
[ different ] let @xmath350 and @xmath351 be two subsets of @xmath114 with cardinality @xmath3 . if the sets @xmath350 and @xmath351 are different , then the corresponding green functions @xmath352 and @xmath353 are different .
suppose that @xmath354 .
if we had @xmath355 , for any continuous function @xmath15 , the function @xmath56 defined on @xmath13 $ ] by @xmath356 would be a solution of the equation @xmath357 satisfying @xmath358 for @xmath71 and @xmath198 for @xmath359 . since @xmath354 and since @xmath360 have the same cardinality , there exists an index @xmath361 which belongs to @xmath362 . for this @xmath361 , we would have @xmath363 for all @xmath364 , or equivalently , @xmath365 this is impossible since the exponent @xmath366 does not appear in the polynomial on the left - hand side of the foregoing equality . as a result , @xmath367 .
[ sym ] let @xmath350 and @xmath351 be two subsets of @xmath114 with cardinality @xmath3 .
the relationship @xmath368 holds for any @xmath40 $ ] ( in other words , the integral operators with kernels @xmath352 and @xmath353 are dual ) if and only if the sets @xmath350 and @xmath351 are linked by @xmath369 .
we have , for any @xmath40 $ ] , @xmath370=\frac{(t - s)^{2n-1}}{(2n-1 ) ! } -\sum_{\iota\in i_1}\frac{(1-s)^{2n-1-\iota}}{(2n-1-\iota ) ! } r_{{\scriptscriptstyle i_1},\iota}(t ) + \sum_{\iota\in i_2}\frac{(1-t)^{2n-1-\iota}}{(2n-1-\iota ) ! } r_{{\scriptscriptstyle i_2},\iota}(s).\ ] ] set , for any @xmath40 $ ] , @xmath371.\ ] ] the polynomial @xmath372 has a degree less than @xmath171 with respect to each variable @xmath373 and @xmath374 and satisfy @xmath375 in particular , * for @xmath145 and @xmath376 , @xmath377 , @xmath378 * for @xmath379 and @xmath380 , @xmath381 , @xmath382 * for @xmath383 and @xmath384 , @xmath385 now we are able to establish the statement of proposition [ sym ] . *
if @xmath386 , there exists @xmath376 such that @xmath387 and then , in view of ( [ partial1 ] ) , @xmath388 the polynomial @xmath372 can not be null , that is , there exist @xmath40 $ ] such that @xmath389 . * if @xmath369 , for any @xmath376 , we have @xmath390 and then , in view of ( [ partial1 ] ) , @xmath391 put , for any @xmath380 , @xmath392 .
the polynomial @xmath393 has a degree less than @xmath171 . by ( [ partial2 ] ) and ( [ partial3 ] )
, we have @xmath394 \tilde{s}_{i_1}^{(i_2)}(1)=0 & \mbox{for } i_2\in i_2 , \end{cases}\ ] ] from which we deduce , invoking lemma [ hermite ] , that all the polynomials @xmath393 , @xmath380 , are null .
finally , the polynomial @xmath372 has a degree less than to @xmath171 with respect to @xmath374 and satisfies @xmath395 \displaystyle\frac{\partial^{i_1}\!s}{\partial t^{i_1}}\,(s,1)=0 & \mbox{for } i_1\in i_1 . \end{cases}\ ] ] we can assert , by lemma [ hermite ] , that @xmath372 is the null - polynomial . the proof of proposition [ sym ] is finished .
a necessary condition for @xmath396 to be the covariance function of a gaussian process is that it must be symmetric : @xmath397 for any @xmath40 $ ] .
the theorem below asserts that if the set of indices @xmath113 is not of the form displayed in the preamble of section [ sect - bridges ] , that is @xmath398 , the green function @xmath396 is not symmetric and consequently this function can not be viewed as a covariance function , that is we can not relate the boundary value problem ( [ bvp - gene - bis ] ) to any gaussian process . [
green - sym ] the green function @xmath396 is symmetric ( and it corresponds to a covariance function ) if and only if the set of indices @xmath113 satisfies @xmath339 .
set @xmath399 . by proposition [ sym ]
, we see that @xmath396 is symmetric if and only if @xmath400{$\scriptscriptstyle{\!i'}$}}(s , t)=g_{_{\!i}}(s , t)$ ] for any @xmath40 $ ] , that is , by proposition [ different ] , if and only if @xmath401 .
we made several verifications with the aid of maple .
below is the program we wrote for this . ....
[ > green_function:=proc(n , seti ) local v , m , s , t , p , seticomp ; v:=i->vector(n,[seq(binomial(k , i),k = n .. 2*n-1 ) ] ) ; m:=liste->stackmatrix(seq(v(i),i = liste ) ) ; s:=(s , liste)->matrix(1,n,[seq((1-s)^(2*n-1-i)/(i!*(2*n-1-i)!),i = liste ) ] ) ; t:=t->matrix(n,1,[seq([t^k],k = n .. 2*n-1 ) ] ) ; p:=(s , t , liste)->multiply(s(s , liste),inverse(transpose(m(liste))),t(t))[1,1 ] ; seticomp:=[op({seq(i , i=0 ..
2*n-1 ) } minus { seq(2*n-1-i , i = seti ) } ) ] ; print(`value of n`=n,`differentiating indices set i_1`=seti ) ; print(`green function for s <
t : gi_1(s , t ) ` = sort(simplify((t - s)^(2*n-1)/((2*n-1)!)-p(s , t , seti)),[s , t],plex ) ) ; print(`green function for s > t : gi_1(s , t ) ` = sort(simplify(-p(t , s , seti)),[s , t],plex ) ) ; print(`symmetry test : gi_1(s , t)=gi_1(t ,
s ) ? ` = evalb(simplify((t - s)^(2*n-1)/((2*n-1)!)-p(s , t , seti)+p(t , s , seti))=0 ) ) ; print(`complementary set of 2n+1-i_1 : i_2`=seticomp ) ; print(`difference between the two green functions for s <
t : gi_1(s , t)-gi_2(s , t ) ` = sort(simplify(-p(s , t , seti)+p(s , t , seticomp)),[s , t],plex ) ) ; print(`difference between the two green functions for s > t : gi_1(s , t)-gi_2(s , t ) ` = sort(simplify(-p(s , t , seticomp)+p(s , t , seti)),[s , t],plex ) ) ; print(`equality test between the two green functions for s < t : gi_1(s , t)=gi_2(t , s ) ? ` = evalb(simplify((t - s)^(2*n-1)/((2*n-1)!)-p(s , t , seti)+p(t , s , seticomp))=0 ) ) ; print(`equality test between the two green functions for s > t : gi_1(s , t)=gi_2(t , s ) ? ` = evalb(simplify((t - s)^(2*n-1)/((2*n-1)!)-p(s , t , seticomp)+p(t , s , seti))=0 ) ) ; end proc ; .... to obtain the two green functions associated with the sets @xmath350 and @xmath351 , @xmath402 and @xmath403 , together with the equality test between them , run the command ` [ > green_function(n , i_1 ) ; ` .
for instance , the return of the command ` [ > green_function(5,[2,3,5,6,8 ] ) ; ` is .... value of n = 3 , differentiating indices set i_1 = [ 1,4,5 ] green function for s < t : gi_1(s , t ) = -1/120 s^5 - 1/72 s^4 t^3 + 1/24 s^4 t + 1/18 s^3 t^3 - 1/12 s^3 t^2 green function for s > t : gi_1(s , t ) = -1/120 s^5 + 1/24 s^4 t - 1/72 s^3 t^4 + 1/18 s^3 t^3 - 1/12
s^3 t^2 symmetry test : gi_1(s , t)=gi_1(t , s ) ?
= false complementary set of 2n+1-i_1 : i_2 = [ 2,3,5 ] difference between the two green functions for s <
t : gi_1(s , t)-gi_2(s , t ) = -1/72 s^4 t^4 + 1/72 s^3 t^4 difference between the two green functions for s > t : gi_1(s , t)-gi_2(s , t ) = 1/72 s^4 t^3 - 1/72 s^3 t^4 equality test between the two green functions for s < t : gi_1(s , t)=gi_2(t , s ) ? = true equality test between the two green functions for s > t : gi_1(s , t)=gi_2(t , s ) ?
= true .... here , we have a look on the particular case where @xmath404 for which the corresponding process @xmath1}$ ] is the twice integrated brownian motion : @xmath405 all the associated conditioned processes that can be constructed are related to the equation @xmath406 with boundary value conditions at time @xmath5 : @xmath407 .
there are @xmath408 such processes .
since the computations are tedious and the explicit results are cumbersome , we only report the correspondance between bridges and boundary value conditions at time @xmath6 through the sets of indices @xmath113 and @xmath100 . these are written in the table below .
@xmath409 & & & & & & & & \\ \hline \end{array}\ ] ] the green functions related to the other sets can not be related to some gaussian processes .
the sets are written in the table below with the correspondance @xmath410 .
@xmath411 & & & & & & \\ \hline \end{array}\ ] ] for @xmath412 , only the @xmath413 following differentiating sets can be related to bridges : @xmath414
[ hermite ] let @xmath415 , @xmath71 , and @xmath416 , @xmath137 , be real numbers .
there exists a unique polynomial @xmath417 such that @xmath418 p^{(i)}(0)=a_i & \mbox{for } i\in\{0,1,\dots , n-1\ } , \\[.5ex ] p^{(i)}(1)=b_i & \mbox{for } i\in i. \end{cases}\ ] ] the conditions ( [ condition - hermite - ter ] ) characterize the hermite interpolation polynomial at points @xmath5 and @xmath6 with given values of the successive derivatives at @xmath5 up to order @xmath419 and given values of the derivatives at @xmath6 with _ selected _ orders in @xmath113 .
when @xmath420 , these polynomials differ from the usual hermite interpolation polynomials which involve the successive derivatives at certain points _
progressively _ from zero order up to certain orders .
we look for polynomials @xmath417 in the form @xmath421 we have @xmath422 we shall adopt the convention @xmath423=0 $ ] for @xmath424 .
the conditions ( [ condition - hermite - ter ] ) yield the linear system ( with the convention that @xmath9 and @xmath425 denote respectively the raw and column indices ) @xmath426 the @xmath427 matrix of this system writes @xmath428 \dotfill \\ \begin{pmatrix}\displaystyle\frac{1}{(j - i ) ! }
\end{pmatrix}_{\!\!\hspace{-4em}i\in i\atop \!\!j\in\{0,1,\dots,2n-1\ } } \end{pmatrix}=\begin{pmatrix } ( \delta_{i , j})_{i\in\{0,1,\dots , n-1\}\atop j\in\{0,1,\dots , n-1\ } } & & \;(0)_{\hspace{-1.4em}i\in\{0,1,\dots , n-1\}\atop j\in\{n , n+1,\dots,2n-1\ } } \\[-0.5ex ] \dotfill & & \hspace*{-.74em } \dotfill \\ \begin{pmatrix}\displaystyle\frac{1}{(j - i ) ! } \end{pmatrix}_{\!\!\hspace{-3.8em}i\in
i\atop \!\!j\in\{0,1,\dots , n-1\ } } & & \;\displaystyle\left(\frac{1}{(j - i)!}\right)_{\!\!\hspace{-5.1em}i\in i\atop \!\!j\in\{n , n+1,\dots,2n-1\ } } \end{pmatrix}\!. \hspace{-13.23em}\begin{array}{c}\\[-2.86ex]\vdots\\[-1.5ex]\vdots\\[-1.5ex]\vdots\\[-2.5ex]\vdots\\[-1.5ex]\vdots\end{array}\hspace{13em}\ ] ] proving the statement of lemma [ hermite ] is equivalent to proving that the matrix @xmath429 is regular . in view of the form of @xmath429 as a bloc - matrix ,
we see , since the north - west bloc is the unit matrix and the north - east bloc is the null matrix , that this is equivalent to proving that the south - east bloc of @xmath429 is regular .
let us call this latter @xmath430 and label its columns @xmath431 , @xmath432 : @xmath433 for proving that @xmath430 is regular , we factorize @xmath430 into the product of two regular triangular matrices .
the method consists in performing several transformations on the columns of @xmath430 which do not affect its rank .
we provide in this way an algorithm leading to a @xmath434-factorization of @xmath430 where @xmath435 is a lower triangular matrix and @xmath436 is an upper triangular matrix with no vanishing diagonal term .
we begin by performing the transformation @xmath437
\displaystyle \mathbf{c}_j^{(0)}-\frac{\mathbf{c}_{j-1}^{(0)}}{n+j - i_1 } & \mbox{if } j\in\{1,2,\dots , n-1\}. \end{cases}\ ] ] the generic term of the column @xmath438 , for @xmath439 , is @xmath440 this transformation supplies a matrix @xmath441 with columns @xmath438 , @xmath432 , which writes @xmath442\begin{pmatrix } \displaystyle\frac{1}{(n - i)!}\end{pmatrix}_{\!\scriptscriptstyle i\in i}\;\ ; & \begin{pmatrix } \displaystyle\frac{i - i_1}{n+j - i_1}\,\frac{1}{(n+j - i)!}\end{pmatrix}_{\!\!\hspace{-3.7em}i\in i\atop \!\!j\in\{1,2,\dots , n-1\ } } \end{pmatrix}\!. \hspace{-18.1em}\begin{array}{c}\\[-3.5ex]\vdots\\[-1.5ex]\vdots\end{array}\hspace{14em}\ ] ] we have written @xmath443 where @xmath444 is the triangular matrix with a diagonal made of @xmath6 below : @xmath445 we now perform the transformation @xmath446
\displaystyle \mathbf{c}_j^{(1)}-\frac{n+j - i_1 - 1}{n+j - i_1 } \,\frac{\mathbf{c}_{j-1}^{(1)}}{n+j - i_2 } & \mbox{if } j\in\{2,3,\dots , n-1\}. \end{cases}\ ] ] the generic term of the column @xmath447 , for @xmath448 , is @xmath449 this transformation supplies a matrix @xmath450 with columns @xmath447 , @xmath432 , which writes @xmath451\vdots\\[-1.5ex]\vdots\end{array}\hspace{32em } \\ &
\hspace{17.5em}\left.\begin{pmatrix } \displaystyle\frac{(i - i_1)(i - i_2)}{(n+j - i_1)(n+j - i_2)}\,\frac{1}{(n+j - i ) ! } \end{pmatrix}_{\!\!\hspace{-3.7em}i\in i\atop \!\!j\in\{2,3,\dots , n-1\ } } \right)\!. \hspace{-23.9em}\begin{array}{c}\\[-3.5ex]\vdots\\[-1.5ex]\vdots\end{array}\end{aligned}\ ] ] we have written @xmath452 where @xmath453 is the triangular matrix with a diagonal made of @xmath6 below : @xmath454 in a recursive manner , we easily see that we can construct a sequence of matrices @xmath455 , @xmath456 , such that @xmath457 where @xmath458 is the triangular matrix with a diagonal made of @xmath6 below : @xmath459 and @xmath460 \!\vphantom{\begin{pmatrix}\frac{1}{(n - i)!}\end{pmatrix}_{i\in i\atop j\in\{k+1,\dots , n-1\ } } } \begin{pmatrix } \displaystyle\frac{1}{(n - i)!}\end{pmatrix}_{\!\scriptscriptstyle i\in i}\;\ ; & \begin{pmatrix } \displaystyle\frac{i - i_1}{n+1-i_1}\,\frac{1}{(n+1-i)!}\end{pmatrix}_{\!\scriptscriptstyle i\in i}\;\ ; & \cdots \end{matrix}\right . \hspace{-15.4em}\begin{array}{c}\\[-3.5ex]\vdots\\[-1.5ex]\vdots\end{array } \hspace{12.5em}\begin{array}{c}\\[-3.5ex]\vdots\\[-1.5ex]\vdots\end{array}\hspace{19.4em } \hspace{-17.7em}\begin{array}{c}\\[-3.5ex]\vdots\\[-1.5ex]\vdots\end{array } } $ } } & \hspace{-4.8em } \begin{pmatrix } \displaystyle\frac{(i - i_1)\dots(i - i_{k-1})}{(n+k-1-i_1)\dots(n+k-1-i_{k-1 } ) } \,\frac{1}{(n+k-1-i)!}\end{pmatrix}_{\!\scriptscriptstyle i\in i}\;\ ; \hspace{-27.55em}\begin{array}{c}\\[-3.5ex]\vdots\\[-1.5ex]\vdots\end{array } \\ & \hspace{-4.8em } \left.\begin{matrix}\\[-1.8ex]\begin{pmatrix } \displaystyle\frac{(i - i_1)\dots(i - i_k)}{(n+j - i_1)\dots(n+j - i_k)}\,\frac{1}{(n+j - i ) ! } \end{pmatrix}_{\hspace{-3.4em}i\in
i\atop \!\!j\in\{k,\dots , n-1\ } } \end{matrix}\right)\!. \hspace{-24.9em}\begin{array}{c}\\[-3.5ex]\vdots\\[-1.5ex]\vdots\end{array}\end{aligned}\ ] ] we finally obtain , since all the @xmath461 , @xmath456 , are regular , that @xmath462 with @xmath463 and @xmath464\vdots\\[-1.5ex]\vdots\end{array } \hspace{12.48em}\begin{array}{c}\\[-3.5ex]\vdots\\[-1.5ex]\vdots\end{array}\hspace{19.6em } \hspace{-17.9em}\begin{array}{c}\\[-3.5ex]\vdots\\[-1.5ex]\vdots\end{array}\hspace{19em } \\ \left.\;\;\begin{pmatrix } \displaystyle\frac{(i - i_1)\dots(i - i_{n-1})}{(2n-1-i_1)\dots(2n-1-i_{n-1 } ) } \,\frac{1}{(2n-1-i)!}\end{pmatrix}_{\!\scriptscriptstyle i\in i } \right)\!. \hspace{-24em}\begin{array}{c}\\[-3.5ex]\vdots\\[-1.5ex]\vdots\end{array}\hspace{28.1em}\end{aligned}\ ] ] it is clear that the matrices @xmath436 and @xmath435 are triangular and regular , and then @xmath430 ( and @xmath429 ) is also regular .
moreover , the inverse of @xmath430 can be computed as @xmath465 the proof of lemma [ hermite ] is finished .
dolph , c. l. and woodbury , m. a. : on the relation between green s functions and covariances of certain stochastic processes and its application to unbiased linear prediction .
amer . math .
72 , 519550 ( 1952 ) lachal , a. : bridges of certain wiener integrals .
prediction properties , relation with polynomial interpolation and differential equations .
application to goodness - of - fit testing . in bolyai society mathematical studies x. limit theorems , balatonlelle ( hungary ) , 1999 , budapest , 151 ( 2002 ) nazarov , a. i. and nikitin , ya .
: exact @xmath466-small ball behavior of integrated gaussian processes and spectral asymptotics of boundary value problems .
probab . theory related fields 129 , no . 4 , 469494 ( 2004 ) | let @xmath0}$ ] be the linear brownian motion and @xmath1}$ ] be the @xmath2-fold integral of brownian motion , @xmath3 being a positive integer : @xmath4$.}\ ] ] in this paper we construct several bridges between times @xmath5 and @xmath6 of the process @xmath1}$ ] involving conditions on the successive derivatives of @xmath7 at times @xmath5 and @xmath6 . for this family of bridges
, we make a correspondance with certain boundary value problems related to the one - dimensional polyharmonic operator .
we also study the classical problem of prediction .
our results involve various hermite interpolation polynomials . |
quantum impurity problems in 1 + 1 dimensions have long been the subject of intense study due to numerous applications in fields such as quantum wires , the quantum hall effect and the kondo problem . in particular ,
much research has been invested into understanding the behavior of particle transfer through an impurity and of growing interest experimentally and theoretically are the statistics of the fluctuations in the transferred charge .
most efforts have been concentrated on the shot noise which is sensitive to the quantization of charge and can give useful information about the system.@xcite in recent years , it was realized that the full counting statistics , probability distribution @xmath2 of the transmitted charge @xmath3 in a given time window @xmath4 , contains more information.@xcite along with the shot noise , the full counting statistics also contains information about the higher order correlations which are of interest experimentally .
it has been suggested that the third moment may give a more reliable measure of the charge than the shot noise.@xcite recent experimental efforts suggest that the higher moments of the full counting statistics may be accessible.@xcite in cases where fermi liquid theory applies , the full counting statistics has been studied extensively ( see ref . [ ] for a review ) .
systems of strongly interacting electrons , though of importance in 1 + 1 dimensional quantum wires , the quantum hall effect and kondo problem have been less well studied .
using advanced methods such as the thermodynamic bethe ansatz , for the laughlin states exact results for the charge current through an impurity,@xcite as well as the noise,@xcite the heat current , @xcite and even attempts at the full counting statistics were reported.@xcite exploiting the power of boundary conformal field theory , the full counting statistics in double quantum dots was studied.@xcite a subject of recent interest is the behavior of quantum entanglement , and in particular of the entanglement entropy , in condensed matter systems ( for a recent review see ref . [ ] . ) the entanglement entropy has been shown to exhibit universal scaling behavior near quantum critical points.@xcite universal scaling behavior of the entanglement entropy is also present in topological phases of matter , such as fractional quantum hall fluids and their generalizations , as well as in the related topological quantum field theories.@xcite however , the entanglement entropy of a macroscopic quantum system is a highly non - local quantity which is difficult to measure .
it has remained a challenge to find an experimentally viable protocol to measure the entanglement entropy .
several recent results have suggested that the behavior of point contacts in quantum critical systems , and topological phases , may offer a way to measure the entanglement entropy .
fendley et al have shown@xcite that the change in the entanglement entropy of topological fqh fluids at a point of constriction is related ( in fact the same ) to the change of the affleck - ludwig entropy@xcite of the coupled edge states of the fqh fluid at the point contact .
however , the affleck - ludwig entropy itself is difficult to measure .
more recently , klich and levitov@xcite have shown that , at least for a system of free fermions , it may be possible to measure the growth ( in time ) of the entanglement entropy upon a quantum quench by monitoring the noise in the charge current through the contact .
specifically , they proposed that the second cumulant of the full counting statistics is related to the entanglement entropy .
they considered the following protocol : first the two subsystems described by non - interacting fermions lying on either side of an infinite strength impurity are completely decoupled . at time @xmath1 , the impurity is removed and the subsystems are allowed to exchange particles . finally at time @xmath5 , the impurity is reinstated and blocks the flow of particles .
they find that the shot noise generated is given by @xmath6 where @xmath7 is some short time cutoff .
such a protocol is a suggestive procedure for measuring the quantum entanglement between two halves of a system . in the initial state
the two subsystems lying on either side of the impurity are completely independent , but following the quench , quantum entanglement is dynamically built up by the exchange of particles with vanishing net flow . a measure of the entanglement between the two subsystems is provided by the entanglement entropy .
the entanglement entropy for such a scenario was recently calculated using conformal field theory ( cft ) and shown to increase logarithmically with time , with a universal coefficient proportional to the central charge @xmath8 characterizing the cft , @xcite @xmath9 where @xmath10 is a short distance cutoff . by comparing eqns .
[ eqn : levitov]-[eqn : cardy ] and noting that for non - interacting fermions , the central charge is @xmath11 , it is quite suggestive that the noise is a measurement of the entanglement .
however , it should be noted that while the methods of calabrese and cardy are general , the results of klich and levitov are specific to non - interacting electrons .
the degree of general validity ( _ e.g. _ for interacting systems ) of the klich - levitov protocol is presently an open question .
recently there has also been a surge of interest in quantum quenching , _
i.e. _ where the parameters that describe the dynamics of the system are changed over a short period of time either locally or globally . in cold atoms ,
it has been seen experimentally that quantum quenches may show interesting features.@xcite theoretically , quantum quenching has also attracted a good deal of attention in recent years.@xcite it was found recently that in a global quench , where the eigenstate state @xmath12 of a hamiltonian @xmath13 is evolved by a different hamiltonian @xmath14 , the correlation functions were found to display universal behavior characteristic of the hamiltonian @xmath14 if it was tuned near criticality.@xcite in this work , we study a _
local _ quantum quench where the point contact between two @xmath0 fractional quantum hall states is instantaneously opened ( see fig .
[ fig : quench ] ) .
we examine the noise and the full counting statistics and extend the results of klich and levitov [ ] to a system of interacting electrons with a dynamic impurity .
no external bias is applied . here
the bare strength of the impurity coupling undergoes a sudden change between two values , the first corresponding to fully reflecting boundary conditions and the second to fully transmitting boundary conditions .
the quantum point contact ( qpc ) generates an effective impurity in the luttinger liquid description of the edge , thus allowing complete control over the impurity strength by tuning the voltage applied to a side gate .
such a system is therefore a promising candidate for an experimental exploration of quantum quenches .
our main result is that the second moment of the full counting statistics , and therefore the noise , has a similar form to the entanglement entropy .
this result holds even for interacting ( luttinger ) systems .
however , we also find that this correspondence appears to be coincidental since , in general , in addition to the noted dependence of the central charge , the noise depends also on other universal quantities of the underlying conformal field theory of the system .
there is however a conceptual difference between static and dynamic entanglement .
static entanglement refers to the measure of the non - local correlations associated with the _ observation _ of a part of the system without disturbing it and is encoded in the von neumann entanglement entropy . on the other hand , when a physical system is changed , as in a quantum quench , the ensuing _ dynamical _ entanglement measures the non - local correlations that develop upon its time evolution in the quenched system .
( see , _ e.g. _ ref . [ ] ) . at the computational level ,
the main method underlying our results is the use of a boundary condition changing operator to generate the transition between the fully closed qpc and the fully open one .
this transition is related to a change in the sign of the odd - boson density operator going through the qpc , thus is marked by the appearance of an orbifold theory .
the method in its most general form will be described in section [ section : orbifolds ] . in sections [ section :
majorana ] and [ section : orbifolds ] we describe the method for @xmath15 and @xmath16 . for the particular case of @xmath15 , the result can be derived explicitly using a majorana fermion representation for the re - fermionized hamiltonian .
finally , in section [ section : noise ] we calculate the noise , and in [ section : fcs ] the full counting statistics . in section [ section : thermal ]
we sketch the calculation of the noise of the energy - momentum in the quantum ising chain , showing that in contrast to the entanglement entropy in a 1 + 1 dimensional critical system , the noise does not always grow logarithmically .
the details of the calculations for the laughlin @xmath15 ( bosonic ) state are presented in appendix [ appendix : nu - half ] .
the modular @xmath17-matrix and fusion rules for the orbifold cft ( used in section [ section : orbifolds ] ) are given in appendix [ appendix : orbifoldalgebra ] .
the relation between the schwinger - keldysh formalism and entanglement entropy is summarized in appendix [ appendix : keldyshentangle ] . .
( i ) fully reflecting for @xmath18 ( @xmath19 ) , ( ii ) fully transmitting for @xmath20 ( @xmath21).,scaledwidth=35.0% ]
the lagrangian describing the edge of the quantum hall effect for the laughlin states at filling factors @xmath0 can be written in the terms of left and right moving bosonic fields , @xmath22 , in the form @xmath23 where @xmath24 $ ] , and the velocity @xmath25 of the edge was set to one . for decoupled edges ,
the physical hilbert space of each chiral boson is generated by @xmath26 , the @xmath27 current and the vertex operator @xmath28 ( see ref . [ ] ) .
this vertex operator generates the charged excited states of the theory which are fundamentally generated by adding or subtracting an electron from either the left or right edge.@xcite requiring that the charged excitations generated by @xmath29 are bosons ( electrons ) implies that they satisfy the ( anti- ) commutation relation sets @xmath30 an even ( odd ) integer .
the requirement that the charge of the operator is precisely @xmath31 forces the relation @xmath32 .
representations of the theory are local with respect to the @xmath27 current and charged excitation @xmath33 which is identified with the electron ( @xmath30 odd ) or boson ( @xmath30 even ) .
the qpc introduces backscattering at @xmath34 , with the most relevant term being the quasi - particle tunneling term @xmath35\ ] ] where @xmath36 is the tunneling strength .
the charge densities at the left and right edges are respectively @xmath37 note that @xmath38 , so in particular the electron operator is well defined provided @xmath39 is an integer ; hence @xmath40 carries the meaning of the number of quasi - particles on the edge . in units of the electron charge @xmath31 , the charge carried by the edge is @xmath41 .
we now perform the following transformation into the odd and even basis @xmath42 the even boson @xmath43 decouples from the impurity , and we ignore it in the rest of the paper .
note that @xmath44 are both left - moving .
the odd and even charges are related to the original @xmath45-charges through @xmath46 , @xmath47 .
the experimental setting we wish to consider is one where the two edges of the quantum hall liquid are separated initially .
no bias is applied and the temperature is taken to be zero . at some time , the two halves of the system are ( suddenly ) connected and we wish to consider the statistics of the charge transferred from one side to the other upon this change , _
i.e. _ a quantum quench . the system is then evolved with the hamiltonian with a hilbert space described by states with transmitting boundary conditions at the quantum point contact .
let @xmath2 be the probability that charge @xmath3 is transmitted through the point contact in time @xmath4 .
then one defines the generating function @xmath48 which encodes all moments ( or cumulants ) @xmath49 of the probability distribution @xmath2 , @xmath50 in particular , the second moment @xmath51 is related to the current fluctuations .
the generating function can be written as @xmath52 where the operators are ordered on the schwinger - keldysh contour.@xcite this can also be written as a trace over a complete set of states .
@xmath53 where @xmath54 is the _ initial _ density matrix , and @xmath55 is the time evolution operator for time @xmath4 after the point contact is opened . in eq .
it should be understood that the trace is taken over the initial states .
if the scattering time at the quantum point contact is short compared with the entire time evolution , then it was shown that @xmath56 can be written as@xcite @xmath57 where @xmath58 is the initial distribution of states and @xmath17 is the reflection - transmission matrix at the quantum point contact . in ref .
[ ] , it was found that for _ non - interacting electrons _ , the generating function is given by @xmath59 where @xmath60 .
in this work , we generalize this protocol to the @xmath0 laughlin states .
we note that while the full counting statistics has been found in many other strongly correlated systems with a static impurity via the thermodynamic bethe ansatz,@xcite the application of this method to finding time and space dependent correlation functions is in general difficult and remains largely not understood .
we take a different approach in tackling this problem .
this relies on the realization that one should consider an extended hilbert space by introducing a boundary changing operator to the edge theory and that physical charged excitations in the initial hilbert space are those conserving total charge . in the next section ,
we examine the noise in @xmath15 which can be refermionized and solved explicitly .
we note an interesting structure which we then extend to the other laughlin states at @xmath0 .
in doing so , we calculate the noise and full counting statistics for the laughlin states .
we start our discussion with an exactly solvable case , the bosonic quantum hall effect at filling factor @xmath15 .
our strategy is as follows : re - fermionizing , we first recast the problem in terms of a quadratic fermionic action .
then , by writing the fermionic field in terms of two real majorana fermion fields , we find that only one of the majorana fermions interacts with the impurity .
the two fixed points associated with the fully transmitting and the fully reflecting impurity , translate into anti - periodic and periodic boundary conditions imposed on this majorana fermion .
this leads us to identify the operator that takes the system between the two fixed points as the @xmath61 operator of the chiral ising model .
armed with this knowledge , we then proceed to use this formalism to calculate the noise for the case of a quantum point contact which is suddenly opened at time @xmath62 . the odd boson theory can be re - fermionized in terms of a fermion field @xmath63 ( see appendix [ appendix : nu - half ] ) leading to the hamiltonian @xmath64\end{aligned}\ ] ] the crucial step is that the hamiltonian can be written in a majorana representation , @xmath65 where @xmath66 and @xmath67 satisfy @xmath68 . in this representation ,
only @xmath67 interacts with the impurity , as is clear by inspecting the equations of motion , @xmath69 integrating the equations leads to the following boundary condition across the impurity , @xmath70 , @xmath71 this boundary condition is continuously interpolating between the fully reflecting and fully transmitting boundary conditions for the odd boson @xmath72 similarly , at @xmath73 , we have the boundary condition , @xmath74 the problem is thus mapped to a single chiral majorana fermion on a circle of circumference @xmath75 . for the reflecting case we have anti - periodic ( ap ) boundary conditions , while for the transmitting case , we have periodic ( p ) boundary conditions .
the free majorana field theory contains altogether three primary fields : the identity field * 1 * , the majorana fermion field @xmath76 , and a @xmath61 operator ( known as the ising twist field ) creating a branch cut on the majorana fermion @xmath76 . as before , we shall label the two copies of fields present here by an index @xmath77 : * 1*@xmath78 , @xmath79 , and @xmath80 . to go from periodic to anti - periodic boundary conditions on @xmath67
, we use the @xmath81 operator to create a branch cut on the fermion : we imagine that the edge majorana fermion encloses a bulk " in which @xmath81 operators can be introduced , appropriately changing the boundary conditions on the majorana fermion .
we thus identify the operator taking the system between the two fixed points associated with the qpc as the @xmath81 operator in the @xmath67 majorana fermion field theory . in particular , to go from anti - periodic to periodic boundary condition , introduce two @xmath81 operators in the bulk " surrounded by the @xmath67 edge ( see fig .
[ fig : boundary - changing ] ) .
one of them then approaches the edge at @xmath34 and a bulk - edge coupling is introduced .
this coupling is mediated by virtual processes involving the tunneling of topologically trivial modes between the bulk and the edge @xcite . finally , when @xmath19 , the edge circumvents the operator completely , and switches to ap boundary conditions . in the sudden approximation
, we can imagine two spatially - separated @xmath81 operators introduced simultaneously : one inside the edge and the other outside .
operators are drawn from the vacuum .
tunneling ( of strength @xmath36 ) is introduced between one of the @xmath81 operators and the edge .
finally , in the limit @xmath19 , the edge circumvents the @xmath81 operator.,scaledwidth=80.0% ] we assume that the qpc is initially closed , then at time @xmath62 it is fully opened , and proceed to calculate the noise produced in that situation . the current operator through the junction can be defined as the rate of transfer of charge between the two edges , measured by the odd boson @xmath82\end{aligned}\ ] ] where @xmath83 .
using the equations of motion , eqns .
, we can write the current in terms of the majorana fermions as@xmath84 in the limits , @xmath85 and @xmath86 , we get @xmath87 where @xmath62 time of the quench , and @xmath88 is what we call the backscattered current . in the following
, we shall define the charge transmitted through the junction during a time window starting at @xmath89 and ending at some arbitrary positive time @xmath4 , @xmath90 since there is no bias , the net transferred charge is zero , @xmath91 .
however , there will be noise due to the opening of the qpc .
the noise is given by @xmath92 besides its role in establishing a non - zero current operator , the operator @xmath81 also changes the boundary conditions on @xmath67 .
however , these boundary conditions ( _ i.e. _ the effects of the operator @xmath81 ) will affect @xmath67 only after a long time - scale set by @xmath93 . in the limit @xmath94 we can safely neglect the effects of these boundary condition changing operators . upon the introduction of a uv cutoff @xmath95 , the short - time switching scale needed to regulate the integral , the second cumulant @xmath51 can be obtained . @xmath96
the expression can be easily integrated to get @xmath97 in the limit of large @xmath4 this reduces to @xmath98
the essential point is that the theory of the two edge states coupled by the local qpc must have a local boundary changing operator @xmath81 that takes the system from one fixed point to the other . in particular
, the operator @xmath81 maps the ground state of the system with a closed qpc to an energy eigenstate of the same system but with the new boundary conditions , an open qpc .
the ensuing growth of the entanglement entropy is due to the evolution of correlations of the excitations in this state.@xcite the previous discussion can also be understood in terms of the boson .
this picture lends itself to generalization to the other laughlin states which are also described by free chiral bosonic theories . here
, we show that from the bosonic picture , one can deduce that @xmath61 operator must be introduced for one of the majorana fermions to change the boundary condition from anti - periodic to periodic .
we then argue that a similar boundary condition changing operator can be included in the bosonic theory .
one can understand the relationship between boson and fermion boundary conditions and the nature of the twist operator by examining the partition functions .
specializing to @xmath15 , the lagrangian becomes @xmath99 the boundary condition at the impurity site changes from @xmath100 perfectly reflecting ( neumann boundary conditions ) to @xmath101 perfectly transmitting ( dirichlet boundary conditions ) .
the neumann boundary condition simply fixes the field @xmath102 to be continuous across the boundary while the dirichlet boundary condition allows the field to be discontinuous .
imposing neumann boundary conditions at the ends of the system for all times , the mode expansion for the odd boson can be found . for neumann ( n ) and dirichlet ( d ) boundary conditions at the origin respectively , @xmath103 the partition function in each sector can then be computed for each set of boundary conditions and one finds @xmath104 where @xmath105 are partition functions for a single @xmath106 holomorphic fermion with boundary conditions @xmath107 in time and @xmath26 in space , a anti - periodic and p periodic . to go from @xmath100 to @xmath101
, we see that exactly one majorana fermion changes anti - periodic to periodic boundary conditions in time .
now , the partition function @xmath108 is also related to the partition function @xmath109 by an insertion of the fermion counting operator @xmath110 in the time direction . by a conformal transformation
, this operator is related to the spin operator of the ising model inserted in the space direction for one of the majorana fermions that compose the theory.@xcite this confirms the results of the previous section .
while this identifies the boundary changing operator as the spin operator in the ising model , this is a special feature of @xmath15 state . at other filling fractions
the boson can no longer be fermionized , but what is independent of the filling is the boson mode expansion and the partition functions @xmath111 . to generalize the calculation of the noise beyond @xmath15 , one should identify the boundary condition changing operator in the bosonic theory that takes neumann boundary conditions to dirichlet boundary conditions . a priori , it may seem strange that a @xmath11 free bosonic theory must include a boundary changing operator , but the necessity of this operator can be seen at the level of the path integral . the path integral should be time ordered on the schwinger - keldysh contour , but by an analytic continuation , one can regard the schwinger - keldysh contour as the time surface for a conformal field theory@xcite ( see fig.[fig : keldysh ] ) . in going from the density matrix to the path integral formulation
, one must consider an _ extended _ hilbert space of the hamiltonian @xmath112 which is spanned by states with all possible boundary conditions .
the boundary changing operators appear as unitary operators that map whole sectors of the extended hilbert space with particular boundary conditions to sectors with different boundary conditions.@xcite details of this analytic continuation can also be found in appendix [ appendix : keldyshentangle ] and in refs .
[ ] . with changing boundary conditions at @xmath113 and @xmath114 .
details in appendix [ appendix : keldyshentangle ] and [ ] .,scaledwidth=47.0% ] folding the system on the @xmath115-world sheet , the local quenching problem can be regarded as a boundary conformal field theory with a changing boundary condition , neumann to dirichlet at the origin .
such a boundary conformal field theory was studied in the context of the x - ray edge singularity and quantum wires.@xcite the neumann boundary condition implies that the odd boson is continuous at @xmath34 while the dirichlet boundary condition implies that the odd boson is discontinuous at @xmath34 . as discussed by affleck and ludwig,@xcite this
can be written as : @xmath116 this has an interpretation of a branch cut running on the @xmath115-world sheet across which the odd boson is discontinuous . in ref .
[ ] it was noted that this branch cut corresponds to an insertion of a dimension @xmath117 operator at the quench times , @xmath118 . observing that the boundary conditions are independent of the radius and following arguments by ref .
[ ] , it is suggestive that there is dimension @xmath117 twist operator in the odd boson theory ( independent of the boson radius ) which changes the boundary condition from neumann to dirichlet
. given such generality , one wonders if there is a larger algebraic structure describing the local quench which includes the boundary condition changing operator in its operator algebra . in the sequel ,
we show that this structure is described by an orbifold conformal field theory .
much of the previous discussion was limited to the @xmath15 case and was special in that it could be formulated in terms of majorana fermions .
we argued previously that to generalize the previous results to the other laughlin states , one needs to understand the theory in terms of free chiral bosons and to work with an extended hilbert space where a dimension @xmath117 boundary condition changing operator is included .
such a hilbert space is described by an orbifold conformal field theory , but before delving into orbifold theories , we need to describe the initial hilbert space of the system which we will extend in the next section .
the total charge of the system is conserved in the quenching process so that physical charged excited states are those preserving the total charge .
initially , one knows that in the limit of two decoupled edge states , the charged excitations that conserve total charge are processes where an electron tunnels from one edge to the other , @xmath119 in the even and odd basis , one can identify the vertex operator @xmath120 with @xmath30 as generating these excited states . by charge conservation , the even sector does not have charged excited states , and its hilbert space is generated by the @xmath27 current . however , charged excited states are allowed in the odd sector by electron tunneling , so the physical hilbert space is generated by the @xmath27 current as well as the vertex operator @xmath121 .
representations of this hilbert space can be constructed with the requirement that they be local with respect to the generators @xmath122 and @xmath123 .
this construction is well known and extends the @xmath27 kac - moody ( km ) algebra to conformal field theories described by the @xmath124-series .
the primaries in the theory can be found in any standard text.@xcite @xmath125 where @xmath126 and @xmath127 .
the twist operator @xmath61 exists explicitly in the fermion theory as spin operators , but by considering a @xmath128 orbifold of the free gaussian theory , a twist operator can be included in the boson theory . on the @xmath115-world sheet of the odd boson , one has that the @xmath129 as @xmath130 approaches the branch cut running between the two quench times .
imposing @xmath131 on the @xmath115-sheet has consequences for the target space.@xcite for a compact boson , @xmath132 $ ] , this means that the target space is identified under the action of the discrete group @xmath128 .
the theory must include an operator which takes @xmath133 . as argued
before , the theory describing the odd boson is a conformal field theory in the @xmath124-series . to this theory , we wish to allow for an anti - periodic boson and need to identify the target space under the group @xmath128 .
one needs to consider the orbifold @xmath134 .
the partition function for orbifold theories under the group @xmath135 was found by ginsparg.@xcite decomposing this partition function into @xmath124 characters and observing the transformation properties under the modular @xmath136 and modular @xmath137 transformation allows one to deduce the primaries , the modular @xmath17-matrix of the theory and fusion rules.@xcite the primaries of the theory are , @xmath138 the @xmath128 transformation @xmath139 acts on these representations as @xmath140 \rightarrow [ \phi_{2m - k}]$ ] and so there are two fixed points @xmath141 $ ] and @xmath142 $ ] .
the operators @xmath143 correspond to a sector invariant under @xmath133 ( i.e. @xmath144 ) and a sector that is broken by the transformation @xmath133 ( i.e. @xmath145 ) .
physically , one can understand them as the tunneling and current operators for the odd boson respectively .
the operator @xmath146 are invariant under the transformation @xmath147 .
@xcite @xmath148 correspond to the trivial and non - trivial representations of @xmath128 . for @xmath149
even , it turns out that the @xmath150 fields are self conjugate while for @xmath149 odd , @xmath151 and @xmath152 are each others conjugates.@xcite using the fusion rules ( see appendix [ appendix : orbifoldalgebra ] ) and decomposing the partition function @xmath153 in terms of the characters of the @xmath134 theory , it can be seen that the @xmath148 operators act as boundary changing operators . for @xmath149 even , fusing with @xmath151 gives @xmath154 while for @xmath149 odd one has @xmath155.@xcite for @xmath15 , the orbifold theory is @xmath156 which is isomorphic to the tensor product of two @xmath106 virasoro algebras ( _ i.e. _ two decoupled ising models).@xcite this was exactly the behavior observed before . the partition function for the odd boson one observed could be decomposed into a product of @xmath106 partition functions .
in addition to this , the primaries of the theory are @xmath157 we see that the algebra naturally includes the twist fields @xmath148 and @xmath158 and that there is an operator corresponding to the quasiparticle tunneling operator , @xmath159 the operator @xmath160 has a fusion rule consistent with its interpretation as @xmath161.@xcite in addition , we see that the dimension @xmath162 field we can regard as the dimension of the two spin field composite . hence , one of the @xmath61 s originally at the center can be taken to be in the trivial representation @xmath151 of @xmath163 non - local with respect to the fermion field @xmath164 and the other in the non - trivial representation and non - local with respect to the fermion @xmath165 ( see fig .
[ fig : boundary - changing ] ) .
when the interaction strength is switched , precisely one spin field interacts with either @xmath166 giving a twisted boundary condition .
this behavior was also seen in the odd boson partition function and explicit calculation in section [ section : majorana ] . for @xmath167 ,
the theory is described by the orbifold theory @xmath168 which corresponds to the @xmath169 parafermion theory.@xcite this can also be seen by comparing the fusion algebra and operator content of the two theories : @xmath170 the fusion rules here are somewhat different than for @xmath149 even .
the operator algebra of , @xmath26 and @xmath171 now has a @xmath169 symmetry , and it is given by @xmath172 the fusion algebra for the @xmath146 vertex operators is unchanged.@xcite the fusion rules and modular @xmath17-matrices for other filling fractions are given in general in appendix [ appendix : orbifoldalgebra ] .
one can then go about computing the noise in the backscattering current in the language of bosons . in the original left - right basis
, the backscattered current is given by @xmath173 @xmath174 are charge densities measured at a position after the point contact while @xmath175 are charged densities measured at a position before the point contact . in the rotated even - odd basis ,
the backscattering current can be written as , @xmath176 where @xmath177 denotes after and before the point contact at @xmath34 . to connect with the boundary conformal field theory formalism
, one can fold the system and regard @xmath178 as a right moving wave @xmath179 .
the backscattering current is then @xmath180 where @xmath181 now . this form has the advantage of explicitly involving a primary field of the theory .
intuitively , the backscattered charge is simply the difference between the ( incoming ) left moving charge density and the ( outgoing ) right moving charge density .
the noise is , @xmath182 where @xmath183 denotes a correlation function with boundary condition @xmath10 at @xmath34 ( neumann / fixed or dirichlet / free ) and fixed ( neumann ) at @xmath73 . here
@xmath184 and @xmath185 when @xmath186 is on the top contour and @xmath187 on the bottom contour and vice versa in the second contribution .
@xmath188 is a short distance cutoff , @xmath189 . in the presence of a boundary condition @xmath10
these correlation functions have been computed using methods of conformal invariance and their relationship to the modular @xmath17-matrix of the bulk theory were found in refs .
[ ] , @xmath190 where @xmath191 and @xmath192 and @xmath193 is a primary field with scaling dimension @xmath194 . by knowing which primary field corresponds to boundary condition @xmath10 one can use cardy s formula to compute the correlation function in the presence of the boundary
. for @xmath15 , we showed that the theory can be thought of as @xmath195 which is isomorphic to a theory of two decoupled ising models .
the boundary state corresponding to free boundary conditions ( dirichlet ) is associated with the primary @xmath61 while the fixed boundary conditions ( neumann ) is associated with the identity or @xmath196 primary operators . using the @xmath17-matrix for the ising model , its easy to see that the correlation functions with boundary condition @xmath10 have the prefactors , @xmath197 for other filling fractions , the identification of boundary states and boundary conditions is more complicated but we find that a similar structure is present .
since the characters for the @xmath198 theory are known , the partition functions @xmath199 and @xmath200 can be decomposed into a sum of characters from which one may identify the boundary states.@xcite note that here we worked with the t - dual ( @xmath201 ) of the theory in refs .
[ ] so that @xmath202 . for @xmath149
even , neumann and dirichlet boundary conditions , it was found that the orbifold boundary states are , @xmath203 for @xmath149 odd , the analogous states can be written down using the appropriate @xmath17-matrix ( see appendix [ appendix : orbifoldalgebra ] ) . from the modular @xmath17-matrix , one then can identify the neumann boundary condition with the highest weight state @xmath146 and the dirichlet ( free ) boundary condition with the @xmath61 field .
note that the boundary states above are _ not _ the usual boundary states but are @xmath163 invariant linear combinations.@xcite however , the fact remains that they transform in the @xmath204 $ ] and @xmath205 $ ] representations respectively .
for @xmath0 one finds that it is generally true that @xmath206 for arbitrary @xmath207 , one finds that in the limit @xmath208 , the noise from the local quench is given by @xmath209 in agreement with the calculation for @xmath15 so that the second cumulant has a similar form as ( [ eqn:2ndcumulant ] ) . a non - trivial check of these results can also be done by considering the boundary condition on the odd currents . in terms of the odd currents ,
the reflecting and transmitting boundary condition were given by @xmath210 and @xmath211 respectively.@xcite imposing either of these boundary conditions , one sees that the left - left and right - right correlation functions are unaffected but the left - right correlation function changes by a minus sign .
@xmath212 so that when the point contact is closed , the noise measured is zero while when it is opened , the noise is non - zero .
this is in agreement with the results using the operator algebra of the @xmath198 orbifold .
note that this correlation function is special in that the boundary condition involves the primary fields @xmath213 and @xmath214 that appear in the desired correlation function but eqn.([eqn : boundarybulkcorr ] ) is more general than this .
while the second cumulant can easily be shown to grow logarithmically in time , our result shows that the coefficient of this logarithm _ depends on the modular @xmath17-matrix elements of the theory and not just to its central charge _ , so that even for @xmath11 the second cumulant does _ not _ fully give the entanglement entropy .
the full counting statistics ( fcs ) is the generating function for the moments of the charge transfered between two reservoirs through a constriction . in ref .
[ ] , the fcs was calculated for the process of opening and closing the qpc , with the result that the distribution function for the transmitted charge is purely gaussian .
here we shall extend our methods developed in the previous sections to find the fcs in the interacting case , and conclude that the non - interacting result is in fact robust to the presence of interactions .
it is plausible that even in the interacting case , there exists a simple relation between the fcs and the entanglement entropy . to avoid boundary effects ,
it is convenient for our purposes to consider a corbino ( annular ) geometry , with a single qpc inducing backscattering between the edges . in this new geometry
, we should actually reverse the protocol appearing in previous sections : first the qpc is open , then is quickly closed , and remains in this state during a time period @xmath4 , at the end of which it is opened again .
going to the odd / even basis for the two edge bosons , the charge carried by the odd boson , @xmath215 , is controlled by the singularity induced by the branch cut for the closed qpc ( assumed to be located at @xmath34 ) .
the full counting statistics is then given by the following correlation function , @xmath216 which , using the observation above , can be reduced to the calculation of vertex operators at the vicinity of the qpc .
this leads to the main result , @xmath217\right\}\end{aligned}\ ] ] which is the typical generating function for a gaussian distribution .
consequently , only the second cumulant ( _ i.e. _ the noise ) is non - zero .
interestingly , except for the explicit dependence on the filling factor @xmath207 , the full counting statistics has the same form as the non - interacting electron case .
writing the boson in the even and odd basis and folding the system , this problem can be mapped onto the boundary sine - gordon model . in the basis of kink and anti - kink solitons of charge @xmath3 and @xmath218 respectively , distribution functions can be found via the thermodynamic bethe ansatz for the entire range of @xmath36.@xcite the problem is then similar to the non - interacting electron problem only the particles are collective modes rather than the original electrons .
however , in this picture , the full counting statistics is difficult to compute .
the reflection and transmission coefficients of the time dependent scattering matrix@xcite are difficult to write in the kink and anti - kink basis .
although in previous sections we showed that the noise generated by a quench can also evolve logarithmically with time , we will show now that this is not always true .
this is in contrast to the static entanglement entropy which depends logarithmically on the size of the observed region@xcite and the dynamic entanglement entropy which evolves logarithmically with time.@xcite consider a one - dimensional quantum ising model that can be split in two by changing with time the strength of just one link .
this effectively implies a change in boundary conditions at that link .
we use the majorana fermion description of the critical quantum ising model , @xmath219 where @xmath220 ( @xmath221 ) is a left - going ( right - going ) majorana fermion , and @xmath222 was flipped . at the center of the ising chain ( @xmath34 )
we modify the link strength , which translates in the effective fermion description to a coupling to the local energy density operator @xmath223 at the same point @xmath224 this marginal perturbation leads to the following relation between the fields on the two sides of the impurity @xmath225 with @xmath226 when @xmath36 is fine tuned to the point @xmath227 , @xmath228 becomes completely off - diagonal @xmath229 .
when @xmath21 , @xmath228 approaches the identity matrix , @xmath230 . the quantum ising model and its effective field theory , the free majorana fermion , has a ( discrete ) global @xmath163 symmetry .
contrary to the luttinger - type models which have a globally conserved @xmath27 charge and a locally conserved ( dimension one ) charge current , in the case of the quantum ising model the only global conservation law is the energy - momentum and its locally conserved ( dimension two ) energy - momentum current .
in contrast with the luttinger model where the quantum quench does not change the total charge , in the ising model the quantum quench changes the hamiltonian and hence the total energy .
although the energy - momentum tensor of the majorana fermion field theory ( and of the ising model ) remains locally conserved everywhere _ except _ at the location of the quantum impurity , these physical differences lead to a distinct asymptotic time dependence of the noise in the energy - momentum current shown below .
the energy density along the edge is defined by @xmath231 the rate of back - scattering of energy by the modified link is @xmath232 using the relations above , for the case that the link is fully transmitting ( @xmath233 ) , the heat current is zero as the contributions on the two sides of the link cancel . in the limit of a fully reflecting link
, we get @xmath234 , and the two contributions add up .
since the scaling dimension of the energy current is @xmath235 we get the following expression for the thermal noise @xmath236 : @xmath237 for large @xmath4 , the thermal noise approaches a non - universal , cutoff dependent , value @xmath238
in the previous sections we identified an extended theory of the bosonic luttinger description of the quantum hall edge states in the presence of an impurity . the extended theory , being an @xmath239 orbifold theory ( for @xmath0 ) , explicitly contains an operator inducing the transition between the zero - backscattering and fully - backscattering fixed points of the impurity .
in particular , for @xmath15 the extended theory decouples into two copies of @xmath163 parafermions , whereas for @xmath167 it coincides with @xmath169 parafermions instead .
the presence of the primary field in the algebra which induces the transition between the two fixed points at the impurity demonstrates that in the sudden approximation equilibrium is maintained through each process of opening or closing of the qpc , and that no transient effects are expected . using these methods we calculated the fcs for the interacting luttinger liquid for the process of a sudden opening ( at @xmath240 ) then closing ( at @xmath5 ) of the qpc .
only the second cumulant of the transferred charge is nonzero , and its logarithmic dependence on @xmath241 suggests that a relation between the entanglement entropy and the fcs should exist even in the interacting case . in that sense , the noise `` measures '' the entanglement entropy .
the only difference from the non - interacting case is the explicit appearance of the filling factor in the noise . in a more general
setting , however , the relation between the fcs and the entanglement entropy does not seems to be structural .
first , it requires the presence of a conserved current with dimension 1 ( essential for a logarithmic behavior of the correlation function ) , and of an associated and strictly conserved global charge .
in contrast , the energy - momentum noise through a weak link in the one - dimensional ising model does not reproduce , or even has the same form , as the entanglement entropy .
we discuss this case briefly in section [ section : thermal ] .
also , for any quantum hall state which contains neutral edge modes , the generated entanglement entropy is proportional to the total central charge , @xmath8 , but in the absence of an impurity the noise involves only the pure current present in the charge carrying @xmath11 theory alone .
this can be amended by adding the central charge explicitly into the relation between the noise and the entanglement entropy , but this argument seems to be ignoring the mechanisms that create entanglement in the neutral theory .
moreover , even for the simple luttinger case the central charge does not appear in the noise ( which depends only on the fusion rules and the conformal dimensions of the cft ) while it certainly appears explicitly in the entanglement entropy . indeed ,
our results show that the noise generated is a measurement of _ dynamical _ entanglement and not of the _ static _ entanglement that is associated with the observation of a subsystem in an eigenstate of the full system .
@xcite interestingly , we also find that the time dependent correlation functions have a universal scaling behavior as in the case of the global quench.@xcite however , the universal behavior is characteristic of an orbifold theory of the original system .
it may be of interest to study this experimentally where corrections to scaling are sure to occur .
the flow away from criticality would presumably involve the more exotic fusion rules . in summary , in this paper
we have computed the noise of the tunneling current generated by a quantum quench of a point contact in a fractional quantum hall fluid .
this was done by explicit solution for the bosonic laughlin state at @xmath15 and by conformal field theory methods for a general laughlin state .
we have also computed the growth of the entanglement entropy due to the quantum quench .
we found that even though the time dependence of the noise has the _ same form _ as the entanglement entropy the latter has a more intricate dependence on the properties of the conformal field theory and the hilbert space of the edge state .
this result suggests that the klich - levitov protocol may not generally supply a procedure to measure the entanglement entropy .
the question of measuring the entanglement entropy remains open .
we thank ian affleck , john cardy , claudio chamon , paul fendley , israel klich , leonid levitov , and chetan nayak for enlightening discussions and suggestions .
this work was supported in part by the national science foundation grant dmr 0758462 at the university of illinois ( ef and bh ) .
eg was supported by the institute for condensed matter theory of the university of illinois .
in this appendix we give some details of the re - fermionization technique used in the exact solution of the @xmath15 ( bosonic ) quantum hall effect . for @xmath15 , the lagrangian for the odd boson in the presence of a qpc at @xmath34
is given by ( @xmath242 ) @xmath243 where the odd density @xmath244/\sqrt{2}$ ] .
we refermionize by defining @xcite @xmath245 so that @xmath246 holds .
the full hamiltonian can now be written as @xmath247\end{aligned}\ ] ] note that @xmath248 , and @xmath249 . here
@xmath250 is a klein factor which appears in a careful handling of the zero modes of the boson field .
assuming for simplicity @xmath251 , the equations of motion can be found and are given by @xmath252 the equations of motion can be solved by expanding the fields in modes as ( see e.g. ) @xmath253 the modes satisfy the following relations , derived by integrating the equations of motions around @xmath34 , @xmath254\end{aligned}\ ] ] defining @xmath255 , we get @xmath256 so that when @xmath86 ( fully transmitting qpc ) , @xmath257 , while when @xmath258 ( fully reflecting qpc ) @xmath259 .
with the convention that the matrix takes the vector labelled by @xmath261 to the same vector with @xmath262 , @xmath263 and @xmath264 , the modular @xmath17-matrix for @xmath149 even can be written as : @xmath265 while for @xmath149 odd , the modular @xmath17-matrix takes a slightly different form .
@xmath266 for @xmath149 even , the elements @xmath267 form a @xmath268 subalgebra .
one has @xmath269 the twist operators have the fusion rules @xmath270 for @xmath149 odd , the elements @xmath271 form a @xmath169 subalgebra .
@xmath272 and for the twist fields , @xmath273 the fusion rules for the vertex operators @xmath146 are the same for @xmath149 even or odd .
@xmath274 the fusion rules for the @xmath275 fields can be found easily by applying @xmath276 .
the non - equilibrium problem of locally quenching the system at the point contact is best dealt with in terms of the density matrix formalism .
initially , the system has a density matrix @xmath277 where the states @xmath278 are eigenstates of the hamiltonian with neumann boundary conditions imposed at the point contact and ends of the system . at time @xmath62 , the point contact is opened and the boundary condition at the boundary is changed to dirichlet . where the regulator @xmath196 has been included to adiabatically cut - off the high momentum modes . here , @xmath14 is the hamiltonian with transmitting boundary condition .
the first term describes forward evolution from the initial state to the final state @xmath280 while the second describes the time reversed evolution back to the initial state . following refs .
[ ] , the schwinger - keldysh time contour can be thought of as the time contour for a conformal field theory . on the forward branch ,
one continues @xmath281 , @xmath282 .
now , @xmath283 , wick rotating , this can be thought of as propagating the initial state for time @xmath7 to the final state .
@xmath284 one can think of the density matrix in terms of a path integral , but to do so , a complete set of states needs to be introduced .
here we work with the _ extended _ hilbert space which is spanned by states with all possible boundary conditions . inserting a complete sets of states @xmath285 at each time slice @xmath286 and also a complete set of conjugate states @xmath287 .
now , @xmath289 so that this is the euclidean path integral with the boundary condition that at @xmath290 , the field takes value @xmath291 and @xmath292 at @xmath293
. a similar set of manipulations can be performed for the backward branch as well . at the point @xmath62 , they differ on a set of measure zero so that the path integral can be thought of as over the entire surface parameterized by @xmath115 with changing boundary conditions at @xmath113 and @xmath114 . at the end of the calculation
, one should continue back to real time @xmath294 and take the limit @xmath295.@xcite here , before the quench @xmath18 , we apply reflecting boundary conditions and after the quench , we have transmitting boundary conditions . in the sequel
, we denote this surface , with changing boundary conditions in time at @xmath34 as @xmath296 . to compute the entanglement entropy , one uses the replica trick to compute , @xmath297 .
the reduced density matrix can be visualized as stitching together the cylinder on the @xmath298 side while leaving the @xmath299 side open .
multiple copies of @xmath300 are stitched cyclically on the cut .
@xcite in ref .
[ ] , it was shown using general transformation properties of the stress energy tensor that this branch cut amounts to a twist operator @xmath301 insertion on the riemann surface with conformal dimension , @xmath302 .
the stitched surface amounts to computing the correlation function of the twist operator .
@xmath303 here @xmath304 denotes an average taken over the surface with changing boundary conditions . to compute this correlation function
, we use the zhukowski mapping which avoids the algebraic structure discussed in the text .
the advantage is that including an @xmath58-twist field as part of an extended hilbert space is difficult and is left as an open question here . using the mapping ,
@xmath305 the surface @xmath296 is mapped to the right half plane with a single boundary condition , @xmath306 . the correlation function on @xmath296
is related to the correlation function on @xmath306 @xmath307 differentiating with respect to @xmath58 at @xmath308 yields the entanglement entropy @xmath309 where @xmath310 is a non - universal constant and @xmath10 a lattice cutoff.@xcite in the limit @xmath311 the entanglement entropy becomes @xmath312 so that the amount of entanglement grows logarithmically in time after the quench at @xmath62 and is proportional to the central charge . | we examine the growth of entanglement under a quantum quench at point contacts of simple fractional quantum hall fluids and its relation with the measurement of local observables . recently
klich and levitov proposed that the noise generated from a local quantum quench provides a measure of the entanglement entropy .
their methods were specific to non - interacting electrons and the generalization to interacting systems was left as an open question . in this work ,
we generalize their result to the laughlin states .
we investigate the noise generated in the current along the edge of a fractional quantum hall state at filling factors @xmath0 , when a quantum point contact , initially closed , is fully opened at some initial time @xmath1 .
we find that local quenching in these systems gives time dependent correlation functions that have universal behavior on sufficiently long time and length scales .
we calculate the noise and full counting statistics for @xmath0 and find that in general , the entanglement entropy and noise generated are unrelated quantities .
we also discuss a generalization of this problem to the critical quantum ising spin chain . |
the discovery of the first tev @xmath0-ray source ( the crab nebula ) in 1989 by the whipple collaboration @xcite initiated a fast development of ground - based @xmath0-ray astronomy .
the imaging air cherenkov technique has been successfully used since the first source was discovered .
imaging air cherenkov telescope ( iact ) measure the cherenkov light from extensive air showers ( eas ) .
the cherenkov photons that are reflected by the telescope mirror are recorded by a matrix of photomultipliers ( the so - called telescope camera ) mounted in the focal plane of iact .
the shower image that is formed in the camera is a two dimensional angular distribution of the cherenkov light .
the number of the registered hadron induced events ( the so - called background ) is several orders of magnitude larger than the number of the registered @xmath0-rays events from a source . in 1985
hillas proposed a method to select @xmath0 rays out of a hadron dominated event sample @xcite .
this method is based on parametrization of the recorded shower image .
the arrival direction of the primary particle is determined by the direction of the image main axis for the data taken with a single iact and by the intersection of the major axes of both recorded images for a stereo observation .
main axes of the primary @xmath0-ray images from point - like source are directed to the source position on the camera , while the hadronic background is isotropically distributed .
the @xmath0/hadron separation methods which are used now are more sophisticated ( such as @xcite ) , but most of them are still based upon the original hillas parameters .
currently operating observatories , such as h.e.s.s .
@xcite , magic @xcite and veritas @xcite use arrays of iacts with large mirror areas .
the potential sources of the @xmath0 rays are observed in stereo mode , in order to improve the sensitivity of telescopes . the cherenkov telescope array ( cta ) collaboration @xcite plans to build arrays of telescopes with different sizes to measure the fluxes of @xmath0-ray sources in a large energy range from a few tens of gev to hundreds of tev .
the detection of a low energy showers is possible by using very large telescopes , with a parabolic shape that avoid broadening of the time profile of the cherenkov signal .
it has been shown , based on real data ( see e.g. @xcite ) , that the @xmath0/hadron separation in iacts becomes much more difficult below 100 gev .
this is motivated by a few effects . at first
, larger fluctuations of the cherenkov light density at ground are expected in a low energy region @xcite which results in larger fluctuations of the image parameters .
second , the geomagnetic field influences the image parameters @xcite .
thirdly , iacts can be triggered by a primary electron or positron initiated shower @xcite .
there are no physical reasons for the differences between the images of a pure electromagnetic subcascade induced by a primary @xmath0 ray and by an electron
. therefore this kind of background can not be rejected by using the hillas parameters describing the image shape .
fourthly , it has been suggested in @xcite and shown in @xcite that a specific type of hardly reducible background occurs when observing the low energy region : telescopes can be triggered by light produced by electrons and positrons from only one or two electromagnetic subcascades , which are products of a single @xmath2 decay in the hadron initiated shower .
shower images formed by cherenkov photons from a single electromagnetic subcascade can be called false @xmath0-ray events because they have a very similar shape to the primary @xmath0-ray images .
these false @xmath0-ray images can be slightly narrower because they start deeper in the atmosphere than real @xmath0 rays .
therefore a narrower angular distribution of secondary @xmath3 and @xmath4 is expected ( see e.g. @xcite ) .
the efficiency of the @xmath0/hadron separation method , based on the parameters describing the image shape , deteriorate at low energy due to the occurrence of the false @xmath0-ray events .
however , the orientation of the major image axis of the false @xmath0-ray events is randomly distributed .
therefore parameters , that determine the shower direction , are still effective variables for the primary @xmath0-rays selection .
the results presented in this paper are based on the monte carlo ( mc ) simulations for a system of four iacts .
we investigate the impact of the occurrence of false @xmath0-ray events on the efficiency of the @xmath0/hadron separation .
the results were obtained for : i ) two hadron interaction models ( gheisha and fluka ) ; ii ) two different altitudes of the observatory ( @xmath5 and @xmath6 a.s.l . ) ; iii ) three azimuth angles of @xmath7 , @xmath8 and @xmath9 ; iv ) three different telescope areas ( @xmath10 , @xmath11 and @xmath12 ) . in the following
we present the mc study and show the fraction of the false @xmath0-ray candidates in the proton initiated showers for different multiplicities of triggered telescopes .
the fraction of single electromagnetic subcascade in the total proton background is estimated for energy range larger than used in the simulations .
this fraction is calculated for different trigger thresholds .
we show how the contribution of single electromagnetic subcascade in the triggered proton events depends on the average of size ( this is the sum of all signals from pixels which belongs the image ) .
the scaled width , scaled length @xcite and the height of the shower maximum are applied for the selection of @xmath0 rays out of the hadron induced showers .
we calculate the quality factor ( qf ) in order to demonstrate the @xmath0-ray selection efficiency .
the strong anti - correlation between qf and the contribution of the false @xmath0 rays was found for all simulated sets of the parameters of the iact system .
therefore , the occurrence of the false @xmath0-ray events is an important reason of the reduction of the @xmath0/hadron separation efficiency in the range of small average sizes .
finally in the conclusions , we compare the results which were obtained from different interaction models , altitudes of the observatory and telescope sizes .
and @xmath6 , respectively.,width=264 ] the investigated system of iacts contains four telescopes placed in the corners of a diamond having a side length of @xmath13 and diagonals of @xmath13 and @xmath14 and with the altitude of the observation equal to @xmath5 a.s.l .
( see filled circles in figure 1 ) .
both the altitude of the observatory and the distance between the two closest telescopes has been chosen as they are for the actual magic telescopes configuration @xcite .
the strength and direction of the geomagnetic field is fixed to that at magic site ( la palma ) .
the side length of the diamond and the distance between the two closest telescopes are reduced to @xmath15 for the simulated altitude equal to @xmath6 a.s.l .
( see open circles in fig . 1 ) .
each simulated telescope is very similar to the magic telescope @xcite .
it has a parabolic shape with a focal length of @xmath16 .
the mirror dish has a diameter of @xmath16 .
the total reflector area covers approximately @xmath10 .
the camera has hexagonal shape that covers in total almost @xmath17 .
it consists of 1141 photomultipliers ( pmt ) with the diameter of @xmath18 .
631 pixels ( in the inner part of the camera ) are used for the trigger .
the trigger requires a time coincidence and neighbour logic .
all four telescopes from the system are exactly the same .
the simulations of telescopes smaller than @xmath10 were obtained by reducing only the mirror area of the described above iact . in reality ,
smaller telescopes are designed with smaller focal length as the davies - cotton design @xcite ( e.g. @xmath19 iacts in h.e.s.s . ) to avoid the so - called coma effect , which is observed for a telescope with the parabolic shape . in cases of the davies - cotton design
one may expect higher trigger rate and better hillas parameter determinations for both hadron and @xmath0-ray events .
however , the assumption of smaller telescope area adopted in this paper leads to reducing of the coma effect .
therefore , it is good approximation to demonstrate the impact of telescope size on single electromagnetic subcascade effects .
the shower development in the atmosphere was simulated using the corsika code @xcite .
two corsika versions 6.023 and 6.99 have been used . in the version 6.023
the gheisha @xcite and venus @xcite interaction models have been applied for the low ( i.e. for particles with primary momentum below 80 gev / c ) and high energy ranges . in the version 6.99 fluka @xcite and qgsjet - ii @xcite
have been used as the low and high energy interaction models , respectively .
the standard corsika code has been adapted in order to study the single subcascade events in the proton initiated shower .
the information about each subcascade that has been produced in the eas was kept in modified code .
additionally , the type of the charged particle , which was responsible for the creation of cherenkov photons , was also saved in the output of the program .
showers initiated by primary protons were simulated in the energy range between 30 gev and 1 tev with a differential spectral index of -2.75 ( which is the index of the primary proton spectrum @xcite ) .
the impact parameter was distributed randomly within a circle of 1.2 km radius ( measured in the plane perpendicular to the shower axis ) around the geometrical center of the telescopes system .
the proton induced showers were simulated within a cone with a half - opening angle of @xmath20 at a zenith angle of @xmath21 .
the impact parameter was randomly distributed within a circle of @xmath22 radius for the simulations of showers induced by the primary @xmath0 rays that have an energy range between 10 gev and 1 tev .
the differential spectral index was chosen to be equal to -2.6 ( which corresponds to the index of the crab spectrum @xcite ) .
the direction of the @xmath0-ray showers was fixed to a zenith angle equal to @xmath21 that is also the direction of the axes of all telescopes .
seven sets of simulations were performed : ( i ) one using fluka and qgsjet - ii as hadron interaction models ( code version from 2011 ) ; ( ii , iii and iv ) three for different azimuth angles ( @xmath7 - showers directed to the north , @xmath8 , @xmath9 ) using the gheisha and venus as the hadron interaction models ; ( v ) one for the altitude of the observatory equal to @xmath6 a.s.l . ; ( vi and vii ) two for smaller telescope sizes ( @xmath11 and @xmath12 ) .
different azimuth angles were simulated in order to show how the geomagnetic field influences the occurrence of false @xmath0-ray events .
it has been shown in @xcite that the effects of geomagnetic field are scaled by the component of b that is perpendicular to the shower axis ( @xmath23 ) .
the azimuth angle of @xmath7 , @xmath8 and @xmath9 correspond to @xmath23 equal to 20 , 25 and 30 mt , respectively .
the distances between the closest telescopes in the plane perpendicular to the shower axis are similar for all simulated azimuth angles and they are in the range between @xmath24 and @xmath13 .
overview of parameters and the numbers of simulated events in all mc sets are presented in the table 1 .
note , that showers initiated by primary @xmath0 rays are simulated in the same way by the two corsika versions considered .
@*6llllll mc&corsika&azimuth&altitude&telescope&number&number + set&version&angle&of observ.&size&of protons&of @xmath0 rays + & & ( deg.)&(km a.s.l . ) & ( m@xmath25 ) & ( in 10@xmath26)&(in 10@xmath26 ) + i&*6.990*&0&2.2 & 230 & 22&0 + ii&*6.023*&*0*&*2.2 * & * 230 * & 22&1 + iii&6.023&*53*&2.2 & 230 & 22&1 + iv&6.023&*82*&2.2 & 230 & 22&1 + v&6.023&0&*4.0 * & 230 & 22&1 + vi&6.023&0&2.2 & * 160 * & 22&1 + vii&6.023&0&2.2 & * 100 * & 22&1 + the detector simulation consists of two parts : the first part includes the full geometry of the mirrors and their imperfections
. additionally in this program , rayleigh and mie scattering of cherenkov light in the atmosphere were taken into account according to the sokolsky formula @xcite .
the camera properties , such as additional reflections in winston cones , photocathode quantum efficiency @xcite and its fluctuation , and the photoelectron collection efficiency were considered in the second part of detector simulations .
the level of the night sky background ( nsb ) has been chosen to that measured at la palma @xcite .
the nsb was included in the simulation for the trigger conditions only . in reality photoelectrons ( p.e . ) produced by light of the nsb are randomly distributed over all pixels of the camera . due to that
, all pixels which contain only the signals from nsb should be subtracted before calculating the parameters of the shower image .
the commonly used cleaning procedure neglects pixels with signals smaller than chosen values for the core and boundary pixels , which belong to the image .
the efficiency of the separation between real and false @xmath0-ray events may depend on both nsb and cleaning levels .
the results obtained without the nsb light and cleaning are more general because they do not depend on observation conditions and the particular cleaning procedure .
a single telescope was triggered by a shower if the output signals in three next neighbouring pixels ( 3 nn ) exceed a certain threshold within a time window of 3 ns . 3 nn trigger logic was applied in all simulations presented in this paper . in reality , different trigger multiplicity might be used for observations with a large single iact .
the simulated pulse of a single photoelectron has a full width at half maximum equal to 3 ns .
three trigger thresholds were studied in this paper : 3 , 4 , 5 photoelectrons , which correspond to the signal of 3 , 4 and 5 p.e .
arriving exactly at the same time , respectively .
finally the possible coincidences between triggers of 2 , 3 or 4 telescopes were also checked .
in the results presented below we call false @xmath0-ray events ( or images ) that fulfilled conditions : i ) the light contribution from the hadronic and muonic part of the shower is less than 10@xmath1 of the image size , ii ) the rest of the image size is the light contribution from a single electromagnetic subcascade in the proton induced shower , regardless of the number of triggered telescopes . according to this definition , a single photoelectron from other electromagnetic subcascades can not give the contribution to recorded images .
therefore results presented below show the smallest possible impact of single electromagnetic events on observations with iacts .
in particular , a total fraction of false @xmath0 rays estimated in this paper should be treated as a lower limit .
the distributions of the number of electromagnetic subcascades giving a contribution to all registered images are presented in figure 2 .
all distributions are normalized to 1 .
the height of the second histogram bin corresponds to the fraction of the false @xmath0-ray candidates only , because conditions i ) and ii ) are not required .
figure 2a shows plots obtained from mc set ii ( using the gheisha model ) for the trigger threshold equal to 3 p.e .. the fraction of the false @xmath0-ray candidates decreases with the multiplicity of the triggered iacts as expected .
we have checked that the presented distributions are similar for all simulated azimuth angles : @xmath7 , @xmath8 and @xmath9 ( mc sets ii , iii and iv ) .
the comparison between the results of the gheisha ( black ) and fluka models ( grey ) are demonstrated in fig .
2b for the multiplicity of triggered telescopes equal to 1 ( long dashed ) and 2 ( solid histograms ) and the trigger threshold of 3 p.e .. the fluka model gives higher fraction of the false @xmath0-ray candidates than the gheisha model only for the required number of triggered telescopes equal to 1 .
the results of both interaction models are very similar when more than one iact is triggered .
the higher fraction of the false @xmath0-ray candidates was obtained for the simulated altitude equal to @xmath6 a.s.l .
( grey histograms in fig .
2c ) than for @xmath5 a.s.l . ( black histograms ) , regardless of the required number of triggered telescopes .
the proton induced showers contain slightly smaller fraction of the false @xmath0-ray candidates for a smaller mirror area ( e.g. @xmath12 in fig .
2d plotted as grey ) than for @xmath10 telescopes ( shown as black histograms in this figure ) .
significantly higher fraction of events without any electromagnetic contribution ( a height of the first histogram bin in fig .
2 ) is expected for smaller than for larger telescopes when less than 4 iacts triggered . and @xmath12 .
long dashed , solid , dashed and dotted histograms correspond to the required multiplicities of triggered telescopes from 1 to 4 , respectively .
all presented distributions were normalized to 1 .
, width=529 ] we have checked that the expected trigger rate of proton showers is higher for fluka in comparison to gheisha model when only at least one or two telescopes fulfilled the trigger conditions . for larger
required multiplicity of triggered iacts , the results of both investigated interaction models do not differ significantly . at the altitude of @xmath6 a.s.l .
more proton induced events triggered the iact system than at @xmath5 , as expected .
the trigger rate is lower for smaller mirror area than for @xmath10 iact , as expected .
we have found that the energy distributions of events that fulfilled both i ) and ii ) conditions are much steeper than that of all triggered showers for all simulation sets .
most of the false @xmath0-ray events are the proton showers with low energy . for less than four triggered iacts , more than 94@xmath1 of the false @xmath0-ray images with the size parameter larger than 20 p.e .
are the proton induced showers with primary energy below 200 gev , regardless on the simulation set .
this fraction is slightly below 90@xmath1 only in case of all triggered telescopes , where obtained trigger energy thresholds for primary proton ( the position of the maximum of the energy distribution ) are the highest .
note , that the discussed fraction almost does not depend on the chosen interaction model ( fluka or gheisha ) , magnetic field , observation altitude and mirror area .
figure 3 shows how the ratio of the number of false @xmath0-ray events to the number of all triggered proton showers depends on the energy of the primary proton for the trigger threshold of 3 p.e .. this ratio was calculated separately in each histogram bin .
the results obtained from mc set ii are presented in fig .
3a for required multiplicities of triggered telescopes from 1 to 4 .
the fraction of the false @xmath0 ray in the energy bin decreases with the multiplicity of triggered iacts .
this fraction also diminishes with energy for all simulation sets ( see comparisons between : interaction models in fig . 3b , different altitudes of the observatory in fig .
3c and different telescope areas in fig .
the first bin of presented histograms corresponds to the energy range between 30 gev and 40 gev . at these energies one should expect that even up to 45@xmath1 of protonic background is single subcascade events at the altitude of @xmath6 a.s.l . for at least two triggered telescopes ( see grey solid histogram in fig .
3c ) . in the energy range between 250 gev and 400 gev , no more than 2@xmath1
of all triggered protonic background are false @xmath0-ray events for the required number of triggered telescopes larger than 1 , regardless on the simulation set ( see the sixth bin of histograms presented in fig . 3 ) .
-ray events to the number of all triggered proton shower versus primary energy for the trigger threshold equal to 3 p.e .. * a ) * the gheisha model for all investigated multiplicities of triggered iacts . *
b ) * comparison between interactions models .
* c ) * comparison between two observation altitudes : 2.2 and 4 km a.s.l .. * d ) * comparison between two telescope sizes : @xmath10 and @xmath11 .
long dashed , solid , dashed and dotted histograms correspond to the required multiplicities of triggered telescopes from 1 to 4 , respectively .
the ratio was calculated in each histogram bin separately.,width=604 ] to estimate a fraction of false @xmath0-ray events in the total proton background , one may evaluate the expected number of the triggered and false @xmath0-ray events for energies above 1 tev .
a simple power - law fits of the energy distribution tail ( for images with the size parameter larger than 20 p.e . )
was used as an extrapolation function .
energies above 500 gev and 200 gev were considered as tails of the energy distribution for all triggered and false @xmath0-ray events .
we have checked that the fraction of false @xmath0-ray events does not depend on the distance between the two triggered telescopes .
similar values were obtained for the shortest and the longest distance between iacts .
figure 4 shows how the fraction of the false @xmath0-ray images in the proton induced showers decreases with the trigger threshold .
four panels in this figure present comparisons between the results obtained from different mc sets .
the fractions of the false @xmath0-ray events for azimuth angles @xmath7 ( black ) and @xmath9 ( grey lines ) are plotted in fig .
the absolute differences are smaller than 1@xmath1 .
therefore we conclude that estimated fractions do not depend on the simulated azimuth angle . in the investigated range of @xmath23 ( between 20 mt and 30 mt )
the geomagnetic field does not influence any of the two numbers : all triggered proton and false @xmath0-ray events .
this conclusion is consistent with the results shown in @xcite , where authors claimed that geomagnetic field has much smaller impact on the proton than on @xmath0-ray initiated showers even for the large size telescope ( lst ) which is proposed in the cta project @xcite .
a comparison of the interaction models is presented in figure 4b . in case of the observations with a single telescope ( long dashed lines ) ,
the contribution of false @xmath0-ray images in all triggered events decreases from approximately 11@xmath1 to 7@xmath1 when the trigger threshold increases from 3 p.e . to 5 p.e . for the fluka model ( grey ) and from 9@xmath1 to 5@xmath1 for the gheisha model ( black lines ) . for
at least 2 triggered iact , the estimated fraction is still higher for fluka than for gheisha , however absolute differences are smaller than 1@xmath1 . for larger multiplicities the differences between used interaction models are negligible .
it has been shown in @xcite that the probability that the majority of the primary energy is deposited in the electromagnetic part of the shower is lower for the gheisha than for fluka model .
due to that , the estimated fraction of the false @xmath0 rays in the protonic background is lower for gheisha than for fluka model for at least one triggered telescope .
however , if more iacts are triggered then similar results were obtained for both interaction models .
this may be explained by the fact that some cherenkov photons from other electromagnetic subcascades can contribute to registered images .
the probability of this effect is higher for fluka which produces larger electromagnetic part of the shower , while , the probability that more telescopes fulfilled trigger conditions decreases in a similar way for both interaction models . the estimated fraction of false @xmath0-ray events for the altitude equal to @xmath6 ( grey lines in fig .
4c ) are higher than for @xmath5 ( black lines in the same figure ) in the investigated ranges of the trigger threshold and telescope multiplicity .
significantly more false @xmath0-ray events are observed at higher altitudes , regardless on the multiplicity of triggered telescope .
therefore , the expected fraction of the false @xmath0-ray events in the proton initiated showers is strongly influenced by the altitude of the observatory .
figure 4d shows a comparison between the contributions of false @xmath0-ray events estimated for two simulated mirror areas : @xmath10 ( black ) and @xmath12 ( grey lines ) . at fixed trigger threshold
, this fraction significantly increases with the mirror area , although the estimated fraction of the false @xmath0-ray candidates only slightly increases with the telescope size ( see fig .
however for smaller mirror areas , large fraction of those candidates contains photoelectrons created by cherenkov photons from hadronic and muonic parts of the shower . in cases of telescope multiplicities
lower than 3 , the fraction of the false @xmath0-ray events obtained for @xmath10 at trigger threshold of 5 p.e .
are very close to results for @xmath11 telescope at trigger threshold of 4 p.e . and for @xmath12 telescope at trigger threshold of 3 p.e .. the simulations of the iact with @xmath12 mirror show that the expected fraction of the false @xmath0-ray events is negligible for the multiplicity of triggered telescope larger than 2 ( see grey dashed and dotted lines in fig .
4d ) . in those cases energy thresholds obtained for proton showers
are above 250 gev . in this energy range
the probability of triggering the iact system by a single electromagnetic subcascade is very small even for at least one triggered telescope .
-ray events in the total protonic background as a function of the trigger threshold for : * a ) * two different azimuth angles ( from mc simulation sets ii and iv ) ; * b ) * gheisha / fluka model ( from mc sets ii and i ) ; * c ) * two different altitudes of the observatory : @xmath5 a.s.l . and
@xmath6 a.s.l .
( from mc sets ii and v ) ; * d ) * two different telescope areas ( from mc sets ii and vii ) . in all panels ,
long dashed , solid , dashed and dotted lines correspond to the multiplicity of the triggered iacts equal to 1 , 2 , 3 and 4 , respectively .
mc sets are defined in table 1.,width=604 ] the size parameter was calculated for each triggered telescope .
an average size ( @xmath27size@xmath28 ) of the event was obtained from parameters measured by each triggered iact .
we have checked that more than 90@xmath1 of false @xmath0-ray events have @xmath27size@xmath28 below 300 p.e . for the multiplicity of triggered telescopes lower than 3 .
figure 5 shows how the ratio of the number of the false @xmath0-ray events to the number of all triggered proton showers depends on the @xmath27size@xmath28 .
this ratio was calculated separately in each histogram bin for the trigger threshold of 3 p.e .. the results of mc set ii are shown in fig .
the fraction of the false @xmath0-ray events in @xmath27size@xmath28 bin does not depend on the multiplicity of triggered iacts ( see long dashed , solid and dashed histograms in this figure ) . in the lowest presented @xmath27size@xmath28 bin
approximately 30@xmath1 - 40@xmath1 of the proton initiated showers are the false @xmath0-ray events .
this fraction decreases with @xmath27size@xmath28 .
the results obtained from other mc sets are also very similar for different multiplicities of triggered telescopes .
therefore the comparison between models ( see fig .
5b ) , altitudes ( see fig .
5c ) and telescope sizes ( see fig .
5d ) is shown only for at least two triggered iacts .
the histograms obtained with gheisha ( black ) and fluka ( grey ) do not differ significantly .
we have checked that in the case of at least one triggered telescope only , the calculated ratio is slightly higher for fluka than for gheisha model in whole range of the average size .
it is seen in fig . 5c
that the contribution of the false @xmath0-ray events in the proton background is higher for the simulated altitude of @xmath6 ( grey ) than for @xmath5 ( black histogram ) .
the contribution of false @xmath0-ray events in each chosen @xmath27size@xmath28 bin is lower in case of a smaller telescope area ( in fig .
5d grey histogram corresponds to @xmath12 telescopes ) .
-ray events to the number of all triggered proton shower versus @xmath27size@xmath28 for the trigger threshold equal to 3 p.e .. * a ) * the gheisha model for different multiplicities of triggered iacts .
* b ) * comparison between interactions models for at least 2 triggered iacts .
* c ) * comparison between two observation altitudes : 2.2 and 4 km a.s.l .. * d ) * comparison between two telescope sizes : @xmath10 and @xmath12 . long dashed , solid and dashed histograms correspond to the required multiplicities of triggered telescopes from 1 to 3 , respectively .
the ratio was calculated in each histogram bin separately.,width=604 ] two hillas parameters relevant for this study are length and width .
they are defined as the standard deviations of the image profiles along the long and short axis of the image , respectively @xcite .
both parameters can be used for the selection of the @xmath0-ray events from the hadron initiated showers .
the more efficient @xmath0/hadron methods use the so - called scaled width ( width@xmath29 ) and scaled length ( length@xmath29 ) ( for definition see @xcite ) .
the scaling factors are calculated from simulated @xmath0-ray images only .
they are applied to the calculations of the width@xmath29 and length@xmath29 for the images of @xmath0-ray and proton initiated showers . as a result ,
the width@xmath29 and length@xmath29 distributions of the @xmath0-ray initiated showers are gaussian distributions with mean equal to 0 and dispersion equal to 1 . for each shower which triggered the system , the mean width@xmath29 ( @xmath27width@xmath29@xmath28 ) and mean length@xmath29 ( @xmath27length@xmath29@xmath28 ) were calculated from scaled parameters obtained from each recorded image .
the distributions of the @xmath27width@xmath29@xmath28 and @xmath27length@xmath29@xmath28 parameters are presented in figure 6a and 6b for at least two triggered telescopes at the trigger threshold equal to 3 p.e .. the results of mc set ii are plotted as an example .
the histograms of all triggered events for primary @xmath0 ray ( dashed black ) and proton ( solid grey ) are normalized to 1 .
appropriate factors were used to scale the histograms of the false @xmath0-ray events .
the distribution of false @xmath0-ray images ( solid black ) is shifted towards lower @xmath27width@xmath29@xmath28 in comparison to the real @xmath0-ray events , because the detected false @xmath0-ray events are subcascades which begin deeper in the atmosphere .
therefore , the observed parts of electromagnetic cascade are younger and the angular distributions of electrons and positrons are narrower than in the showers initiated by primary @xmath0 rays ( see e.g. @xcite )
. it can be seen in figure 6 , that one can expect very low efficiency of the primary @xmath0-rays selection from false @xmath0-ray events if the width@xmath29 and length@xmath29 parameters only are used in the separation method . the height of the shower maximum ( h@xmath30 ) can be used to separate @xmath0 rays from proton events @xcite .
the distributions of the h@xmath30 parameter ( in km a.s.l . ) obtained from mc set ii are presented in figure 6c for at least two triggered iacts at the trigger threshold of 3 p.e .. the height of the shower maximum is a good parameter to use in the @xmath0/hadron separation .
however , there is no clear separation between real and false @xmath0-ray results .
therefore , the real and false @xmath0-ray events can not be effectively distinguished by using the height of the shower maximum .
the occurrence of the false @xmath0-ray events affects the efficiency of the @xmath0/hadron separation . in order to estimate this effect quantitatively ,
whole sample of triggered events was split into five intervals of @xmath27size@xmath28 .
next , the @xmath0/hadron separation method was applied in each @xmath27size@xmath28 bin .
primary @xmath0-ray candidates are selected as events that fulfilled conditions : i ) have all recorded images with width@xmath29 and length@xmath29 in the range between -1.5 and 1.5 , ii ) have the height of the shower maximum above 9.5 km a.s.l .. the estimated energy of primary @xmath0 rays ( e@xmath31 ) that correspond to each @xmath27size@xmath28 interval is given in the table 2 for different simulation sets .
the presented energy was estimated as the peak position of the energy distribution when at least 2 telescopes triggered at trigger threshold of 3 p.e .. note , that for fixed @xmath27size@xmath28 bin different values of e@xmath31 were obtained from different mc sets .
thus the efficiency of the @xmath0-ray selection presented in this paper was calculated in energy ranges which depend on the simulations set .
@*6llllll @xmath27size@xmath28 & & ( 20,50)&(50,100 ) & ( 100,200 ) & ( 200,300 ) & @xmath28 300 + mirr
. area & altitude & e@xmath31 & e@xmath31 & e@xmath31 & e@xmath31 & e@xmath31 + ( m@xmath25 ) & a.s.l .
( km ) & ( gev ) & ( gev ) & ( gev ) & ( gev ) & ( gev ) + 230 & 2.2 & 15 . & 25 .
& 50 . & 90 . & 135 .
+ 160 & 2.2 & 20 . & 35 . & 65 . & 120 . & 195 .
+ 100 & 2.2 & 30 . & 50 . & 100 .
+ 230 & 4.0 & 11 . & 18 . & 30 . & 60 . & 90 .
+ the quality factor ( qf ) was used to demonstrate the efficiency of the @xmath0-ray selection from the hadronic background @xcite .
this factor is related to the significance of the measurement ( the ratio of the @xmath0-ray signal and the dispersion of the poisson - like hadronic background ) .
the quality factor can be defined as , @xmath32 where @xmath33,@xmath34 are the total numbers of triggered events for chosen @xmath27size@xmath28 bin and @xmath35 , @xmath36 are the numbers of events surviving the @xmath0/hadron separation cuts .
the qf , which were calculated in this paper , take into account only the differences between parameters describing image shapes and differences in h@xmath30 of the showers initiated by primary @xmath0 rays and protons .
other parameters that are used in the primary @xmath0-rays selection are not connected with the occurrence of the false @xmath0-ray events .
the quality factor versus the @xmath27size@xmath28 parameter is shown in figure 7a for at least two triggered iact at the triggered threshold of 4 p.e .. the same trigger conditions were chosen to prepare plots in all panels of the figure 7 .
the efficiency of the @xmath0/hadron separation increases with the @xmath27size@xmath28 for all simulation sets as expected .
the value of qf decreases with the mirror area - compare the results for @xmath12 ( dashed ) , @xmath11 ( dotted ) and @xmath10 telescopes ( solid line ) .
the quality factors obtained using the fluka model ( grey solid ) are very similar to values calculated using the gheisha model ( black solid ) .
the efficiency of the primary @xmath0-ray selection is worse for the altitude equal to @xmath6 ( dash - dotted ) than for @xmath5 ( solid line ) .
we have check that the qf obtained for different azimuth angles ( i.e. different @xmath23 ) do not differ in contrast to the results presented in @xcite .
this disagreement can be explained by the fact , that in the @xmath0/hadron separation method we have not use parameters describing the image orientation that were applied by authors in @xcite .
the quality factor versus the contribution of false @xmath0-ray events in protonic background is presented in figure 7b .
the contribution of false @xmath0-ray events before @xmath0/hadron separation was calculated in each @xmath27size@xmath28 bin separately .
the quality factors decreases with the fraction of the false @xmath0-ray events in a similar way for all simulation sets if the contribution of the false @xmath0-ray events is larger than 0 . in order to present how the quality factor is reduced due to occurrence of the background caused by the false @xmath0-ray events ,
the ratio between qf and the maximal value of qf ( qf@xmath30 ) is shown as a function of the contribution of the false @xmath0-ray events ( see figure 7c ) .
it can be seen , that the results of all mc sets contain the points with negligible contribution of false @xmath0-ray events ( points with coordinates close to ( 0,1 ) ) , except the altitude of the observatory equal to @xmath6 a.s.l .. therefore for this altitude , qf@xmath30 is underestimated because it was calculated from the sample of showers with the contribution of the false @xmath0-ray events approximately equal to 2@xmath1 .
the most rapid reduction of a quality factor with the contribution of the false @xmath0-ray events is expected for iact system with the smallest telescopes ( diamonds ) . in this case
the calculated qf for @xmath27size@xmath28 larger than 300 p.e .
( i.e. e@xmath31 equal to 300 gev ) is approximately 5 times higher than for @xmath27size@xmath28 between 20 and 50 p.e .
( i.e. e@xmath31 equal to 30 gev ) .
qf parameter is reduced by a similar factor in case of 160 m@xmath25 telescopes ( stars ) and by a factor of 4 for gheisha / fluka model ( filled / open circles ) with the corresponding reduction of e@xmath31 ( see numbers in the table 2 ) .
the slowest decline of qf / qf@xmath30 was obtained for the iact system located at the altitude equal to @xmath6 a.s.l .
( crosses ) where the e@xmath31 is lower than 100 gev for all @xmath27size@xmath28 bins . however , as it was mentioned before , qf@xmath30 is underestimated for this simulation set . finally , we calculated the correlation coefficient between qf and the contribution of false @xmath0-ray events in protonic background before @xmath0-ray selection .
correlation coefficient changes significantly with increase of the multiplicity of triggered telescopes from 1 to 4 . at the trigger threshold of 4 p.e .
it varies from -0.95 to -0.5 and from -0.97 to -0.55 for the gheisha and fluka model , respectively . for the altitude of the observatory equal to @xmath6 ,
the correlation is stronger as its coefficient changes from to -0.86 . in case of 160 m@xmath25 iacts ,
calculated coefficient changed from -0.94 to -0.71 with decreasing the multiplicity of triggered telescopes from 1 to 3 .
the weakest correlation was obtained for the smallest investigated telescopes .
the correlation coefficients are equal to -0.92 and only -0.6 for at least one and 2 triggered telescopes , respectively . in cases of higher multiplicity of triggered iacts ,
the correlation coefficient was not calculated because the total fraction of the false @xmath0-ray events is negligible ( see fig .
the occurrence of false @xmath0-ray events plays an important role in the observations using an imaging air cherenkov telescopes system .
the study brought two important conclusions .
first , this kind of background is mainly caused by proton induced showers with the primary energy below 200 gev .
second , most of the false @xmath0-ray events have @xmath27size@xmath28 smaller 300 p.e .. both of those features do not depend on : the interaction model , altitude of the observatory and size of the telescope mirror .
those conclusions are consistent with the results obtained for a smaller trigger area of the camera @xcite .
the quality factor depends much stronger on the multiplicity of triggered telescopes than on the trigger threshold because the energy threshold of the iact system is more sensitive on the required number of triggered telescopes than on the trigger threshold .
we have checked , that the effect of the false @xmath0-ray events does not depend on the geomagnetic field for @xmath23 in the range between 20 and 30 mt .
the simulations presented in this paper allowed to compare the results obtained with two models of the hadron interactions .
we have shown , that for the multiplicity of triggered iacts larger than 1 , there are not significant differences between results obtained using gheisha and fluka model . in case of at least one triggered iacts , both the estimated total trigger rate and the expected fraction of false @xmath0-ray events are larger for fluka than for gheisha model . in this case the contribution of the single electromagnetic subcascade events in each @xmath27size@xmath28 intervals is also slightly higher for the fluka model .
comparison of the results for altitudes of @xmath5 and @xmath6 a.s.l .
leads to the following conclusions .
the system of iacts located at higher altitude can be triggered by protons at lower energies . due to that , the fraction of false @xmath0-ray events is significantly larger for altitude equal to @xmath6 than for @xmath5 , regardless on the multiplicity of triggered telescopes and trigger threshold .
the contribution of single subcascade events in chosen interval of @xmath27size@xmath28 is also larger for iact system located at higher altitude . as a consequence ,
the efficiency of the primary @xmath0-ray selection is lower for higher altitude of the observatory independently on the number of triggered telescopes .
the system of iacts with smaller mirror areas can be triggered by proton showers at higher energies than in the case of a system with larger telescopes . as a result
the fraction of single subcascade events also decreases significantly when the mirror area changes from @xmath10 to @xmath12 .
however , for the multiplicity of triggered iacts lower than 3 the fraction obtained for @xmath10 at trigger threshold 5 p.e .
is comparable with results for @xmath11 and @xmath12 at trigger threshold of 4 p.e . and 3 p.e .
, respectively .
the fraction of hardly reducible background for the observations with at least three or at least four triggered telescopes are not simply shifted in the trigger threshold level .
the estimated fraction of the false @xmath0-ray events is negligible ( lower than 1@xmath1 ) for the multiplicity of triggered iacts larger than 2 and the mirror area equal to @xmath12 .
telescopes with such mirror areas are proposed to be applied as the medium size telescopes ( mst ) in the cta project .
the data taken by at least three msts should contain negligible fraction of the background caused by single electromagnetic subcascade events .
we have shown that the efficiency of the @xmath0/hadron separation is strongly anti - correlated with the contribution of the false @xmath0-ray events in the proton induced showers . additionally , the occurrence of the false @xmath0-ray events has less influence on the data that are taken by at least 3 or 4 triggered telescopes because in this case , the system is triggered mostly by higher energy protons .
the results presented in this paper were obtained using a basic reconstruction of the shower . a new , more sophisticated reconstruction algorithm may lead to better @xmath0/hadron separation .
however , the occurrence of the false @xmath0-ray events is one of the main causes of the deterioration of both the @xmath0/hadron separation efficiency and as a result sensitivity of a system of iacts in the low energy range .
this work was supported by polish grant from narodowe centrum nauki 2011/01/m / st9/01891 | the sensitivity of imaging air cherenkov telescopes ( iacts ) worsens significantly at low energies because the @xmath0/hadron separation becomes much more complex . in this paper
we study the impact of the single electromagnetic subcascade events on the efficiency of the @xmath0/hadron separation for a system of four iacts using monte carlo simulations .
the studies are done for : two different altitudes of the observatory , three different telescope sizes and two hadron interaction models ( gheisha and fluka ) .
more than 90@xmath1 of the single electromagnetic proton - induced subcascade events are showers with primary energy below 200 gev , regardless on the trigger threshold .
the estimated efficiency of the @xmath0/hadron separation using the fluka model is similar to results obtained using the gheisha model .
nevertheless , for at least one triggered telescope only , a higher fraction of single electromagnetic subcascade events was obtained from the fluka model .
finally , the calculated quality factors are anti - correlated with the contributions of the false @xmath0-ray events in the proton initiated showers .
therefore , the occurrence of single electromagnetic subcascade events is one of the main reasons of the worsening of the primary @xmath0-ray selection efficiency at low energies . _
keywords _ : @xmath0-rays : general methods : observational instrumentation : detectors telescopes |
the inner few degrees of the galactic center show a large concentration of molecular , atomic hydrogen and dust clouds ( pierce - price et al . 2000 ; lang et al . 2010 ; molinari et al . 2011 ) .
the molecular gas toward the galactic center is considered to reside in the so - called central molecular zone ( cmz ) and consists of a mixture of diffuse and dense components ( morris and serabyn 1996 ; martin et al . 2004 ; sawada et al . 2004
; oka et al . 2005 ; yusef - zadeh et al .
radio continuum emission from this region is also extended and is produced by a mixture of thermal and nonthermal processes ( nord et al .
2004 ; yusef - zadeh et al . 2004 ; law et al .
2008 ) . given the confusing region of the inner galaxy due to large number of foreground and background sources along the line of sight as well as the complex motion of the gas in the galactic center , it is difficult to use the kinematic distance method to identify molecular clouds associated with hii regions and supernova remnants .
we describe a new technique to identify neutral clouds that show a deficiency in the distribution of radio continuum emission .
these clouds are embedded in a bath of radiation or cosmic ray particles produced by thermal or nonthermal sources , respectively .
the strong radiation field in the environment of cloud complexes with high column densities , such as infrared dark clouds ( irdcs ) , allow us to identify their dark counterparts in radio continuum images at cm and mm wavelengths .
the origin of radio dark clouds ( rdcs ) is unlike the x - ray shadowing and irdcs which are caused by strong absorption of background light by dense clouds ( egan et al .
1998 ; andersen et al .
rdcs are also unlike optically thick hii regions seen in absorption against the strong background nonthermal emission at low frequencies .
this effect is due to a free - free absorption coefficient which increases at low frequencies as @xmath2 ( nord et al .
the origin of the deficiency in radio continuum emission at high frequencies is due to the high column of embedded molecular gas that does not allow an external radiation field or cosmic ray particles to penetrate through the cloud .
this implies that this subset of molecular clouds is interacting with its surrounding hot medium .
we first demonstrate the physical situation in which radio dark clouds are produced followed by five examples demonstrating the reality of radio dark clouds .
since background radio continuum radiation is transparent when passing through neutral clouds , one would expect uniform background emission across the face of neutral clouds .
however , if a cloud is surrounded by hot synchrotron or thermal emitting plasma , the continuum emission is depressed due to the shorter path length of the continuum emission integrated along the line of sight toward the center of the molecular cloud .
one possibility involves external heating and ionization of neutral gas by ultraviolet continuum radiation that falls off rapidly from the edge to the center of the cloud with where the visual extinction ( a@xmath3 ) is much larger than one magnitude .
thus , free - free radio emission is substantially reduced in clouds with high column densities and their imprint can be identified as dark features in radio continuum images . in other words , atomic or molecular gas clouds or dust clouds suppress the continuum emission and create the appearance of a `` hole '' in their distribution .
it is expected that neutral clouds embedded in a hot plasma are edge brightened outlining the boundary of the cloud , thus can be distinguished from a cavity devoid of gas .
the presence of spectral line and/or continuum dust emission can also distinguish rdcs .
we consider a cloud with a diameter @xmath4 located at a distance d from us and is embedded within an ionized medium characterized to have electron density @xmath5 ( @xmath6 ) and with the emission measure @xmath7 @xmath8 along a path length @xmath9 .
the surface brightness toward the center and away from the cloud are defined as s@xmath10 and s@xmath11 , respectively , at a given frequency @xmath12 .
the flux deficiency @xmath13 is the difference between the flux density of the ambient gas toward and away from the cloud .
the ratio of the diameter of the cloud to the path length @xmath9 is @xmath14 the differential emission measure @xmath15em between the cloud center and the ionized medium can be estimated by @xmath16 where t@xmath17 is the electron temperature in k , @xmath18 is the beam size in arcsecond and @xmath19 is the flux deficiency in mjy .
if the electron density of the ionized medium is measured independently , then the depth of the molecular cloud @xmath4 along the line of sight can be estimated .
another possibility that could produce rdcs is the deflection of non - thermal particles as they diffuse inside a molecular cloud . in this case , the path length over which nonthermal particles travel are limited by the magnetic field geometry of the cloud which could shield the electrons penetrating into the cloud .
the ionization losses of nonthermal particles could also suppress the emission from high energy particles as they interact with the gas .
thus , the flux of nonthermal emission at high frequencies is expected to be reduced with respect to the background nonthermal emission .
if we assume that the magnetic field is in equipartition with the particles , the ratio of the magnetic field in the diffuse medium to the molecular cloud is @xmath20 where @xmath21 and @xmath22 are the magnetic fields in the ambient medium and dense cloud , respectively .
the spectral index of the emission @xmath23 , where s@xmath24 , is assumed to be constant .
multi - wavelength images presented here are based on observations that have already been described elsewhere .
the data that we have used are taken by mopra telescope ( jones et al .
2011 ) , green bank telescope ( gbt ) of the national radio astronomy observatory ( nrao ) , ( law et al . 2008 ) , very large array ( vla ) ( yusef - zadeh et al . 2004 ) , antarctic submillimeter telescope and remote observatory ( ast / ro ) ( martin et al . 2004 ) , irac on spitzer space telescope ( arendt et al .
2008 ) , scuba on james clark maxwell telescope ( pierce - price et al .
2000 ) , nobeyama radio observatory ( nro ) ( tsuboi et al . 2011 ) , and nicmos of hubble space telscope ( hst ) ( yusef - zadeh et al . 2001 ) .
we present below multi - wavelength observations of five sources toward the inner galaxy .
* g359.750.13 : * figure 1a shows the distribution of velocity integrated hcn ( 1 - 0 ) line emission from g359.75 - 0.13 which runs for @xmath25 parallel to the galactic plane ( jones et al .
this cloud is part of a layer of molecular gas associated with the central molecular zone .
this elongated cloud is detected at mid - ir and submm images as an irdc ( molinari et al .
2011 ; arendt et al .
2008 ; yusef - zadeh et al .
figure 1b shows the radio continuum counterpart to the molecular cloud at 3.5 cm . a dearth of emission coincides with the cloud tracing hcn line emission .
figure 1c shows a composite color image of hcn ( 1 - 0 ) emission and 3.5 cm continuum image .
radio continuum emission is present to the north and south of the elongated molecular gas layer . to determine the anti - correlation between radio continuum and molecular line emission , cross cuts along a line drawn on fig .
1a are made across the hcn and 3.5 cm radio continuum images and are presented in figure 1d .
the deficiency in the flux density of radio continuum emission is @xmath260.1 jy implying that the depth of the ionized gas and molecular gas is similar to each other .
these images demonstrate a clear anti - correlation in the distribution of molecular gas that is detected as an irdc and radio dark cloud .
the cross cuts combined with images of this cloud at radio and millimeter wavelengths suggest that the ionized and molecular gas in g359.750.13 are in the same environment , most likely in the galactic center region .
this is because diffuse radio continuum emission from the galactic center is much stronger than in the galactic disk .
* g0.13 - 0.13 and the radio arc : * adjacent to the nonthermal filaments of the arc near galactic longitude @xmath27 lies the molecular cloud g0.13 - 0.13 we compare the distribution of radio continuum and molecular line emission from g0.13 - 0.13 .
figure 2a - b show the distributions of a 20 cm continuum emission and integrated intensity of hcn ( 1 - 0 ) line , respectively .
the vertical filaments associated with the radio arc are known to be magnetized structures running perpendicular to the galactic plane . a 3.5 cm image of the same region mapped by the gbt shows similar morphology to that of the 20 cm continuum image .
using hcn line intensity and 3.5 cm continuum images , figure 2c shows cross cuts , along a line drawn on fig .
2b , demonstrating the depression in the continuum flux across the length of the cloud .
the dip at the location of g0.13 - 0.13 corresponds to @xmath28 mjy at 3.5 cm .
the molecular cloud g0.13 - 0.13 is surrounded by vertical nonthermal filaments of the arc to the east ( yusef - zadeh et al .
1987 ) and to the west of g0.13 - 0.13 ( reich 2003 ) .
the images and cross cuts made from hcn line and radio continuum data show anti - correlation between molecular line and radio continuum distributions .
the largest deficiency of radio continuum emission is located where molecular line emission peaks .
these intensity profiles imply the interaction of nonthermal radio filaments and g0.13 - 0.13 and are consistent with earlier studies measurements ( tsuboi et al . 1997 ) .
the flux of the continuum emission at 3.5 cm is reduced by a factor 2 where the molecular cloud g0.13 - 0.13 is located .
this implies that the region of nonthermal emission surrounding rdc is twice the depth of the molecular cloud .
* g359.160.04 and the snake filament : * one of the most prominent nonthermal filamentary structure in the galactic center region is the snake filament which extends for more than 20@xmath29 and runs almost perpendicular to the galactic plane ( e.g. , gray et al .
figure 3a shows the northern half of this striking filament terminating at the radio continuum source g359.16 - 0.04 ( uchida et al .
this 6 cm continuum image shows two dark features rdc-1 and rdc-2 to the e. and w. edges of the extended continuum source giving the appearance of a butterfly " .
rdc-2 lies to the north of the well - known radio jet 1e1740 - 2942 ( mirabel et al . 1992 ) .
given that the 6 cm data is produced by the vla and that incomplete _ uv _ coverage may be responsible for producing dark features , figure 3b shows a grayscale 3.5 cm continuum emission mapped by the gbt .
we note that the dark features are also identified in single dish observations , thus establishing the reality of rdcs in interferometric images .
contours of integrated emission of co ( 4 - 3 ) line are superimposed on the 6 cm continuum image , as shown in figure 3c .
given the mismatch between the 2@xmath29 resolution of molecular line data based on ast / ro observations and radio continuum data , the distribution of molecular gas is similar to that of dark radio clouds .
the continuum image reveals detailed morphological structures of a giant molecular cloud on arcsecond spatial resolution especially at the boundary of the molecular cloud near the continuum source .
we also made cross cuts at b=@xmath30 across the 6 cm and molecular line images , as shown on figure 3d .
the cross cuts show two peaks of molecular line emission coinciding with two dips in the 6 cm image with deficient radio flux of roughly 400 and 300 mjy .
these dips correspond to rdc-1 and rdc-2 , and imply that the diameters of rdc-1 and rdc-2 are roughly 4/5 and 3/5 of the depth of the ionized medium surrounding the cloud .
* g0.6 - 0.0 sgr b2 : * the sgr b2 cloud is a well - studied giant molecular cloud which lies near the galactic center ( reid et al . 2009 ) and is part of a continuous dust ridge that is viewed in absorption at mid - ir wavelengths ( lis & carlstrom 1994 ; molinari et al .
figure 4a shows the irdc associated with sgr b2 to the n. and ne of the cluster of stars and the nebula .
figure 4b , c show the distribution of 20 and 6 cm continuum emission from the same region , respectively .
the brightness of the 20 cm continuum emission is saturated to bring out an elongated dark feature near b@xmath31 .
radio continuum images show rdc-1 which appears to coincide with the southern edge of irdc that surrounds the cluster of uc hii regions in sgr b2 ( de pree et al .
we also note a decrease in the radio flux at 6 and 20 cm between sgr b2 and the isolated radio continuum feature g0.73 - 0.10 .
however , the image quality in this region is poor , thus may show artifacts .
cross cut plots along a line drawn on fig .
4c across the rdc-1 are made using the 6 cm continuum and 450@xmath32 m images .
the profiles of emission , as shown in figure 4d , show flux deficiencies of 5 and 2 mjy which imply that the depths of the ionized gas for the two dips are similar to and twice larger than those of molecular gas , respectively . because of the large number of bright compact hii regions in sgr b2 , the cross cuts also show fluctuations due to emission from individual compact sources .
* sgr a east snr & sgr a west : * given its gaseous environment with a strong radiation field , the inner ten parsecs of the galactic center is an excellent site to search for rdcs tracing molecular clouds that interact with the strong radiation field or with expanding supernova remnants .
the strong radiation field in the galactic center produces a thick layer of ionized gas at the edge of dark clouds .
one of the most prominent giant molecular clouds in the galactic center is the 50 g0.02 - 0.07 .
this cloud is physically interacting with the expanding shell of the sgr a east snr g0.0 + 0.0 which is thought to lie near the galactic center ( e.g. , tsuboi et al .
this complex region is also the site of young massive star formation as a chain of compact hii regions that lie to the east of the sgr a east remnant .
figures 5a - b show the inner 4@xmath29 of the galactic center at 6 cm hosting the shell - type snr g0.0 - 0.0 and four compact hii regions a - d at its eastern boundary ( zylka et al .
the 1.3 mm continuum distribution traces dust emission from the 50 molecular cloud .
we note several dark features rdc-1 to 4 surrounding the snr shell of figure 5a , all of which coincide with dust and molecular features ( tsuboi et al .
the largest scale dark cloud rdc-1 is noted to the ne of the chain of hii regions where we note an oval - shaped structure with an extent of [email protected] .
figure 5c shows contours of integrated emission of sio ( 2 - 1 ) superimposed on the 6 cm image , supporting the suggestion that rdc-1 coincides with an oval - shaped molecular gas .
another chain of hii regions to the ne of rdc-1 near @xmath34 might be associated with a second site of star formation in this cloud . to estimate the level at which the continuum emission is suppressed
, figure 5d shows cross cuts of rdc-3 made at 1.3 mm and 6 cm , supporting a clear anti - correlation between dust and radio continuum emission .
the largest deficient flux at 6 cm is 25 mjy implying that the size of the hot plasma is twice the size of the molecular cloud . on a smaller scale , the inner pc of the galactic center hosts diffuse ionized gas sgr a west orbiting the massive black hole ( e.g. , ferriere et al .
sgr a west consists of three arms of ionized gas and is surrounded by the circumnuclear molecular ring ( cmr ) .
figure 5e , f show a 3.5 cm and 1.87@xmath32 m continuum images of the eastern half of sgr a west whereas figures 5 g show molecular h2 1 - 0 s(1 ) counterparts ( yusef - zadeh et al .
several dark features are noted in the 3.5 cm continuum image , especially in the region between the n. and e. arms .
it turns out rdc-1 and rdc-2 lying between the two arms coincide with an extinction feature , as revealed in figure 5h where a drop in stellar density is noted .
cross cuts across the 3.5 cm and h2 images show that the h2 emission coincides with a dip in the continuum peak at 3.5 cm .
the deficient flux at 3.5 cm suggests that the depth of the ionizing gas is similar to that of the molecular gas . from equation ( 2 ) , the depth of the cloud is estimated to be 0.11 pc if @xmath35 @xmath6 and t@xmath36k .
the association of a `` tongue '' of neutral gas ( rdc-3 ) , an extinction cloud with the n. arm of sgr a west ( jackson et al .
1993 ; yusef - zadeh et al .
2001 ) , as well as the association of the e. arm with neutral gas suggest that there is considerable molecular gas in the ring and that the arms of sgr a west trace the ionized surface of neutral clouds .
if these clouds are massive , the presence of molecular gas inside the cmr can have important consequences in the formation of stars and the dynamics of stars near sgr a*. * conclusions : * we have presented five examples of dark features in radio continuum images of molecular complexes toward the inner galaxy .
we illustrated that these dark features are anti - correlated with molecular line and dust emission , thus implying that cold gas associated with radio continuum features can be detected in radio images .
given the new generation of radio telescopes with their broad band capability , future continuum measurements could potentially be effective in identifying cold gas clouds in star forming sites in the local and distant universe .
some of the dark radio clouds are surrounded by nonthermal radio emission , thus , radio continuum imaging can identify the interaction of nonthermal particles from radio jets or jet - driven outflows or supernova remnants with the cold ism or the igm .
lastly , rdcs can potentially be used as off positions in total power technique of observations with radio telescopes .
acknowledgments : i am grateful to m. wardle , d. roberts , r. arendt , w. cotton , m. royster , m. tsuboi and other colleagues for discussions and for providing me with their data over the years .
this work is partially supported by the grant ast-0807400 from the nsf . .
_ ( b ) middle - _ similar to ( a ) except showing the distribution of radio continuum emission at 3.5 cm based on gbt observation with a spatial resolution of 88@xmath37 .
_ ( c ) bottom - _ a color composite image showing ( a ) and ( b ) in red and green , respectively .
_ ( d ) right - _ a cross cut , as drawn on ( a ) , made at constant l=16@xmath38 showing the intensity profiles of hcn line emission in red and and of 3.5 cm continuum data in blue . , title="fig : " ]
_ ( b ) middle - _ similar to ( a ) except showing the distribution of radio continuum emission at 3.5 cm based on gbt observation with a spatial resolution of 88@xmath37 . _
( c ) bottom - _ a color composite image showing ( a ) and ( b ) in red and green , respectively . _
( d ) right - _ a cross cut , as drawn on ( a ) , made at constant l=16@xmath38 showing the intensity profiles of hcn line emission in red and and of 3.5 cm continuum data in blue .
, title="fig : " ] .
_ ( b ) middle - _ similar to ( a ) except showing the distribution of radio continuum emission at 3.5 cm based on gbt observation with a spatial resolution of 88@xmath37 .
_ ( c ) bottom - _ a color composite image showing ( a ) and ( b ) in red and green , respectively .
_ ( d ) right - _ a cross cut , as drawn on ( a ) , made at constant l=16@xmath38 showing the intensity profiles of hcn line emission in red and and of 3.5 cm continuum data in blue . , title="fig : " ] . _ ( b ) middle - _ similar to ( a ) except showing the distribution of radio continuum emission at 3.5 cm based on gbt observation with a spatial resolution of 88@xmath37 . _ ( c ) bottom - _ a color composite image showing ( a ) and ( b ) in red and green , respectively . _
( d ) right - _ a cross cut , as drawn on ( a ) , made at constant l=16@xmath38 showing the intensity profiles of hcn line emission in red and and of 3.5 cm continuum data in blue . , title="fig : " ] . _
( b ) top right - _ the integrated emission of the hcn line over velocities between 0 and 50 with a spatial resolution of 39@xmath37 ( jones et al .
_ ( c ) bottom left - _ cross cuts aslong a line drawn on ( b ) are made at constant l@xmath39 showing the profiles of hcn line emission in red and of 3.5 cm continuum emission based on gbt observation in blue .
, title="fig : " ] . _
( b ) top right - _ the integrated emission of the hcn line over velocities between 0 and 50 with a spatial resolution of 39@xmath37 ( jones et al .
_ ( c ) bottom left - _ cross cuts aslong a line drawn on ( b ) are made at constant l@xmath39 showing the profiles of hcn line emission in red and of 3.5 cm continuum emission based on gbt observation in blue .
, title="fig : " ] and rms noise 10@xmath32jy .
_ ( b ) bottom left - _ similar to ( a ) except that a 3.5 cm continuum image taken with the gbt with a spatial resolution of 89@xmath37 . _
( c ) top right - _ a weighted average contours of co ( 4 - 3 ) line emission integrated between -150.5 and -140.5 with a resolution of @xmath40 superimposed on ( a ) . _ ( d ) bottom - _ cross cuts along a line drawn on ( a ) at b=@xmath30 show two co peaks and two 6 cm continuum dips , corresponding to rdc-1 and rdc-2 . the vla image at 6 cm is primary beam corrected . , title="fig : " ] and rms noise 10@xmath32jy . _ ( b ) bottom left - _ similar to ( a ) except that a 3.5 cm continuum image taken with the gbt with a spatial resolution of 89@xmath37 . _
( c ) top right - _ a weighted average contours of co ( 4 - 3 ) line emission integrated between -150.5 and -140.5 with a resolution of @xmath40 superimposed on ( a ) . _ ( d ) bottom - _ cross cuts along a line drawn on ( a ) at b=@xmath30 show two co peaks and two 6 cm continuum dips , corresponding to rdc-1 and rdc-2 . the vla image at 6 cm is primary beam corrected . , title="fig : " ] and rms noise 10@xmath32jy . _ ( b ) bottom left - _ similar to ( a ) except that a 3.5 cm continuum image taken with the gbt with a spatial resolution of 89@xmath37 . _
( c ) top right - _ a weighted average contours of co ( 4 - 3 ) line emission integrated between -150.5 and -140.5 with a resolution of @xmath40 superimposed on ( a ) . _ ( d ) bottom - _ cross cuts along a line drawn on ( a ) at b=@xmath30 show two co peaks and two 6 cm continuum dips , corresponding to rdc-1 and rdc-2 . the vla image at 6 cm is primary beam corrected . , title="fig : " ] m with a resolution of @xmath41 _ ( b ) top right - _ a continuum image of sgr b2 at 20 cm based on combining data taken from the c and d configurations of the vla and the gbt before the image was convolved to a resolution of 30@xmath37 .
_ ( c ) bottom left - _ similar to ( b ) ( both vla and gbt data combined ) but at a resolution of 10.8@xmath42 _ ( d ) bottom right - _ cross cuts centered at @xmath43 along a line drawn on ( c ) show profiles of emission at 450@xmath32 m and 6 cm in blue and red , respectively . ] , stellar subtracted h2 1 - 0 s(1 ) emission at 2.12@xmath32 m , an image of 1.87@xmath32 m emission from stars and ionized gas , the cross cuts along the line drawn on ( f ) show the emission profiles of 3.5 cm and molecular h2 line .
, title="fig : " ] , stellar subtracted h2 1 - 0 s(1 ) emission at 2.12@xmath32 m , an image of 1.87@xmath32 m emission from stars and ionized gas , the cross cuts along the line drawn on ( f ) show the emission profiles of 3.5 cm and molecular h2 line .
, title="fig : " ] | we show radio continuum images of several molecular complexes in the inner galaxy and report the presence of dark features that coincide with dense molecular clouds . unlike infrared dark clouds , these features which we call
`` radio dark clouds '' are produced by a deficiency in radio continuum emission from molecular clouds that are embedded in a bath of uv radiation field or synchrotron emitting cosmic ray particles .
the contribution of the continuum emission along different pathlengths results in dark features that trace embedded molecular clouds .
the new technique of identifying cold clouds can place constraints on the depth and the magnetic field of molecular clouds when compared to those of the surrounding hot plasma radiating at radio wavelengths .
the study of five molecular complexes in the inner galaxy , sgr a , sgr b2 , radio arc , the snake filament and g359.75 - 0.13 demonstrate an anti correlation between the distributions of radio continuum and molecular line and dust emission .
radio dark clouds are identified in gbt maps and vla images taken with uniform sampling of _ uv _ coverage .
the level at which the continuum flux is suppressed in these sources suggests that the depth of the molecular cloud is similar to the size of the continuum emission within a factor of two .
these examples suggest that high resolution , high dynamic range continuum images can be powerful probes of interacting molecular clouds with massive stars and supernova remnants in regions where the kinematic distance estimates are ambiguous as well as in the nuclei of active galaxies . & pdflatex # 1 ( g10 ^ 3 g_0 ) ^#1 # 1 10^#1 km s@xmath0 # 1 10^#1 # 1 # 2 ^#2
u#1 @xmath1 i v |
overcharging is a surprising phenomenon in which conterions adsorbed to a surface exceed the number of fixed surface charges @xcite . as a consequence
, coions are drawn from the bulk toward the overcharged surface leading to a formation of a `` triple - layer '' .
because the underlying mechanism of overcharging relies on electrostatic correlations @xcite , it has become synonymous with the strong - coupling limit .
indeed , a mean - field treatment which accurately captures the weak - coupling limit , can not describe overcharging .
( however , the mean - field in combination with the excluded volume interactions can induce overcharging if the bulk volume fraction of an electrolyte is sufficiently large to generate a depletion force that pushes particles against a surface .
this effect is seen in uncharged systems and persists for weakly charged surface charges @xcite . ) in the strong - coupling limit overcharging is the result of increased structuring within the layer of counterions .
the electrostatic correlations between the condensed counterions lead to formation of correlation `` holes '' within the layer of condensed ions which can attract excess of counterions from the bulk .
the value of the coupling constant @xmath0 , which is the ratio between the bjerrum and gouy - chapman length ( @xmath1 is the valency ) , estimates the extent of correlation effects . in the limit @xmath2 , the counterions are said to freeze into a 2d wigner crystal @xcite . at large , but finite @xmath3 , the local structure of an ionic fluid remains wigner - like @xcite .
the above mechanism is specific to coulomb interactions that diverge as @xmath4 and , therefore , exhibit the excluded volume effects @xcite . if , however , the divergence in the pair interaction is removed ( the pair potential is bounded as @xmath4 ) , particles can interpenetrate and the usual excluded volume interactions underlying the crystal formation are eliminated ( at sufficiently high temperature and/or density ) .
for some family of bounded potentials , particles can form stacks where two or more particles occupy the same position and act as an effective single particle .
this is possible only if a pair potential is sufficiently flat around @xmath5 @xcite .
one example is the penetrable sphere model where the pair potential is the step function @xcite .
the stacking formations stabilize the liquid phase @xcite , since doublets , triplets , etc . effectively decrease the number of particles .
the presence of stacked formations is signaled as a positive peak in the pair correlation function at @xmath4 @xcite . extrapolating these ideas to ions , which in addition to penetrable cores have long - range coulomb interactions , we ask what influence penetrability has on the structure of a double - layer .
can the restructuring invoked by the penetrability lead to overcharging in the weak - coupling limit ? in the present work , the divergence in the coulomb potential is removed by smearing - out the central charge of an ion over a finite region .
the penetrating core , then , depends on the weight function used to smear out the charge .
this procedure does not require going beyond the framework of electrostatics and the weak - coupling limit can be described by the modified version of the poisson - boltzmann equation .
bulk properties and phase diagram of penetrable ions have been investigated in @xcite .
the main feature is the formation of the bjerrum pairs of two opposite ions that function as polarizable particles .
the formation of these pairs leads to an insulator - conductor transition @xcite , which does not exist in systems of hard - core ions in three dimensions @xcite .
thus , penetrability dramatically affects the phase transition and the topology of the phase diagram .
the model of penetrable particles is not only of theoretical interest .
various macroparticles can exhibit interpenetration .
marquest and witten @xcite suggested a penetrable sphere model for micelles .
polymer coils and dendrimers in good solvent can be represented by a gaussian core model @xcite .
if in addition these macroparticles are charged , as is often the case for real systems @xcite , then the model of penetrable ions can be of genuine physical relevance .
recently , ionic microgels have been modeled as uniformly charged spheres @xcite , allowing interpenetration at short- and coulomb interactions at long - separations .
the charge @xmath6 of an ion @xmath7 is smeared out over the finite region according to the weight function @xmath8 such that @xmath9 .
the charge density operator for @xmath10 smeared - out ions is @xmath11 in this work we consider a symmetric electrolyte , @xmath12 .
the averaged charge density for this system is @xmath13 , \label{eq : rhoc}\ ] ] where @xmath14 and @xmath15 denotes the number density of anions and cations , respectively .
the poisson equation is @xmath16,\ ] ] where @xmath17 is the background dielectric constant . to obtain a closed equation
, we need an expression for @xmath15 and @xmath14 in terms of the mean electrostatic potential @xmath18 . for point charges
this leads to @xmath19 where the subscript @xmath20 is either @xmath21 or @xmath22 , and @xmath23 is the bulk salt concentration .
the number density depends locally on an electrostatic potential . however , if charge is smeared around the ion center at @xmath24 , the entire distribution @xmath25 interacts with the mean electrostatic potential , @xmath26 and the number density becomes @xmath27 we may now write down the mean - field equation for the electrostatic potential produced by smeared - out ions , @xmath28 we refer to this modified poisson - boltzmann equation the finite - spread pb equation ( fspb ) .
the fspb equation complements the already quite sizable set of modified pb equations : the pb that incorporates the excluded volume interactions @xcite , the dipolar interactions @xcite , the nonlinear solvent contributions @xcite , and the polarizability of ions @xcite .
the idea of finitely spread - out ions was considered in @xcite to study the self energy contributions beyond the mean - field .
mathematically , the fspb equation has the structure of a convoluted equation .
each ion is convoluted according to the weight function which determines the composition of a single ion .
( [ eq : fspb ] ) can also be obtained by minimizing the grand potential , @xmath29 & = & \frac{1}{2}\int d{\bf r}\int d{\bf r}'\ , \frac{\rho_c({\bf r})\rho_c({\bf r}')}{4\pi\epsilon|{\bf r}'-{\bf r}| } \nonumber\\ & + & k_bt \sum_{\alpha}\int d{\bf r}\,\rho_{\alpha}({\bf r } ) \big[\log\rho_{\alpha}\lambda^3 - 1\big ] \nonumber\\ & -&\sum_{\alpha}\mu\int d{\bf r}\,\rho_{\alpha}({\bf r } ) , \label{eq : omega}\end{aligned}\ ] ] where @xmath30 is the thermal de broglie wavelength and @xmath31 is the chemical potential .
the three contributions are the electrostatic energy , entropy , and we allow the number density to fluctuate through the contact with a reservoir .
the minimization @xmath32 recovers eq .
( [ eq : rho_alpha ] ) and the application of the poisson equation yields the fspb model .
the concept of charge smearing is not novel to this work , but it has been evoked many times in the past both as a physical representation and a mathematical construct .
the best known example ( and to our knowledge the earliest ) of mathematical construct is the calculation of ewald summation to treat periodic charges @xcite , today the most practiced method to account for contributions due to periodic boundary conditions of a simulated system @xcite , where by spreading the charge one achieves separation of interactions into the short- and long - range counterpart .
another instance of mathematical convenience is the onsager smearing optimization to obtain the variational free energy of the strongly correlated one component plasma @xcite .
the idea of charges which at short separations exhibit soft repulsion and at long separations the coulomb interaction appeared in @xcite to represent semi - classical hydrogen plasma at high temperature and density , yet no smearing procedure was used to construct this potential .
the actual smearing procedure to represent physical particles was used to model electrons set in cyclotron motion by a uniform magnetic field @xcite . in soft matter
, the smearing out procedure was used to represent microgels in @xcite , and recently to represent charged polymers @xcite .
we consider a few simple , spherically symmetric distributions @xmath33 and the interactions that result from these distributions .
a spherically symmetric charge distribution generates the following electrostatic potential , @xmath34.\ ] ] to derive the above formula we applied the gauss law to the poisson equation , and afterwards we integrated the resulting electrostatic field ( the integration by parts was evoked ) .
if the distribution is uniform inside a spherical cavity , @xmath35 where @xmath36 is the heaviside function and @xmath37 , then the potential inside a sphere is @xmath38 .
\label{eq : psi_ps}\ ] ] outside the sphere the coulomb potential is recovered .
the interaction between two ions with this charge distribution is @xcite @xmath39 .
\label{eq : u_ss}\ ] ] for non - overlapping separations , ions behave like point charges .
next , we consider the distribution @xmath40 where a charge is smeared - out over a spherical shell . inside a sphere ,
the electrostatic potential is constant , @xmath41 the interaction between two of these distributions for overlapping separations is @xmath42.\ ] ] a sphere and a shell distributions interact via @xmath43.\ ] ] finally , we consider a gaussian distribution @xcite , @xmath44 which leads to the following pair interaction @xmath45
we primarily focus on the distribution @xmath46 where @xmath47 is the ionic volume .
the fspb equation in reduced units reads @xmath48 where @xmath49 @xmath50 , @xmath51 is the screening length , and @xmath52 is the bjerrum length .
the number density of each species is @xmath53 all the ions occupy the half - space @xmath54 .
the charged hard wall at @xmath55 determines boundary conditions , @xmath56 where @xmath57 is the surface charge .
note that the number density is limited to the region @xmath54 , but the charge density , because of the distribution @xmath33 , reaches all the way to @xmath55 , the location of the charged wall .
consequently , we assume the ( hard - sphere)-(hard - wall ) potential between an ion and the wall .
subsequent figures show data points for @xmath54 , the region available to ion centers . in the dilute limit ( @xmath58 ) , overcharging does not happen , the solution of the pb equation for point ions in the weak - coupling limit yields an algebraic density profile , @xmath59 in the strong - coupling limit the electrostatic correlations modify the functional form of the distribution @xcite , @xmath60 where @xmath61 is the gouy - chapman length .
the barometric - like distribution in the transverse direction is a consequence of high degree of ordering in the lateral direction reminiscent of the wigner crystal @xcite .
although the fspb equation is purely mean - field , we find similar modification of the density profile , but this time the relevant parameter is @xmath62 .
the limit @xmath63 corresponds to the ionic distribution given by eq .
( [ eq : wcl ] ) , and the limit @xmath64 to the one given by eq .
( [ eq : scl ] ) , see fig .
( [ fig : rho_counter_scl2 ] ) . in the limit @xmath64 ,
( [ eq : u_ss ] ) , we recover an ideal gas particles in a uniform gravitational field .
( [ fig : rho_counter_r8_z0_1.0 m ] ) shows the counterion profiles for an electrolyte solution .
this introduces a length scale @xmath65 .
counterions from the bulk can now overcharge the surface .
the counterion profile beyond @xmath66 dips below the bulk value , which indicates an overcharged surface .
( [ fig : rho_co_r8_z0_1.0 m ] ) shows how the coion density rises above the bulk value , another signature of charge reversal
. finally , fig .
( [ fig : phi_r8_z0_1.0 m ] ) shows the non - monotonic electrostatic potential which goes to negative values and has a minimum around @xmath67 , at which point the electrostatic field vanishes and , farther on , it changes sign
. under what conditions penetrability of ions leads to charge reversal ?
for penetration to take place , either the thermal fluctuations must exceed the interaction energy of the overlapped particles , @xmath68 , or the counterion concentration at the wall must be sufficiently large to involve numerous overlaps , such that the effective 2d density within the counterion layer is @xmath69 . for the distribution in eq .
( [ eq : h1 ] ) the first requirement translate into @xmath70 , and , assuming @xmath71 , the second one becomes @xmath72 . in fig .
( [ fig : rho_counter_r8_z0_1.0 m ] ) @xmath73 and @xmath74 , and so both criteria are met . as a parenthetical note ,
we point out that the pb and the fspb models satisfy the same contact value relation , @xmath75 where @xmath76 is the ideal gas pressure in the bulk , @xmath77 and @xmath78 is the total density in the bulk and at the contact with a wall , respectively .
the sole constraint that @xmath25 needs to satisfy is that its integral recovers the charge of an ion .
this leaves sufficient room to manipulate the short - range interactions . as an example we can take a mixed distribution @xmath79,\nonumber\\ \label{eq : h2}\end{aligned}\ ] ] comprised of a charged shell and sphere .
@xmath80 is the surface area of a sphere .
@xmath81 is the additional parameter and @xmath82 corresponds to the distribution in eq .
( [ eq : h1 ] ) .
the number density is @xmath83},\ ] ] where @xmath84 fig .
( [ fig : u_z ] ) plots the pair potentials for different @xmath81 .
the parameter @xmath81 controls the strength of the repulsion and , by the same token , the strength of the excluded volume interactions . by way of example , fig .
( [ fig : rho_counter_r8_z4_1.0 m ] ) shows the density profile for @xmath82 and @xmath85 , for @xmath86 . the excluded volume contributions for @xmath85 expel counterions from the first layer , which is the opposite of overcharging seen for @xmath82 .
with this example we try to demonstrate possible extensions of the model based on charge spreading . for example , the repulsion controlled by the parameter @xmath81 could represent the interaction between polymers on account of the self - avoiding walk of polymer chains .
to test the mean - field approximation implicit in the fspb equation , we carry out monte carlo simulations .
the size of the simulation box is @xmath87 and @xmath88 .
the periodic boundary conditions are in the lateral @xmath89-directions .
the bounding plates at @xmath90 and @xmath91 have opposite charge .
the box encloses @xmath92 ions , @xmath93 . to check for the finite size effects , we double the simulation size for some parameters .
the standard ewald summation is applied for contributions of periodic images @xcite . if particles overlap , we supplement the interaction energy with the term : @xmath94 .
( a physical picture is slightly modified when doing simulations . in simulations
we tend to think of particles as point charges which at separations @xmath95 switch to the non - coulomb interactions @xmath96 . within the mean - field theory based on the poisson equation ,
we tend to think of an ion as a charged ball .
the two conceptions are , however , identical . )
( [ fig : rho_counter_sim ] ) compares counterion profiles obtained from the simulation and the fspb equation . for @xmath86
the agreement is virtually exact .
for the smaller radius @xmath97 we see the two results deviate : in the simulation , the correlations generate a more concentrated layer of counterions ( but no overcharging ) . in fig .
( [ fig : rho_co_sim ] ) we plot the coion density profile for @xmath86 to further confirm the accuracy of the mean - field in this regime .
what causes charge reversal for smeared - out ions ? do counterions merely penetrate or the pair potential is sufficiently flat in the overlap region and counterions collapse into stacked formations , so that the pair correlation function exhibits positive peak as @xmath4 ? in fig .
( [ fig : c1 ] ) we show the configuration snapshots for counterions adsorbed on a charged surface for different values of @xmath62 .
the difference in structure is perceptible .
ionic penetration favors smaller separations between counterions .
this gives the impression of string - like formations .
[ cols="^,^ " , ] as mentioned in the introduction , not all penetrable potentials can lead to stacked configurations , where two or more particles occupy `` the same '' space .
for example , the gaussian core model precludes @xcite , while the penetrable sphere model favors the stacked configurations @xcite .
stable stacked aggregates require that a pair potential be sufficiently flat .
a more rigorous test involves the fourier transform of the pair interaction .
if @xmath98 oscillates , and , therefore , involves negative values , then the stacking takes place under certain conditions @xcite . on the other hand ,
if @xmath98 is non - negative , stacking does not occur under any conditions .
this criterion is arrived at by considering the ornstein - zernike relation which , in the fourier space , is @xmath99 where @xmath100 and @xmath101 is the direct and pair correlation function , respectively . in the mean - field approximation @xmath102
[ the exact definition is @xmath103 , and in the mean - field @xmath104 is the first term in eq .
( [ eq : omega ] ) ] , so that @xmath105 if we take @xmath106 to be the signature of stacking , and @xmath107 then it becomes immediately clear that a non - negative @xmath98 always yields @xmath108 , and no stacking occurs .
only an oscillating @xmath98 can yield @xmath109 . in the present model particles are smeared - out charges , @xmath110 and in the fourier space @xmath111 this
yields @xmath112 we see that @xmath113 regardless of the distribution @xmath33 .
we conclude that soft interactions generated by charge spreading can not lead to stacked configurations . in fig .
( [ fig : corr ] ) we show the lateral correlation function for the adsorbed coutnerions , @xmath114 .
penetration reduces the degree of electrostatic correlations between the ions , however , @xmath114 always remains negative and decreases monotonically all the way to @xmath5 .
the small difference between the shell and the sphere distributions indicates that the exact shape of the pair potential is unimportant . in fig .
( [ fig : bond2 ] ) we show the probability distribution of the nearest neighbor separation for various @xmath33 functions .
differences are rather small and only quantitative .
the gaussian @xmath33 shows preference for smaller interionic separation .
finally , we explore the validity of the mean - field for multivalent symmetric ions , @xmath115 .
we suppose that the mean - field should deteriorate quickly for @xmath115 since the coupling constant @xmath116 .
( [ fig : rho_counter_q_sim ] ) shows the counterion density profiles for @xmath117 .
for @xmath118 the mean - field is virtually exact and for @xmath119 there are small deviations .
this surprising agreement can be explained with the following .
increasing @xmath1 naturally reduces the number of counterions required to neutralize the wall . on the other hand ,
a larger @xmath1 generates stronger overcharging so the density drop is partially compensated .
the number of overlapping configurations is still large and the mean - field retains its validity . to confirm this conjecture
, we check the 2d density of the counterions adsorbed on the charged wall . for @xmath119
, we find @xmath120 .
the overlapping configurations are , therefore , still significant .
for overcharging to occur , there must exist a mechanism that either reduces the free energy of counterions condensed on a charged surface ( it becomes favorable for a counterion to leave its bulk environment ) or increase the free energy of a bulk ( it becomes less favorable for ions to stay in the bulk ) .
it is the exchange between the two environments , the bulk and the double - layer , that leads to overcharging .
we consider the environment of a charged surface with its adsorbed counterions and consider the conditions that lead to lowering of its energy .
the surface charge of a plate is @xmath57
. we do not violate the condition of neutrality , thus the 2d density of adsorbed counterions is @xmath121 .
we focus on the energy of a single counterion . for sake of simplicity , we assume counterions and the charged wall occupy the same plane .
an attraction to the wall yields a negative energy contribution , @xmath122 \nonumber\\ & = & \lim_{r_c\to\infty}\bigg[-2\pi z\lambda_b\sigma_c r_c\bigg ] \end{aligned}\ ] ] which is countered by the repulsive interactions with other counterions , @xmath123,\ ] ] where @xmath124 is the pair distribution function . for absence of correlations @xmath125 , and the attractive contribution
is completely canceled out , @xmath126.\ ] ] but the cancellation will not be exact if correlations are present . separating correlations from the pair distribution function , @xmath127 , the uncanceled correlation energy is @xmath128 the relation between the correlation energy and the correlation function is quite clear . an ion will generate a negative correlation hole in its neighborhood , which will lead to negative energy contributions . we assume the following simple correlation hole , @xmath129 and @xmath130 , that is , by fixing a position of an ion , we introduce a circular hole in the density profile with the correlation length @xmath131 .
the conservation of mass condition requires that an area of the hole is related to the density via @xmath132 , therefore , the correlation length is @xmath133 .
this correlation function is the low temperature limit of the correlation hole obtained by construction from the linear mean - field treatment @xcite . inserting this @xmath101 into eq .
( [ eq : e_h ] ) we can approximate the correlation energy in the strong - coupling limit , @xmath134 recalling the definition of the coupling constant @xmath135 , we get @xmath136 but for penetrable ions correlations are not required for reducing the electrostatic energy of counterions the fact demonstrated by the validity of the mean - field theory .
consequently , we set @xmath137 .
the reduction in electrostatic energy comes from a different source , from the fact that at overlapping separations , @xmath95 , the electrostatic interactions are reduced on account of smearing procedure , which leads to the energy gain @xmath138\nonumber\\ & = & -c\pi z\lambda_b\sigma_c r , \label{eq : overlap}\end{aligned}\ ] ] where the constant @xmath139 depends on the pair potential @xmath96 , which , in turn , is determined by the distribution function @xmath33 .
for the distributions considered in this work : @xmath140 , @xmath141 , and @xmath142 .
comparing eq .
( [ eq : correlation ] ) with eq .
( [ eq : overlap ] ) we see the different dependence of each mechanism on different parameters .
the stabilization based on penetration has stronger dependence on the surface charge and the bjerrum constant , on the other hand , its dependence on valency is weaker . in the strong - coupling limit , penetrable ions will exhibit correlated motions . if the correlation length is larger than the diameter of an ion , @xmath143 , penetration may be neglected and the former mechanism comes to the fore .
on the other hand , if @xmath144 , we expect the two contributions to mix . to recap , both mechanisms depend on eliminating the energy contributions coming from short separations between ions . for point ions in the strong - coupling limit
, configurations with short separations are eliminated through correlated motion .
the price is sacrifice in entropy , despite this , the total free energy is lowered . for penetrable ions the problem of high energy contributions at short separations
does not exist to begin with . due to smearing out procedure of an ion charge and the removal of the divergence from pair interactions , these contributions
are taken out of the picture .
consequently , there is no entropy price to be paid , as all separations are explored `` equally '' and the mechanism is valid in the weak - coupling limit .
the present work studies the structure of a double - layer composed of ions whose central charge is smeared over a finite region in accordance with a weight function @xmath33 .
the smearing - out procedure removes the divergence as @xmath4 from the pair interaction , allowing for interpenetration between the ions .
the conditions under which penetration is favored are large temperature and high density .
this regime is suitable for the mean - field treatment .
accordingly , we derive a modified poisson - boltzmann equation for spread - out charges ( the fspb equation ) . the fspb equation predicts that for sufficiently large spreading radius @xmath62 , overcharging takes place the mc simulations verify this prediction .
this suggests an alternative mechanism for charge reversal that is not related to correlations and the strong - coupling limit .
using simulations and the mean - field approximation , we can exclude the ionic stacking as the underlying mechanism of overcharging .
in fact , any soft repulsion obtained by charge spreading can not lead to stacked configurations . as a final consideration
, we address the physical relevance of the smeared - out ion model . the spreading - out of the charge
may capture the interactions between charges distributed along the polymer chains , but a more realistic representation would involve a non - electrostatic component produced by the self - avoidance of the polymer chains . within the electrostatics framework ,
we have suggested a plausible distribution @xmath33 composed of a charged shell and sphere , which generates an additional repulsion inside a penetrable core . in the end , however , a physically accurate pair potential requires corroboration with experimental analysis .
d.f . is grateful to nathan berkovits for an invitation to and short sojourn in the ictp - saifr where , prompted by the theoretical atmosphere of the institute , the formal aspects of the fspb equation were laid down .
this work was partially supported by the cnpq , fapergs , inct - fcx , and by the us - afosr under the grant fa9550 - 12 - 1 - 0438 .
g. m. torrie and j. p. valleau , j. chem .
phys . * 73 * , 5807 ( 1980 ) ; e. gonzales - tovar , m. lozada - cassou and d. henderson , j. chem . phys . * 83 * , 361 ( 1985 ) ; m. lozada - cassou , r. saavedra - barrera and d. henderson , j. chem . phys . * 77 * , 5150 ( 1982 ) .
grosberg , t. t. nguyen , and b. i. shklovskii , rev .
phys . * 74 * 329 ( 2002 ) ; y. levin , rep .
phys . * 65 * , 1577 ( 2002 ) ; w.m .
gelbart , r. f. bruinsma , p. a. pincus , and v. a. parsegian , phys . today * 53 * , 38 ( 2000 ) ; f. j. solis and m. o. de la cruz , ibid . *
54 * , 71 ( 2001 ) .
s. pianegonda , m. c. barbosa , and y. levin , europhys . lett . * 71 * , 831 ( 2005 ) ; a. diehl and y. levin , j. chem . phys . * 129 * , 124506 ( 2008 ) ; a. p. dos santos , a. diehl , and y. levin , j. chem . phys . * 132 * , 104105 ( 2010 ) . d. frenkel and b. smit , understanding molecular simulation : from algorithms to applications , academic press , 2nd edition , ( 2002 ) ; m. p. allen and .
d. j. tildesley , computer simulation of liquids , oxford university press , ( 1989 ) . | we investigate a double - layer of penetrable ions near a charged wall . we find a new mechanism for charge reversal that occurs in the weak - coupling regime and , accordingly , the system is suitable for the mean - field analysis .
the penetrability is achieved by smearing - out the ionic charge inside a sphere , so there is no need to introduce non - electrostatic forces and the system in the low coupling limit can be described by a modified version of the poisson - boltzmann equation .
the predictions of the theory are compared with the monte carlo simulations . |
commonly , the energy gap @xmath2 of a superconducting film is determined in tunneling experiments where a conducting electrode and the superconductor are separated by an insulating layer . in early tunneling experiments ,
planar tunneling junctions with a thick oxide layer @xcite or point contacts @xcite were utilized to determine the superconducting properties of many typical superconductors . with the advancement of low temperature scanning tunneling spectroscopy ( sts ) in ultra - high vacuum ( uhv ) , it has become possible to probe _ in - situ _ fabricated superconductors , and to determine @xmath2 with atomic - scale spatial resolution @xcite . here
, @xmath2 can be determined with a high degree of accuracy as a result of higher sample quality resulting from _ in - situ _ preparation , and utilizing the vacuum barrier which acts as a perfect insulator . in recent years
, such investigations have shown that seemingly well understood , elemental bcs superconductors , like pb or in , behave dramatically different at the thin film limit . for example , the onset of quantum well states allows superconductivity to remain robust in the thin film limit @xcite , or even persist down to a single layer @xcite . for other systems , like pb - bi alloys
, it has been shown that @xmath1 can be engineered by modifying the fermi wavevector @xmath3 @xcite .
a thorough understanding of superconductivity at these length scales is not only interesting from an academic point of view , but also required for possible applications of superconductivity in nanoscale devices , mandating local probe investigations with high spatial and energy resolution .
although the preparation quality is especially crucial for reactive materials as , e.g. , lanthanides , this technique has not been applied to lanthanum yet , one of the few elemental superconductors with @xmath4 k @xcite .
the superconductivity of bulk la was investigated in a number of experimental studies @xcite , including planar tunneling @xcite and point contact spectroscopy @xcite , and by theory @xcite .
most notably , la is an intermediate - coupling superconductor @xcite with a critical temperature at atmospheric pressure of @xmath5 k and @xmath6 k for the stable double hexagonally close packed ( dhcp ) and metastable fcc phase , respectively , and an extraordinary enhancement under compression @xcite .
studies about the low - dimensional properties of la are lacking . in this letter , we report on a sts study of the superconducting properties of dhcp la islands .
we observe that @xmath7 and @xmath1 of la are larger than commonly believed for clean bulk la @xcite . approaching the thin film limit , namely where the film thickness is comparable to the coherence length @xmath8 , we determine a monotonous decrease of the superconducting properties , in agreement with a theoretical model that considers the boundary condition for the superconducting wavefunction @xcite , ruling out significant quantum size effects in the superconductivity .
we investigated samples of different purity and show that a reduction in @xmath1 is correlated with increased surface contamination and quenching of the unoccupied surface state of the la islands .
_ in - situ _ prepared la films were studied in a commercially available uhv stm @xcite at a base temperature of @xmath9 k , unless otherwise specified , and a base pressure @xmath10 mbar .
the w(110 ) surface was cleaned by cycles of annealing in an oxygen atmosphere and subsequent thermal flashing @xcite .
la films were grown by electron beam evaporation on a clean w(110 ) substrate held at room temperature .
afterwards , each sample was annealed for 5 to 12 minutes at temperatures in the range of 700 to @xmath11 and slowly cooled to room temperature , avoiding a rapid thermal quenching to bypass the metastable fcc phase .
two types of la sources with different nominal purity were used for the experiments @xcite .
the corresponding la films are named 1@xmath12 and 2@xmath13 generation samples in the following .
stm topographs were recorded in constant - current mode , with a sample bias voltage @xmath14 v and a tunneling current in the range of @xmath15 to @xmath16 pa . sts was performed using a standard lock - in technique , adding a modulation voltage @xmath17 to @xmath18 .
the @xmath19 spectra dedicated to study superconductivity were taken using normal metal and superconducting nb tips @xcite with @xmath20 to 150 pa at @xmath21 mv , @xmath22 to 0.07 mv . in this junction resistance range , we did not observe any effects of josephson supercurrents or andreev reflections .
the spectra for the surface state were recorded with @xmath23 pa at @xmath24 v , @xmath25 mv .
annealing of la / w(110 ) leads to stranski - krastanov growth @xcite , i.e. , a la wetting layer ( wl ) with a thickness of one monolayer ( ml ) on the w(110 ) surface [ fig .
[ fig : figure_1](a ) ] , generation sample with heights in ml as depicted by the numbers.,scaledwidth=80.0% ] while the additional material forms flat - top dhcp la(0001 ) islands [ fig . [ fig : figure_1](b ) ] . height and lateral extension of the islands depend on the combination of deposition time and rate , and annealing time and temperature . in total , islands with thicknesses @xmath26 in the range between @xmath27 nm and @xmath28 nm ( 8 to 460 ml ) were grown , which covers a broad range from the thin film to the bulk limit , with respect to the coherence length in the clean limit , @xcite . in order to avoid lateral size effects
, we only studied islands with a diameter @xmath29 .
sts on la islands reveals a symmetric superconducting gap , @xmath0 , around @xmath30 resulting from the superconductivity of the probed island [ fig .
[ fig : figure_2](a ) ] , while the wl shows no gap , indicating a normal metal .
moreover , @xmath0 is reduced with decreasing @xmath26 .
spectra taken with nb tips [ fig . [ fig : figure_2](b ) ] correspondingly show a gap on the wl stemming from the tip density of states ( dos ) , and a larger gap on the la islands stemming from the interplay of the tip and sample dos . for la islands with different thicknesses @xmath26 , using ( a ) normal metal and ( b ) superconducting nb tips .
experimental curves ( blue solid lines ) were taken on the wl to characterize the tip dos and with the same tip on the la islands with indicated @xmath26 .
the fitted calculations are shown as red dotted curves .
( a ) experimental data : , @xmath31 k. fitted calculations : , , @xmath32 , @xmath33 mev , @xmath34 , @xmath35 , @xmath36 mev ( from top to bottom ) .
( b ) experimental data : @xmath37 mv , @xmath38 k. fitted calculations : @xmath39 , @xmath40 mev , @xmath41 mev , @xmath42 mev , @xmath43 mev .
the curves are vertically shifted for visual clarity.,scaledwidth=80.0% ] in order to precisely extract @xmath0 for a given @xmath26 , the experimental curves were each fitted with a numerically calculated differential conductance ( appendix a ) , involving a bcs - like dos for the sample , or for both electrodes in the case of nb tips , @xmath44 here , @xmath45 is the dos of the electrode in its normal metal state which is assumed to be constant , @xmath46 is the energy , @xmath2 is the familiar energy gap from bcs theory , and @xmath47 is a broadening parameter which was originally introduced to describe the finite lifetimes of quasiparticles in the tunneling process @xcite
. the fitted calculations ( appendix a ) yield @xmath7 and @xmath48 as free fit parameters , while @xmath49 and @xmath50 were determined from the wl spectra taken with the same micro - tip . the excellent fit quality [ fig . [
fig : figure_2](a , b ) ] permits an accurate determination of these quantities , resulting in values of up to @xmath51 mev and @xmath52 to @xmath53 mev .
there is no strong lateral variation of superconductivity on the islands , at least for distances @xmath54 from the islands edges ( appendix b ) .
hence , we conclude that @xmath0 depends only on @xmath26 and @xmath55 .
the obtained @xmath56 is shown in fig .
[ fig : figure_3](a ) for both sample generations . below , we first focus on the results from the cleaner @xmath57 generation samples , and discuss the effect of impurities on superconductivity later .
going from the bulk limit @xmath58 to the thin film limit , the values of @xmath56 show a linear decrease as a function of inverse thickness with no obvious saturation above @xmath59 .
@xmath60 is determined from the experimental @xmath56 using bcs theory ( appendix c ) . in the la bulk limit
, the relation between the zero - temperature energy gap @xmath61 and @xmath60 was shown to equal @xmath62 @xcite .
we assume the same constant value for the whole range of la thicknesses studied here .
in fact , the only two effects which could alter its value for decreasing @xmath26 are ( i ) phonon softening @xcite , which would emerge as a shift of phonon modes in experimental @xmath19-curves @xcite , and ( ii ) electronic quantum size effects , which would appear as discrete states in the higher voltage range of the spectra .
there are no clear indications for such features ( fig .
[ fig : figure_2 ] , fig .
[ fig : figure_4 ] ) .
therefore , both effects can be ruled out , and should only emerge in very thin films of up to a couple of monolayers ( 5 to 10 ml in pb @xcite ) . for our thinnest films , where @xmath1 becomes comparable to @xmath55 , the finite experimental temperature is considered by calculating @xmath61 from @xmath56 by a numerical integration according to bcs theory ( appendix c ) . and ( b ) @xmath60 on the inverse film thickness @xmath63 .
sample generations and measurement temperatures are indicated .
the error bars are due to uncertainties in the measured film thickness and in the fit parameters .
the literature bulk values @xmath64 , @xmath65 , and @xmath66 are indicated by vertical and horizontal lines , respectively , in ( a , b ) . dashed and solid curves in ( b ) show power law fits and linear fits according to the simonin model ( eq . [ eq : t_c_vs_d ] ) , respectively.,scaledwidth=80.0% ] the obtained @xmath60 shows a linear behavior in @xmath63 with highest values of @xmath67 k for the thickest islands [ fig . [
fig : figure_3](b ) ] .
note that @xmath60 calculated from the @xmath68 k experimental data roughly coincides with @xmath60 calculated from the @xmath69 k data , justifying the determination of @xmath60 via bcs theory .
in order to explain the observed thickness dependence of @xmath1 , at first the very general relation @xmath70 is fitted to the experimental data [ dashed line in fig .
[ fig : figure_3](b ) ] .
the resulting fit parameter , @xmath71 , is close to @xmath72 .
this strongly suggests that the experimental data can be explained by the so - called simonin model @xcite , where the boundary conditions for the superconducting wavefunction , imposed by the interface to the w substrate and the surface , corresponds to an additional term in the ginzburg - landau free energy .
this leads to a reduction in @xmath1 for thin superconducting films , resulting in @xmath73 here , the critical thickness @xmath74 can be interpreted as a threshold thickness , which is required for a la film to develop superconductivity , if the simonin model holds for such very thin films . keeping @xmath75 and @xmath74 as free parameters , eq .
[ eq : t_c_vs_d ] is fitted to the experimental data [ black solid line in fig .
[ fig : figure_3](b ) ] .
the fitted value for the critical thickness is ( about 25 ml in the [ 0001 ] direction ) .
note that , although no superconducting gap is observed for @xmath76 nm under our experimental conditions , it can not be ruled out that these islands have a finite @xmath1 which is below the measurement temperature of 1.2 k. an expected value for @xmath74 can be determined via @xmath77 $ ] @xcite from literature values for the inverse thomas - fermi screening length @xmath78 nm@xmath79 , and for the product of the electronic density of states at the fermi level and the electron - phonon coupling potential @xmath80 @xcite , giving @xmath81 , which is close to the experimental result .
the relatively large value of @xmath82 , compared to other superconducting materials such as pb , thus originates from the comparatively small fermi velocity @xcite of la .
the value for the bulk critical temperature extracted from the fit to the simonin model is .
the aforementioned relation @xmath83 then yields a bulk energy gap of @xmath84 mev , which coincides with the value extrapolated from fig .
[ fig : figure_3](a ) . as a main result of this work ,
both @xmath75 and @xmath85 are almost 40% higher in comparison to reported bulk values @xcite .
this result is surprising , since in previous experimental studies of the thickness - dependent @xmath60 in other materials , which also revealed a linear dependence in agreement with the simonin model , the extrapolated values usually are in agreement with the respective bulk values @xcite . in the following ,
we discuss the origin of the large @xmath1 value .
the effect of strain on the la islands can be neglected in the thickness dependency @xmath60 , even though la is a relatively soft metal : for gd / w(110 ) , which has a similar lattice constant to la / w(110 ) , only the monolayer and bilayer have slightly enhanced lattice constants of 2% and 0.3% @xcite , while thicker films are already relaxed to the bulk lattice parameter . moreover
, the residual strain is tensile in nature and would lead to a reduction in @xmath1 @xcite .
therefore , we can rule out that the experimentally observed increase in @xmath85 and @xmath75 is related to strain effects .
-like surface state that forms on la(0001 ) .
the curves are normalized such that the differential conductance at coincides .
the negative @xmath19 values originate from an interplay between the strongly peaked sample dos and a tip dos with negative slope @xcite .
( b ) quantitative analysis of fwhm @xmath87 and intensity @xmath88 of the surface state extracted from lorentzian fits to sts spectra like shown in ( a ) ( appendix d).,scaledwidth=80.0% ] since impurities are known to play a crucial role for superconductivity , we finally investigate the effect of purity of the sample on @xmath56 and @xmath60 by an investigation of the @xmath89 generation samples , where stm topographs revealed residual surface contamination . the surface contamination is quantified by the intensity and width of the @xmath90-like surface state which forms on the lanthanide ( 0001 ) surfaces @xcite and is very sensitive to adsorbates @xcite .
the spectra in fig .
[ fig : figure_4](a ) indicate a well developed surface state by a strong resonance at @xmath91 v in the @xmath57 generation samples , while the surface state is partly quenched in the @xmath89 generation samples .
this finding is supported by a quantitative analysis ( appendix d ) of the full width at half maximum ( fwhm ) @xmath87 and intensities @xmath88 of the resonances measured on all investigated samples [ fig .
[ fig : figure_4](b ) ] . for the @xmath57 generation samples
, @xmath87 has a value comparable to or even lower than the reported intrinsic lifetime broadening due to electron - electron and electron - phonon scattering for very clean samples @xcite ( appendix d ) .
in contrast , on the @xmath89 generation samples , the surface state lifetime is further reduced by defect scattering as revealed by a roughly doubled @xmath87 .
we can therefore conclude that the surface purity of the @xmath89 generation samples is strongly reduced with respect to that of the @xmath57 generation samples . as shown in fig .
[ fig : figure_3](a , b ) , the reduced purity of the @xmath89 generation samples is correlated with overall reduced values of @xmath56 and @xmath60 .
. [ eq : t_c_vs_d ] fitted to this experimental data results in and [ red straight line in fig .
[ fig : figure_3](b ) ] .
since @xmath92 enters the boundary condition for the suppression of the superconducting order parameter @xcite , the enhanced value of @xmath92 indicates a stronger decay of the superconducting order parameter at the dirtier surface boundary of the @xmath89 generation samples .
these results suggest that the superior sample quality of the @xmath57 generation samples with respect to previous studies @xcite is responsible for the enhanced values of @xmath85 and @xmath75 of these samples .
this result is astonishing : while the surface state and the observed contamination are localized in the topmost atomic layer of the la islands , and thus , their superconductivity is expected to be influenced by the surface only within a region of @xmath93 , we observe a reduction of @xmath85 and @xmath75 even for the thickest islands with @xmath94 nm .
however , it is likely that not only the surface , but also the interior and the w - la interface of the @xmath89 generation la islands are dirtier than for the @xmath57 generation samples .
this might explain the correlation between @xmath1 and surface contamination for the thick islands with @xmath95 , where the surface is not expected to play a crucial role for the superconducting properties .
in summary , our observations reveal that the intrinsic bulk energy gap and critical temperature of dhcp la are 40% larger as compared to the values cited in the literature .
in addition , we quantitatively determined the thickness dependence of @xmath0 and @xmath1 , which is in good agreement with a theoretical model that considers the boundary conditions for the superconducting wavefunction .
we find that superconductivity does not persist below a critical thickness of 25 ml .
we consider the effects of sample purity , as correlated with modifications to the unoccupied surface state of la , and find , regardless of the thickness , that superconductivity is reduced at increased surface contamination .
our results suggest , that a superior purity of the samples investigated here explains why we observe an enhancement of @xmath1 as compared to previous reports .
this highlights the challenge in the investigation of the superconducting properties of the notoriously reactive lanthanides .
we acknowledge financial support from the erc advanced grant `` astonish '' , and from the dfg via graduiertenkolleg 1286 .
acknowledges project no .
kh324/1 - 1 from the emmy - noether - program of the dfg .
we thank a. kamlapure for fruitful discussions .
the energy gap of lanthanum was investigated by sts with superconducting nb tips and normal metal tungsten and ptir tips . a lock - in technique
gives direct access to @xmath19 .
there , the finite modulation voltage @xmath17 ( rms value ) limits the energy resolution and broadens the coherence peaks , which adds to the usual thermal broadening . in order to determine the superconducting energy gap @xmath2 and lifetime broadening parameter @xmath47 ,
the differential conductance is calculated numerically in analogy to the working principle of the lock - in amplifier @xcite : @xmath96 in this formula , @xmath97 is the tunneling current , defined as @xmath98\ , \mathrm d e\ .
\label{iv_sc}\ ] ] @xmath99 and @xmath100 are the densities of states ( dos ) of the two tunneling electrodes , and @xmath101 is the fermi function . in the simplest case , where the tunneling junction consists of a superconducting and a normal metal electrode ( i.e. , superconducting la island and normal metal tip or superconducting tip and normal metal wetting layer ) , the dos of the superconducting electrode , defined by eq . 1 of the main paper , is used as @xmath99 .
furthermore , we assume that the dos @xmath100 of the normal metal electrode is constant on the relevant scale of @xmath102 . in that case , a fitting of eq .
[ didv_sc_2 ] to the experimental data , with only @xmath102 and @xmath103 as free parameters , yields these quantities with good accuracy . in sts measurements with two superconducting electrodes ( i.e. , superconducting la island and superconducting tip )
, @xmath99 is the same as before .
we assume a similar dos for the second superconducting electrode , @xmath100 , with its normal metal dos @xmath104 .
in addition , we consider the parameters @xmath105 , @xmath106 , and the applied voltage : @xmath107 ^ 2-\delta_2 ^ 2}}\right ) \label{eq : rho_2}\ ] ] as @xmath102 and @xmath103 are already known from the characterization experiment we performed for each tip on the wetting layer , fitting eq .
[ didv_sc_2 ] to the experimental data yields @xmath105 and @xmath106 with good accuracy . while most sts studies of superconductors fit experimental @xmath19-curves by using only eq .
[ iv_sc ] , the approach introduced by eq .
[ didv_sc_2 ] naturally considers a finite modulation voltage .
moreover , a possible electronic noise added to the bias voltage can be described by this approach when using an increased effective modulation voltage @xmath108 .
this enables to decouple several effects that broaden the @xmath19-curves , and to determine the `` intrinsic '' lifetime broadening parameter @xmath47 for tip and sample .
mv , @xmath109 k ) of a 137 nm thick lanthanum island taken with a superconducting nb tip ( @xmath110 mev , ) .
( b ) @xmath19-curves taken at increasing separations from the rim , as depicted by the colored dots in ( a ) ( @xmath20 pa , @xmath21 mv , @xmath111 mv ) .
the spectra are each shifted by 5.5 a.u . for better clarity .
numerical calculations ( dotted curves ) , obtained with @xmath109 k and @xmath39 , reveal the very same parameters , @xmath112 mev and @xmath113 mev , for all experimental @xmath19-curves .
slight deviations at positive voltages can be ascribed to an insufficient stabilization of the tunneling junction .
the dips in @xmath19 at voltages above the coherence peaks originate from the nb tip and are not related to the lanthanum island.,scaledwidth=80.0% ]
a possible lateral variation of the superconducting properties of lanthanum islands is inspected in a control experiment on a particular 137 nm thick island [ fig . [
fig : lateral_superconductivity](a , b ) ] .
@xmath19-spectra were taken at different positions on the island [ colored dots in fig .
[ fig : lateral_superconductivity](a ) ] .
even the blue point is 58 nm away from the rim , which is clearly above the coherence length .
the @xmath19-spectra do not show any lateral variation on the relevant energy scale up to [ fig .
[ fig : lateral_superconductivity](b ) ]
. hence , the superconducting properties of the lanthanum islands , @xmath7 and @xmath48 , are only a function of thickness @xmath26 and temperature @xmath55 .
in the main paper , the following relation between @xmath60 and @xmath114 is given @xcite : @xmath115 in order to obtain @xmath60 from the experimental @xmath116 , an understanding of the temperature dependence of @xmath116 is required .
for the intermediate - coupling superconductor lanthanum @xcite , the relations between @xmath116 and @xmath60 are given by @xmath117 } \label{eq : tc}\ ] ] @xmath118 here , @xmath119 .
[ eq : fraction ] is used to determine @xmath60 as a function of @xmath114 [ dashed line in fig .
[ fig : figure_s_tcvsdelta ] ] .
the dependence of @xmath60 as a function of @xmath116 is calculated by a numerical integration of eq .
[ eq : delta_t ] using eq .
[ eq : tc ] in order to relate @xmath120 to @xmath60 . from the experimental values of @xmath56 , for @xmath121 ,
@xmath122 k , and @xmath123 k.,scaledwidth=80.0% ] in fig .
[ fig : figure_s_tcvsdelta ] , @xmath60 is shown as a function of @xmath124 ( green curve ) and @xmath125 ( blue curve ) . concerning @xmath9 k , in the range of the relevant data , @xmath126 @xmath127ev ,
there is basically no deviation between @xmath114 and @xmath116 , hence @xmath60 is immediately given by @xmath128 .
in contrast , @xmath129 k leads to a significant deviation between @xmath61 and @xmath56 .
the calculated relations are used to determine @xmath60 for all measured @xmath56 , as shown in the main paper .
as described in the main text , the @xmath90-like surface state which forms on la(0001 ) is sensitive to surface contamination ( fig . 4 ) .
in the following , the reduction of intensity and enhancement of broadening is quantitatively determined from the experimental sts curves of the two sample generations .
this enables to relate these values to the surface state lifetime and quality of la films for comparison with previous experimental studies @xcite . in order to describe the differential conductance @xmath130 analytically
, a simple model in analogy to @xcite is introduced . neglecting any voltage dependence of the transmission coefficient @xmath131 , and assuming a constant electronic density of states of the tip , an expression for @xmath19
is given as @xmath132 here , @xmath133 is the density of states of the sample , @xmath134 denotes the differentiation of the fermi function with respect to @xmath18 .
the transmission coefficient @xmath135 is rather constant .
the ( negative ) effective mass of the la(0001 ) surface state was determined by _ ab initio _ calculations as @xmath136 @xcite . here
, this rather weak dispersion is neglected completely , and @xmath137 is first approximated by a @xmath138-function @xmath139 , where @xmath140 equals the band maximum at @xmath141 .
now , the finite lifetime of the surface state @xmath142 is taken into account .
the lifetime broadening @xmath143 consists of electron - electron ( @xmath144 ) , electron - phonon ( @xmath145 ) , and defect scattering ( @xmath146 ) , that all add up to the overall broadening @xmath147 .
the energy dependence of these different scattering channels is neglected for simplicity .
therefore , the @xmath138-function needs to be replaced by a lorentzian @xcite , which leads to @xmath148 @xmath88 is a factor of proportionality , which is called intensity in the following .
the width ( fwhm ) of the differentiated fermi function @xmath134 is about @xmath149 mev at @xmath9 k , which is orders of magnitude smaller compared to the energy scale of the surface state .
therefore , @xmath134 is replaced by a @xmath138-function @xmath150 , resulting in @xmath151 when fitting this formula to the experimental @xmath19-curves , there are three free parameters : intensity @xmath88 , peak maximum @xmath152 , and width ( fwhm ) @xmath87 . before fitting ,
the individual sts curves were normalized to have the same differential conductance at @xmath153 v. in addition , a constant offset was subtracted , which reflects tunneling into bulk states . as depicted by the fitting result for two exemplary measured curves in fig .
[ fig : s1](a , b ) , this procedure leads to a good fit of the experimental data . ) to experimental @xmath19-curves .
( a ) @xmath19-curve ( dots ) taken on a 34 nm thick la island of the 1@xmath12 sample generation .
fit results ( straight line ) : , @xmath154 mv , @xmath155 mev .
( b ) @xmath19-curve ( dots ) on a 129 nm thick island of the 2@xmath13 sample generation .
fit results ( straight line ) : @xmath156 a , , @xmath157 mev .
( c , d ) compilation of fit results including additional measurements .
data from 1@xmath12 ( 2@xmath13 ) generation samples are marked by green ( red ) dots .
the particular results shown in ( a , b ) are marked by colored circles .
( c ) width ( fwhm ) @xmath87 vs. island thickness .
( d ) @xmath87 vs. intensity @xmath88 .
( a - d ) tunneling parameters : @xmath23 pa at @xmath24 v , , @xmath9 k.,scaledwidth=80.0% ] the fit results for all measured @xmath19-spectra are shown in fig [ fig : s1](c , d ) color - coded according to the sample generation .
@xmath87 is found to be almost independent of the island thickness .
even the thinnest films have a well - pronounced surface state , which is in agreement with previous studies @xcite .
the 1@xmath12 sample generation always exhibits a very narrow surface state peak with a maximum at @xmath158 mv and a width given by @xmath159 mev , corresponding to a lifetime of @xmath160 fs . in a previous analysis of the width of the experimental surface state peak , which even considered energy - dependent lifetime broadening , a value of @xmath161 mev
was reported @xcite . in comparison with the result reported here for the 1@xmath12 sample generation
, this implies that no additional energy broadening arises due to impurity - induced scattering .
it is therefore assumed that the defect - induced broadening for this very clean sample generation can be completely neglected , and hence @xmath162 is the `` intrinsic '' lifetime broadening . in summary , this shows that la(0001 ) surfaces of the 1@xmath12 sample generation are of the same or even better quality compared to previous studies @xcite .
the sts spectra taken on the la islands of the 2@xmath13 sample generation exhibit surface state peaks with a reduced intensity and a larger width [ fig .
[ fig : s1](b - d ) ] .
therefore , the modified @xmath163 must be affected by a defect - induced contribution @xmath164 .
this quantity is about twice as large as the `` intrinsic '' broadening @xmath87 .
the influence of surface contamination on superconductivity is analyzed in the main part of the paper .
10 shen l y l 1972 _ aip conf . proc . _ * 4 * 31 lou l and tomasch w 1972 _ phys . rev . lett . _
* 29 * 858 whl h , eichler a and wittig j 1973 _ phys .
lett . _ * 31 * 1393 levinstein h j , chirba v g and kunzler j e 1967 _ phys .
lett . _ * 24a * 362 de lozanne a l , elrod s a and quate c f 1985 _ phys .
lett . _ * 54 * 2433 eom d , qin s , chou m - y and shih c k 2006 _ phys . rev
. lett . _ * 96 * 027005 zhang r _ et al _ 2010 _ nature phys . _ * 6 * 104 zer m m , jia y , zhang z , thompson j r and weitering h h 2007 _ science _ * 316 * 1594 buzea c and robbie k 2005 _ supercond .
technol . _ * 18 * r1 johnson d and finnemore d k 1967 _ phys
* 158 * 376 maple m b , wittig j and kim k s 1969 _ phys .
* 23 * 1375 pan p h , finnemore d k , bevolo a j , shanks h r , beaudry b j , schmidt f a and danielson g c 1980 _ phys .
b _ * 21 * 2809 nixon l w , papaconstantopoulos d a and mehl m j 2008 _ phys . rev .
b _ * 78 * 214510 baci s , ttnc h m t , duman s and srivastava g p 2010 _ phys . rev .
b _ * 81 * 144507 simonin j 1986 _ phys . rev .
b _ * 33 * 7830 specs surface nano analysis gmbh , berlin , germany bode m , krause s , berbil - bautista l , heinze s and wiesendanger r 2007 _ surf . sci . _ * 601 * 3308 chempur gmbh , karlsruhe , germany ; mateck gmbh , jlich , germany wiebe j , wachowiak a , meier f and wiesendanger r 2004 _ rev .
instrum . _ * 75 * 4871 wegner d , bauer a , koroteev y m , bihlmayer g , chulkov e v , echenique p m and kaindl g 2006 _ phys . rev .
b _ * 73 * 115403 ; wegner d , bauer a , kaindl g 2007 _ phys . rev .
b _ * 76 * 113410 ; wegner d 2004 phd thesis freie
universitt berlin dynes r c , narayanamurti v and garno j p 1978 _ phys .
lett . _ * 41 * 1509 dickey j and paskin a 1968 _ phys .
* 21 * 1441 brun c , hong i - p , patthey f , sklyadneva i y , heid r , echenique p m , bohnen k p , chulkov e v and schneider w - d 2009 phys
* 207002 fbrega l , camn a , fernndez - martnez i , ses j , parra - borderas m , gil o , gonzlez - arrabal r , costa - krmer j l and briones f 2011 _ supercond .
technol . _
* 24 * 075014 song c
- l , wang y - l , jiang y - p , li z , wang l , he k , chen x , ma x - c and xue q - k 2011 _ phys . rev .
b _ * 84 * 020503 nepijko s , getzlaff m , pascal r , zarnitz c , bode m and wiesendanger r 2000 _ surf .
sci . _ * 466 * 89 getzlaff m , bode m , pascal r and wiesendanger r 1999 _ phys .
b _ * 59 * 8195 | the thickness dependence of the superconducting energy gap @xmath0 of double hexagonally close packed ( dhcp ) lanthanum islands grown on w(110 ) is studied by scanning tunneling spectroscopy , from the bulk to the thin film limit .
superconductivity is suppressed by the boundary conditions for the superconducting wavefunction at the surface and w / la interface , leading to a linear decrease of the critical temperature @xmath1 as a function of the inverse film thickness . for thick , bulk - like films ,
@xmath0 and @xmath1 are 40% larger as compared to literature values of dhcp la measured by other techniques .
this finding is reconciled by examining the effects of surface contamination as probed by modifications of the surface state , suggesting that the large @xmath1 originates in the superior purity of the samples investigated here .
_ keywords _ : lanthanum , superconductivity , scanning tunneling spectroscopy |
let @xmath5 be an open and bounded set . for a given function @xmath6
we denote the gradient of @xmath7 as @xmath8 and its hessian @xmath9 .
the @xmath1-laplacian is the partial differential equation ( pde ) @xmath10 where `` @xmath11 '' is the tensor product between @xmath12-vectors and `` @xmath13 '' the frobenius inner product between matrices .
this problem is the prototypical example of a pde from calculus of variations in @xmath14 , arising as the analogue of the _ euler lagrange _ equation of the functional @xmath15 : = \norm{\d u}_{{\ensuremath{\ll^{\infty}}}(\w)}\ ] ] @xcite and as the ( weighted ) formal limit of the variational @xmath0-laplacian @xmath16 the @xmath0-laplacian is a divergence form problem and appropriate _
weak _ solutions to this problem are defined in terms of duality , or integration by parts . in passing to the limit ( @xmath17 ) the problem loses its divergence structure . in the nondivergence
setting we do not have access to the same stability framework as in the variational case and a different class of `` weak '' solution must be sought .
the correct concept to use is that of viscosity solutions ( * ? ? ?
* ; * ? ? ?
* c.f.).the main idea behind this solution concept is to pass derivatives to test functions through the maximum principle , that is , _ without _ using duality .
the design of numerical schemes to approximate this solution concept is limited , particularly in the finite element context , where the only provably convergent scheme is given in @xcite ( although it is inapplicable to the problem at hand ) . in the finite difference setting techniques have been developed @xcite and applied to this problem and also the associated eigenvalue problem @xcite .
in fact both in the finite difference and finite element setting the methods of convergence are based on the discrete monotonicity arguments of @xcite which is an extremely versatile framework .
other methods exist for the problem , for example in @xcite , the authors propose a biharmonic regularisation which yields convergence in the case ( [ eq : inflap ] ) admits a strong solution . in @xcite
the author proposed an @xmath4-adaptive finite element scheme based on a residual type error indicator .
the underlying scheme was based on the method derived in @xcite for fully nonlinear pdes . in this note
we examine a different route .
we will review and use the known theory used in the derivation of the @xmath1-laplacian ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
where a _ @xmath0-limiting _ process is employed to derive ( [ eq : inflap ] ) .
we study how well galerkin approximations of ( [ eq : introplap ] ) approximate the solutions of ( [ eq : inflap ] ) and show that by forming an appropriate limit we are able to select candidates for numerical approximation along a `` good '' sequence of solutions .
this is due to the equivalence of weak and viscosity solutions to ( [ eq : introplap ] ) @xcite . to be very clear about where the novelty lies in this presentation , the techniques we use are not new .
we are summarising existing tools from two fields , one set from pde theory and the other from numerical analysis . while both sets of results are relatively standard in their own field , to the authors knowledge , they have yet to be combined in this fashion .
we use this exposition to conduct some numerical experiments which demonstrate the rate of convergence both in terms of @xmath0 approximation approximation we use here should not be confused with @xmath0-adaptivity which is local polynomial enrichment of the underlying discrete function space . ] and @xmath4 approximation .
these results illustrate that for practical purposes , as one would expect , the approximation of @xmath0-harmonic functions for large @xmath0 gives good resolution of @xmath1-harmonic functions .
the numerical approximation of @xmath0-harmonic functions is by now quite standard in finite element literature , see for example @xcite .
there has been a lot of activity in the area since then however . in particular , the quasi - norm introduced in @xcite
gave significant insight in the numerical analysis of this problem and spawned much subsequent research for which @xcite form an inexhaustive list . while it is not the focus of this work , we are interested in this approach as it allows us to extend quite simply and reliably into the vectorial case . when moving from scalar to vectorial calculus of variations in @xmath14 viscosity solutions
are no longer applicable .
one notion of solution currently being investigated is @xmath18-solutions @xcite which is based on concepts of young measures .
the ultimate goal of this line of research is the construction of reliable numerical schemes which allow for various conjectures to be made as to the nature of solutions and even what the correct solution concepts in the vectorial case are @xcite .
the rest of the paper is set out as follows : in [ sec : plap ] we formalise notation and begin exploring some of the properties of the @xmath0-laplacian . in particular , we recall that the notion of weak and viscosity solutions to this problem coincide , allowing the passage to the limit @xmath17 . in [ sec : fem ] we describe a conforming discretisation of the @xmath0-laplacian and its properties .
we show that the method converges to the weak solution for fixed @xmath0 .
numerical experiments are given in
[ sec : numerics ] illustrating the behaviour of numerical approximations to this problem .
in this section we describe how @xmath1-harmonic functions can be approximated using @xmath0-harmonic functions .
we give a brief introduction to the @xmath0laplacian problem , beginning by introducing the sobolev spaces @xcite @xmath19 which are equipped with the following norms and semi - norms : @xmath20 where @xmath21 is a multi - index , @xmath22 and derivatives @xmath23 are understood in a weak sense .
we pay particular attention to the case @xmath24 and @xmath25 for a prescribed function @xmath26 .
let @xmath27 be the _
lagrangian_. we will let @xmath28 { \cj[\ \cdot \ ; p ] } \phi { { \ensuremath{\ww^{1,p}_g}}(\w ) } { \cj[\phi ; p ] : = \int_\w l(\geovec x , \phi , \d \phi ) \d \geovec x } { \reals}\ ] ] be known as the _ action functional_. for the @xmath0laplacian the action functional is given as @xmath29 : = \int_\w l(\geovec x , u , \d u ) = \int_\w \norm{\d u}^p . \quad
\footnote{typically $ l(\geovec x , u , \d u ) = \tfrac{1}{p } \norm{\d u}^p$. note here the rescaling of $ l$ has no effect on the resultant euler -- lagrange equations as to $ l$ is independent of $ u$.}\ ] ] we then look to find a minimiser over the space @xmath30 , that is , to find @xmath31 such that @xmath29 = \min_{v\in{\ensuremath{\ww^{1,p}_g}}(\w ) } \cj[v;p].\ ] ] if we assume temporarily that we have access to a smooth minimiser , @xmath32 , then , given that the lagrangian is of first order , we have that the euler
lagrange equations are ( in general ) second order .
the euler
lagrange equations for this problem are @xmath33 : = \div\qp{\norm{\d u}^{p-2}\d u } = 0.\ ] ] note that , for @xmath34 , the problem coincides with the poisson problem @xmath35 . in general , the @xmath0-laplace problem is to find @xmath7 such that @xmath36 the problem ( [ eq : plap ] ) is associated to a weak formulation , set @xmath37 to be a semilinear form , then @xmath38 is a _ weak solution _ of ( [ eq : plap ] ) if it satisfies @xmath39 [ pro : existence - uniqueness ] there exists a unique weak solution to ( [ eq : plap ] ) .
the proof is standard and can be found in ( * ? ? ?
* thm 5.3.1 ) for example .
it is based on the strict convexity of @xmath40 $ ] , yielding uniqueness , together with appropriate growth conditions for existence .
[ def : visc - soln ] a function @xmath41 is a viscosity sub - solution of a general second order pde @xmath42 at a point @xmath43 if for any @xmath44 satisfying @xmath45 , and touching @xmath7 from above , that is @xmath46 in a neighbourhood of @xmath47 , we have @xmath48 similarly , a function @xmath41 is a viscosity super - solution of ( [ eq : fnl ] ) at a point @xmath49 if for any @xmath44 satisfying @xmath50 and touches @xmath7 from below we have @xmath51 the function @xmath7 is a viscosity solution of ( [ eq : fnl ] ) in @xmath52 if it is a viscosity super and sub - solution at any @xmath53 .
[ the : weak - visc ] let @xmath26 and suppose @xmath54 is fixed . then weak solutions of @xmath55 are viscosity solutions of @xmath56 we begin by noting that by expanding the derivatives @xmath57 is a renormalisation of ( [ eq : normalised ] ) .
the two formulations ( [ eq : norm ] ) and ( [ eq : normalised ] ) of the @xmath0-laplacian are equivalent in the viscosity sense , see for example ( * ? ? ?
* lemma 3 ) . it remains to show that weak solutions of ( [ eq : p - lap ] ) are viscosity solutions of ( [ eq : norm ] ) .
as @xmath26 we have @xmath58 < \infty.\ ] ] since @xmath7 solves ( [ eq : p - lap ] ) weakly , it minimises the functional @xmath59 $ ] and hence the minimiser must be of finite energy . in view of the existence and uniqueness of the minimisation problem from proposition [ pro : existence - uniqueness ] and morrey s inequality , we may infer @xmath60 and hence @xmath41 .
now assume by contradiction that @xmath7 is not a viscosity subsolution of @xmath61 then by definition [ def : visc - soln ] we can find an @xmath49 , a @xmath62 and an @xmath63 such that @xmath64 on @xmath65 , @xmath66 and @xmath67 for some @xmath68 . since @xmath69 has a strict maximum at @xmath47 we may find an @xmath70 such that @xmath71 hence @xmath72 as @xmath73 on @xmath74 .
now by the convexity of the lagrangian @xmath75 we have @xmath76 hence @xmath77 - \cj[\psi-\epsilon ; p]\ ] ] and we see @xmath29 \leq \cj[\psi - \epsilon ; p].\ ] ] this means that @xmath78 and we have a contradiction .
the complete proof in full generality for convex minimisation problems can be found in @xcite .
see also @xcite where the author extends the arguments of @xcite to singular pdes .
the converse to theorem [ the : weak - visc ] is also true , thus weak and viscosity solutions are an equivalent concept for the @xmath0-laplacian and its evolutionary relative .
this has been shown in @xcite . [
the : ptoinf ] let @xmath79 denote a sequence of weak / viscosity solutions to the @xmath0-laplacian then there exists a subsequence such that as @xmath17 that sequence converges to a candidate @xmath1-harmonic function @xmath80 , that is , @xmath81 we denote @xmath82 as the weak solution of ( [ eq : plap ] ) . in view of proposition
[ pro : existence - uniqueness ] we know that @xmath83 minimises the energy functional @xmath84 = \int_\w \norm{\d u_p}^p.\ ] ] hence in particular @xmath84 \leq \cj[g;p],\ ] ] where @xmath85 is the associated boundary data to ( [ eq : plap ] ) . using this
fact @xmath86 \leq \cj[g;p ] = \norm{\d g}_{{\ensuremath{\ll^{p}}}(\w)}^p,\ ] ] and we may infer that @xmath87 now fix a @xmath88 and take @xmath89 , then using hlders inequality @xmath90 with @xmath91 and @xmath92 such that @xmath93 . hence @xmath94 and we see @xmath95 using the triangle inequality @xmath96 in view of the poincar inequality . using the triangle inequality again we have @xmath97 by ( [ eq : limitpf2 ] ) . hence using ( [ eq : limitpf1 ] )
@xmath98 this means that for any @xmath88 we have uniformly that @xmath99 hence , in view of weak compactness , we may extract a subsequence @xmath100 and a function @xmath101 such that for any @xmath102 @xmath103 and @xmath104 taking the limit @xmath105 we have @xmath106 and thus @xmath107 .
the result follows from morrey s inequality , concluding the proof .
we note that an alternative sequence of solutions is given in @xcite , where rather than studying the limit of the @xmath0-dirichlet functional , the authors propose @xmath108 = \int_\w \exp\qp{{p}\norm{\d u}^2}.\ ] ] this functional may have some merit over the @xmath0-dirichlet functional since the euler lagrange equations @xmath109 yield a clearer relation between @xmath110 and @xmath111 .
we will not explore this issue further in this work .
the `` candidate '' @xmath1-harmonic function @xmath112 is the unique viscosity solution to the @xmath1-laplacian ( [ eq : inflap ] ) .
the proof is detailed in @xcite .
roughly , existence of @xmath112 has been shown in theorem [ the : ptoinf ] .
for uniqueness one must prove and make use of the maximum principle for ( [ eq : inflap ] ) .
note also the result of @xcite where the authors use difference equations to prove the same result in a simpler fashion .
in this section we describe a conforming finite element discretisation of the @xmath0-laplacian .
let @xmath113 be a conforming triangulation of @xmath52 , namely , @xmath113 is a finite family of sets such that 1 .
@xmath114 implies @xmath115 is an open simplex ( segment for @xmath116 , triangle for @xmath117 , tetrahedron for @xmath118 ) , 2 . for any @xmath119 we have
that @xmath120 is a full lower - dimensional simplex ( i.e. , it is either @xmath121 , a vertex , an edge , a face , or the whole of @xmath122 and @xmath123 ) of both @xmath122 and @xmath123 and 3 .
@xmath124 .
the shape regularity constant of @xmath113 is defined as the number @xmath125 where @xmath126 is the radius of the largest ball contained inside @xmath115 and @xmath127 is the diameter of @xmath115 .
an indexed family of triangulations @xmath128 is called _ shape regular _
if @xmath129 further , we define @xmath130 to be the piecewise constant _ meshsize function _ of @xmath113 given by @xmath131 a mesh is called quasiuniform when there exists a positive constant @xmath132 such that @xmath133 . in what follows
we shall assume that all triangulations are shape - regular and quasiuniform although the results may be extendable even in the non - quasiuniform case using techniques developed in @xcite .
we let @xmath134 be the skeleton ( set of common interfaces ) of the triangulation @xmath113 and say @xmath135 if @xmath136 is on the interior of @xmath52 and @xmath137 if @xmath136 lies on the boundary @xmath138 and set @xmath139 to be the diameter of @xmath136 .
further , we define the broken gradient @xmath140 , laplacian @xmath141 and hessian @xmath142 to be defined element - wise by @xmath143 , @xmath144 , @xmath145 for all @xmath146 , respectively , for respectively smooth functions on the interior of @xmath115 , we let @xmath147 denote the space of piecewise polynomials of degree @xmath148 over the triangulation @xmath113,@xmath149 and introduce the _ finite element space _
@xmath150 to be the usual space of continuous piecewise polynomial functions of degree @xmath148 .
a finite element sequence @xmath151 is a sequence of discrete objects indexed by the mesh parameter , @xmath4 , and individually represented on a particular finite element space @xmath152 , which itself has a discretisation parameter @xmath4 , that is @xmath153 .
[ eq : l2proj ] the @xmath154 projection operator , @xmath155 is defined for @xmath156 such that @xmath157 it is well known that this operator satisfies the following approximation properties for @xmath158 @xmath159 we consider the galerkin discretisation of ( [ eq : plap ] ) , to find @xmath160 with @xmath161 such that @xmath162 there exists a unique solution of ( [ eq : plapdis ] ) .
the proof is standard and , in fact , equivalent to that of the smooth case , as in ( * ? ? ?
* thm 5.3.1 ) .
[ the : convergence - of - scheme ] let @xmath163 be the finite element sequence generated by solving ( [ eq : plapdis ] ) and @xmath83 , the weak solution of ( [ eq : p - lap ] ) , then for fixed @xmath0 we have that @xmath164 we begin by noting the discrete weak formulation ( [ eq : plapdis ] ) is equivalent to the minimisation problem : find @xmath165 such that @xmath166 = \min_{v\in\fes } \cj[v ; p].\ ] ] using this , we immediately have @xmath167 \leq \cj[p g ; p ] \leq \norm{\d \qp{p g}}_{{\ensuremath{\ll^{p}}}(\w)}^p.\ ] ] in view of the stability of the @xmath168 projection in @xmath169 @xcite we have @xmath170 uniformly in @xmath4 .
hence by weak compactness there exists a ( weak ) limit to the finite element sequence , which we will call @xmath171 . due to the weak semicontinuity of @xmath40
$ ] we have @xmath172 \leq \cj[u ; p].\ ] ] in addition , in view of the approximation properties of @xmath173 given in [ eq : l2proj ] we have for any @xmath174 that @xmath175 = \liminf_{h\to 0 } \cj[p_k v ; p].\ ] ] using the fact that @xmath176 is a discrete minimiser of ( [ eq : dismin ] ) we have @xmath172 \leq \cj[u ; p ] \leq \cj[p_k v ; p],\ ] ] whence sending @xmath3 we see @xmath177 \leq \cj[v;p].\ ] ] now , as @xmath178 was generic we may use density arguments and that @xmath83 was the unique minimiser to conclude @xmath179 , concluding the proof . in view of theorem [ the : weak - visc ] the discrete scheme converges to viscosity solutions of the @xmath0-laplacian .
[ eq : lem - dis - p - to - inf ] let @xmath180 solve the discrete problem ( [ eq : plapdis ] ) , then for fixed @xmath4 along a subsequence we have @xmath181 .
the proof follows similarly to theorem [ the : ptoinf ] .
since @xmath180 is the galerkin solution to ( [ eq : plapdis ] ) , it minimises @xmath182 $ ] over @xmath152 .
hence we know @xmath183 in view of the stability of @xmath173 in @xmath169 .
in addition , analogously to ( [ eq : limitpf3])([eq : limitpf4 ] ) we may find a constant such that @xmath184 allowing the extraction of a subsequence @xmath185 and a limit @xmath186 such that for @xmath88 @xmath187 the rest of the proof parallels that of theorem [ the : ptoinf ] .
up to this point we have shown the green ( solid ) lines on the following diagram ( 1)2 ( 1)4 ( 2)3 ( 4)(3 ) node [ above ] thm [ the : convergence - of - scheme ] ; ( 4)(3 ) node [ below ] @xmath188 ; ( 4)(3 ) ; ( 3)(2 ) node [ right ] @xmath189 ; ( 3)(2 ) node [ left ] thm [ the : ptoinf ] ; ( 3)(2 ) ; ( 4)(1 ) node [ left ] @xmath189 ; ( 4)(1 ) node [ right ] lem [ eq : lem - dis - p - to - inf ] ; ( 4)(1 ) ; ( 4)(2 ) ; hold .
we would like to select a route for which we can pass the limits together , that is , we want to select an appropriate route for which the red ( dashed ) line is true .
[ the : convergence ] let @xmath180 be the galerkin solution of ( [ eq : plapdis ] ) and @xmath112 the unique viscosity solution of ( [ eq : inflap ] ) then along a subsequence @xmath190 the proof is a consequence of theorems [ the : ptoinf ] and [ the : convergence - of - scheme ] noting that along the same subsequence used in theorem [ the : ptoinf ] we have that @xmath191 and hence @xmath192 as @xmath2 and @xmath3 . an immediate consequence of theorem [ the : convergence ] and the previous arguments are that for @xmath193 with appropriate conditions ( convexity for example ) , finite element approximations to the @xmath0-functional @xmath29 = \norm{h(\cdot , u , \d u)}_{{\ensuremath{\ll^{p}}}(\w)}\ ] ] can be used as approximations to @xmath15 = \norm{h(\cdot , u
, \d u)}_{{\ensuremath{\ll^{\infty}}}(\w)}.\ ] ] all the above results can be extended into the discontinuous galerkin framework .
this is based on the discrete action functional @xmath194 : = \int_\w \norm{g(u)}^p + \int_\e h_e^{1-p } \norm{\jump{u}}^p,\ ] ] where @xmath195 where @xmath196 denotes the _ jump _ over an edge @xmath136 shared by neighbouring elements @xmath197 and @xmath198 and @xmath199 , the _ average _ of a quantity over an edge . using the results of @xcite discrete minimisers to ( [ eq : action - dg ] )
satisfy the equivalent weak convergence results to the conforming finite elements .
in this section we summarise numerical experiments validating the analysis done in previous sections .
the computation of @xmath0-harmonic functions is an extremely challenging problem in its own right . the class of nonlinearity in the problem results in the algebraic system , which ultimately yields the finite element solution , being extremely badly conditioned .
one method to tackle this class of problem is the use preconditioners based on descent algorithms @xcite . for extremely large @xmath0 , say @xmath200 this may be required , however for our purposes we restrict our attention to @xmath201 .
this yields sufficient accuracy for the results we want to illustrate . even tackling the case
@xmath202 is computationally tough .
our numerical approximation is based on a newton solver . as is well known ,
newton solvers require a sufficiently close initial guess to converge . for large @xmath0
a reasonable initial guess is given by numerically approximating the @xmath203-laplacian for @xmath204 sufficiently close to @xmath0 .
this leads to an iterative process in the generation of the initial guess .
we begin by approximating the viscosity solution derived by aronsson using separation of variables @xcite .
the function @xmath205 is a viscosity solution of the @xmath1-laplacian .
notice that this is a weighted version of the aronsson solution . we have chosen this as @xmath206 on the domains we consider to try to overcome the severe restrictions in computing @xmath0-harmonic functions with large @xmath0 . in this test
we take @xmath207 ^ 2 $ ] and triangulate with a criss - cross mesh .
this is so the singularity will not be aligned with the mesh .
we approximate the solution of the @xmath0-laplacian with boundary data given by ( [ eq : aronson ] ) for a variety of increasing @xmath0 .
examples of solutions are given in figure [ fig : aronson ] . in figure
[ fig : fixh ] we plot the error against @xmath0 for a various levels of mesh refinement . in table
[ table1 ] we demonstrate the convergence of the finite element approximations as @xmath3 .
the error of the finite element approximation to the @xmath0-laplacian compared to the viscosity solution to the @xmath1-laplacian for various @xmath0 .
the colours represent different mesh refinement levels .
the darker the colour , the more refined the mesh . in this experiment
the meshsize ranges from @xmath208 to @xmath209 notice as the mesh is refined , best approximation is achieved for higher and higher @xmath0 . ] .
[ table1 ] in this table we show the convergence of the finite element approximation @xmath180 to @xmath112 , a viscosity solution of ( [ eq : inflap ] ) , as the meshsize is decreased .
we study the @xmath14 error of the approximation , the associated convergence rate and give @xmath210 , the smallest such @xmath0 for which @xmath211 is attained . notice that as the mesh is refined , the critical value increases . [ cols="^,^,^,^",options="header " , ] | in this note we show that conforming galerkin approximations for @xmath0-harmonic functions tend to @xmath1-harmonic functions in the limit @xmath2 and @xmath3 , where @xmath4 denotes the galerkin discretisation parameter . |
more than half of the clusters of galaxies observed at x ray wavelengths exhibit evidence for cool gas in their cores ( edge , stewart , & fabian 1992 ) .
the cooling time of this gas is often short compared to cluster ages ; therefore , the gas cools at rates that are often very large , @xmath3 yr@xmath4 . in the absence of a heating mechanism to balance cooling , gravity and pressure from the hotter ,
surrounding gas will drive the cooling material into the core of the cluster .
if cooling flows are long lived phenomena , these rates imply that @xmath5@xmath6 of material would cool over the lifetime of the cluster .
determining the final state of this cooling material remains the fundamental problem concerning the cooling flow theory .
the most obvious repositories for the cooling gas are cold molecular and atomic clouds , and stars .
the central dominant galaxy in cooling flow clusters often have blue stellar colors in their central regions , which indicate that stars are currently being formed there ( mcnamara & oconnell 1992 ) .
however , the observed star formation rates are generally @xmath7 of @xmath8 .
therefore star formation can not account for the cooling material without appealing to unusual initial mass functions .
similarly , extensive efforts have been made to detect the accumulated cooled material either as gas at some temperature below the x ray emitting regime ( @xmath9 k ) .
gas is seen in cooling flows at @xmath10 k ( heckman et al .
1989 ; baum 1992 ) and in a few cases , as neutral atomic or molecular gas ( lazareff et al . 1989
; mirabel , sanders , & kazes 1989 ; mcnamara , bregman , & oconnell 1990 ; jaffe 1992 ; odea , baum , & gallimore 1994a ) .
dust is also seen in a number of cooling cores ( bregman , mcnamara , & oconnell 1990 ; wise et al .
1993 ) . in all cases , however , the detected levels of atomic and molecular gas are too low ( @xmath11@xmath12 ) to account for the cooled gas which would accumulate over the age of the cluster . the detection by white et al . ( 1991 ) of excess x ray absorption in a sample of cluster cooling flows was the first direct evidence for a sufficiently large mass of cold material .
using _ einstein _
sss ( solid state spectrometer ) spectra , these authors found that many cooling flow clusters exhibit significant levels of excess absorption over that expected from the galaxy with typical excess column densities of @xmath13 @xmath14 .
evidence for excess absorption in cooling flows has also been found in analyses of x - ray spectra taken with detectors on _ einstein _ , _ rosat _ , _ bbxrt _ , and _ asca _ ( lea , mushotzky , & holt 1982 ; miyaji 1991 ; allen et al .
1993 ; fabian et al .
1994 ; allen & fabian 1997 ) .
the excess absorption columns detected by white et al .
( 1991 ) were only observed in clusters with spectral evidence for cooling flows and show some evidence for a correlation between @xmath0 and @xmath15 .
this correlation and the observed spatial coincidence between excess absorption and cooling flows suggests that the absorbing material is intrinsic to the cluster and probably located within the cooling flow region . integrating these excess column densities over the area of the cooling region in cluster cores implies the presence of large quantities of cold absorbing material ( m@xmath16@xmath17 ) and may represent the first direct evidence for the large amounts of cooled material which current cooling flow models predict ( white et al .
1991 ) .
on the other hand , very extensive searches have been made to detect the excess absorber in emission or absorption at radio wavelengths in lines of h i or co and have not succeeded ( e.g. , mcnamara & jaffe 1993 ; antonucci & barvainis 1994 ; odea et al .
it is becoming difficult to understand how so much x ray absorbing gas could have escaped detection in other wavebands ( e.g. , voit & donahue 1995 ) .
often the _ rosat _ pspc spectra of cooling flows are inconsistent with large columns of excess foreground absorption ( e.g. , sarazin , wise , & markevitch 1998 ) , but are consistent with and may require large amounts of internal absorption in the cooling flow ( allen & fabian 1997 ) . for nearby clusters where the _ rosat _ or _ asca _ observations can resolve the cooling flow region , the excess absorption appears to be concentrated to the center of the cluster and cooling flow ( @xmath18 kpc ) ( allen et al .
1993 ; irwin & sarazin 1995 ; fabian et al .
1994 ; allen & fabian 1997 ) . in the standard data analysis of x ray spectra ( e.g. , in xspec or iraf / pros ) ,
it is conventional to treat absorbers as lying in the foreground of emission components .
this assumption allows the simple arithmetic combination of additive emission sources and multiplicative absorption components .
however , x ray observations suggest that the excess absorbing material in cluster cooling cores occupies the same spatial region as the x ray emitting gas ( allen et al . 1993 ;
irwin & sarazin 1995 ; allen & fabian 1997 ) .
certainly , one would expect such a spatial correlation if the absorber originated as cooling x ray gas . including the effects of absorbing material which is intermixed with the x ray emitting gas
is not easily treated within the framework of conventional x ray spectral modeling .
allen & fabian ( 1997 ) used de - projection techniques based on _ rosat _ pspc x - ray colors to attempt to determine the three dimensional distribution of the absorption . in order to provide more accurate models of the x ray spectra of cooling flows and to assess
the effect of intermixed absorbing material , we have calculated the emergent x ray properties for a set of inhomogeneous cooling flow models including the opacity due to accumulated cooled gas . for a given cooling flow model , we have self - consistently included the x ray absorption due to varying fractions of the total cooled material generated by the model .
the details of the input models and the numerical solution of the radiative transfer equation are described in
[ sec : calc ] . in
[ sec : results ] , we present x ray luminosities , spectra , integrated surface brightness profiles , and emission line equivalent widths , and spectral profiles for the models .
the implications of these results for the interpretation of existing x ray observations and their application to observations with _
axaf _ and astro - e are discussed in
[ sec : derive ] . in the final section ( [ sec :
conclude ] ) , we summarize our conclusions .
we demonstrate that such large quantities of absorbing material can have a significant impact on the observed x ray properties of cluster cooling flows .
as an input to the transfer calculations , we have selected a number of representative models for inhomogeneous cooling flows , some of which were presented previously in wise & sarazin ( 1993 ) .
these models were calculated using the numerical technique described in white & sarazin ( 1987 ) and a subset have been used previously by sarazin & graney ( 1991 ) to estimate the optical coronal line emission from cooling flows .
the optically thin x ray properties for some of these models have been discussed previously in wise & sarazin ( 1993 , hereafter paper i ) while the effects of opacity due to cooling gas were discussed in wise & sarazin ( 1999 , hereafter paper ii ) .
we note that the opacity of gas which is in the process of _ cooling _ below x ray emitting temperatures ( as opposed to the accumulated _ cooled _ material described in this paper ) was shown previously in paper ii to be insufficient to account for the observed excess absorption .
these cooling flow models have been discussed in detail in these papers and , in the interest of brevity , we will only briefly describe their relevant characteristics here . for a more detailed discussion ,
the reader is referred to these sources .
the set of models discussed in papers i and ii span a range of gas deposition including both homogeneous models as well as models with varying levels of inhomogeneity .
homogeneous models , by definition , exhibit a single density and temperature at any given radius .
it is important to realize that homogeneous cooling flow models can not reproduce the observed properties of the excess absorption . in a homogeneous model
, all of the gas cools below x ray emitting temperatures just within the sonic radius , @xmath19 , which is the point where the inflow velocity exceeds the sound speed in the gas ( see 2.1 of paper i and references therein ) .
values for the sonic radii are typically quite small ( @xmath20 kpc ) ; therefore , any accumulated cooled material would be confined to the very innermost regions of the cooling flow .
in contrast , the spectral results of white et al .
( 1991 ) indicate that the absorbing material must cover a significant fraction of the cooling flow region ( @xmath21 kpc ) , and the spatially resolved spectral measurements of allen et al .
( 1993 ) and irwin & sarazin ( 1995 ) confirm this extended absorption .
consequently , in choosing models for these calculations , we have only considered inhomogeneous models which feature cooling material ( and by implication _ cooled _ material ) distributed over large spatial scales .
the models used in this paper are listed in table [ tab : models ] and assume a spherically symmetric , steady state inflow with only gravitational and gas pressure forces included .
the outer boundary is taken to be the `` cooling radius '' , @xmath22 , which corresponds to the point at which the instantaneous isobaric cooling time , @xmath23 , is equal to the age of the cluster . in all cases ,
the cluster age was taken to be @xmath24 years .
as described in paper i , the models are completely determined by three quantities : the total inflow rate @xmath8 ; the gas temperature @xmath25 ; and the mass deposition profile , @xmath26 , which describes how inhomogeneous gas cools below x ray emitting temperatures as a function of radius in the cluster . both @xmath8 and @xmath25
are evaluated at the cooling radius @xmath22 and values of @xmath27 or @xmath28 @xmath6 yr@xmath4 and @xmath29 k were used for all models . to specify the mass deposition profile , @xmath26 , two forms for the mass inflow rate were used . for most of the models , the simple two phase cooling flow model of white & sarazin ( 1987 )
was employed . in this prescription ,
the mass deposition profile is parameterized as @xmath30}{p(r ) } \label{eq_4:2}\ ] ] where @xmath23 is the instantaneous isobaric cooling time , @xmath31 is the gas density , @xmath32 is the gas temperature , and @xmath33 is the gas pressure . the total emission and cooling rate per unit volume in the gas is @xmath34 , where @xmath35 is the total emissivity coefficient . the parameter @xmath36 is referred to as the `` gas loss efficiency parameter . ''
the values of @xmath36 for each of the models are given in column 5 of table [ tab : models ] .
in addition to models using the white & sarazin mass deposition formula ( eq . [ eq_4:2 ] ) , one model was included which used a form for the mass deposition profile derived by various groups from deconvolutions of cluster x ray surface brightness distributions .
fabian et al . ( 1984 ) , stewart et al . ( 1984 ) , and thomas et al . (
1987 ) have argued on the basis of these observations that the inflow rate varies as @xmath37 . in the c300_8_fb model , this assumption was adopted , which implies a mass deposition profile of the form @xmath38 we note that this model is quite similar to a white & sarazin model with @xmath39 .
the various physical parameters for each of the models is given in table [ tab : models ] .
the models use a naming convention of c@xmath40_@xmath41_@xmath42 , where @xmath43 ( in units of m@xmath44 yr@xmath4 ) , @xmath45 ( in units of @xmath46 k ) , and @xmath47 . for the `` fabian et al . ''
style model with @xmath48 , @xmath49 `` fb . ''
three of the models were presented previously in sarazin & graney ( 1991 ) and paper i , and the second column in table [ tab : models ] gives their identifications in the nomenclature of those papers .
the remaining columns give the cooling rate @xmath50 and temperature @xmath25 at the cooling radius , the gas loss parameter , the cooling radius @xmath22 , and the sonic radius @xmath19 .
a value of @xmath51 indicates a fully subsonic solution .
the calculation of the x ray emission from the gas is discussed in considerable detail in papers i and ii . we included bound - bound , bound - free , free - free , and two - photon emission from ions of the elements h , he , c , n , o , ne , mg , si , s , ca , fe , and ni .
elemental abundances for these species were set at one half of the solar values as defined by meyer ( 1979 ) .
the emissivities were calculated under the usual `` coronal limit '' conditions , as described in paper i. of course , the one major exception to these assumptions was the inclusion of the opacity of the gas .
we assume that the cooled material is at temperatures below @xmath52 k , and produces no appreciable x ray emission . at the column densities of interest , fluorescent line emission by the cooled gas
is not expected to be significant . however , if the columns of cold material in the absorbing clouds were larger , fluorescent emission of fe k lines might be detectable ( churazov et al . 1998 ) .
in calculating the opacity in the cooling flow , we have included contributions from the ambient hot gas , the _ cooling _ hot gas , and the accumulated _ cooled _ material .
of these three components , the first two ( opacity due to ambient and cooling material ) were discussed in detail in paper ii and are treated in the same manner here .
processes contributing to the opacity of the hot gas include resonant line absorption , photoelectric absorption , and electron scattering .
as discussed in paper ii , processes such as resonant scattering in the x ray lines or electron scattering merely redistribute photons within the cluster and produce no net loss in the spatially integrated spectrum .
we have also shown previously that the photoelectric opacity in an inhomogeneous cooling flow model due to ambient and cooling intracluster gas is small with @xmath53 ( see paper ii ) .
thus , the opacity from these gas phases can not account for the observed excess soft x
ray absorption in cooling flows .
in the present paper , we include a new source of opacity associated with the accumulated cooled gas .
we assume that this material is relatively cold , so that its opacity is primarily due to photoabsorption .
values for the photoionization cross - sections were taken from an updated version of the raymond & smith ( 1972 ) emissivity code provided by john raymond .
the opacity of the cooled material will depend somewhat on its ionization state .
we have assumed that the cooled material is cold and neutral .
although the physical state of the cooled material has not been established by observations , white et al .
( 1991 ) , ferland , fabian , & johnstone ( 1993 ) , and daines , fabian , & thomas ( 1994 ) have argued that the absorbing material must be in the form of small , cold clouds to have avoided detection at other wavelengths .
as mentioned above , the gas must be at a temperature @xmath54 for oxygen to retain k
shell electrons and thus present a significant photoelectric cross - section ( shull & van steenberg 1982 ) .
in fact , the total opacity should remain fairly constant for temperatures much below this value , since the cross - sections are fairly insensitive to changes in the ionization state once the ion has regained a full complement of k shell electrons . for a given model
, we must also specify the amount and spatial distribution of the cooled material in order to calculate its contribution to the photoelectric opacity . as discussed in
[ sec : intro ] , our approach is to construct self - consistent models in which the absorbing material is assumed to represent a fraction of all the gas which has cooled below x ray emitting temperatures over the lifetime of the cooling flow .
thus , we have taken the density of accumulated cold absorbing material as a function of radius in the cluster to be @xmath55 where @xmath56 is the rate at which material is cooling below x ray emitting temperatures and @xmath57 is the lifetime of the cooling flow , taken here to be @xmath58 years .
the factor @xmath59 is a free parameter which allows us to vary the fraction of the cooling material which is stored as cold absorbing gas . since it is essentially a deposition efficiency , we assume that @xmath60 .
inherent in the adoption of equation ( [ eq : rhocold ] ) is the assumption that the cold material accumulates where it cools . obviously , this assumption is a simplification , since in practice a cloud of gas which is colder ( and thus denser , assuming pressure equilibrium ) than the ambient gas will tend to fall in towards the center of the cluster .
white & sarazin ( 1987 ) have calculated the distribution of the cooled material assuming it moves on ballistic orbits . because such material spends a significant fraction of its orbital period at the largest distances from the cluster center ,
the resulting distributions are reasonably approximated by equation ( [ eq : rhocold ] ) .
alternatively , nulsen ( 1988 ) and fabian et al .
( 1991 ) have argued that the cold clouds must comove with the background cooling flow .
such co - motion also results in a distribution similar to equation ( [ eq : rhocold ] ) . in any event ,
the likely departures from equation ( [ eq : rhocold ] ) are in the sense that the profile of the cold gas is more centrally condensed , which would tend to magnify the effects discussed in this paper .
we have solved the transfer of radiation through the cooling flow numerically using the basic algorithm described by yorke ( 1980 ; 1986 ; 1988 ) .
the transfer equation is cast in the cylindrical @xmath61 coordinate system of hummer & rybicki ( 1971 ) .
note , we use @xmath62 rather than @xmath63 to represent the impact parameter which is fixed for a given `` ray '' through the flow and corresponds to a given projected radius on the sky .
the coordinate @xmath64 measures the depth into the cluster along each ray .
with this choice of coordinates , the transfer equation can be written @xmath65 where @xmath66 is the intensity of radiation at photon energy @xmath67 at any point in a given direction along a ray .
the emissivity , @xmath68 , contains contributions due to scattering and therefore depends implicitly on integrals of the intensity @xmath66 .
having specified the variation of @xmath68 and @xmath69 with radius within the flow , we solve equation ( [ eq : ray ] ) using the same numerical code employed in paper ii to calculate the effects of opacity due to the cooling gas .
spherical symmetry and a steady radiation field are assumed .
the code is non - relativistic and assumes that velocity @xmath70 , temperature @xmath71 , and photon energy @xmath67
satisfy @xmath72 , @xmath73 , and @xmath74 .
the effects of the flow velocity due to the cooling flow have been included .
this code is described in detail in paper ii and wise ( 1992 ) . for relevant details of the numerical technique ,
the reader is referred to these sources .
the models presented here are limited to the cooling flow region of the cluster ( @xmath75 ) . in practice ,
the cooling flow region is immersed within and smoothly connected to the general intracluster gas which extends out at least another order of magnitude in radius .
we have chosen not to include the ambient cluster component to our models in order to focus on the properties of the cooling flow region and avoid introducing additional parameters associated with the cluster distribution . in practice
, this decision means that some of the results of our models will differ from the equivalent properties directly observed in real clusters .
for example , the x ray surface brightnesses in our models drop to zero at the cooling radius ( e.g. , see figure [ fig : spect_tot ] below ) .
thus , to compare these profiles with actual cluster surface brightness profiles , the emission due to the foreground and background cluster gas must be removed .
absorption due to the accumulated cold material will diminish the total amount of x ray emission from the cooling flow .
the magnitude of this reduction in the total cooling flow luminosity is shown in figure [ fig : flux_tot ] as a function of the deposition fraction @xmath59 .
curves are given for three cooling flow models with a small degree of inhomogeneity @xmath76 ( c300_8_01 ) , a moderate degree of inhomogeneity @xmath77 ( c300_8_10 ) , and a high degree of inhomogeneity @xmath78 ( c300_8_40 ) .
three different bandpasses are presented , corresponding to the total x ray flux ( all photon energies @xmath79 kev ) , the flux observed with the _ rosat _ pspc ( position sensitive proportional counter ) instrument , and the harder bandpass of the _ asca _ gis ( gas imaging spectrometer ) . also shown for comparison are the effects of a column @xmath80 of foreground absorption .
for values of the deposition fraction @xmath81 , the opacity due to the accumulated cold material can produce a significant reduction in the total x ray luminosity on the order of 40% , and an even larger reduction in the flux seen in the soft _ rosat _ pspc bandpass ( as large as 60% ) .
the excess absorption columns seen by white et al .
( 1991 ) correspond to values of @xmath82 to 0.4 ( see figure [ fig : ncol_eta ] and [ sec : spect_fit ] below ) .
we note that these smaller values of @xmath59 still result in appreciable reductions of @xmath8330% in the total flux and @xmath8345% in the softer _ rosat _ band .
the differences between the three models in figure [ fig : flux_tot ] are relatively small . at first sight , this lack of distinction is surprising , since a more homogeneous model ( @xmath84 , c300_8_01 ) deposits more absorbing material in the center of the cluster where it covers less of the cooling flow region . on the other hand
, a more homogeneous model will also produce more of its x ray emission in this region as well .
in fact , equation ( [ eq : rhocold ] ) implies that there is always a correlation between the distribution of the x ray emissivity of the cooling gas ( proportional to @xmath85 ) and the opacity of the absorption due to cooled gas .
this correlation decreases the model - to - model differences in the reduction in the x ray luminosity .
the primary evidence for substantial amounts of cold , absorbing material in the data of white et al . ( 1991 ) is a deficit of x ray emission for photon energies below about 2 kev over that expected from purely galactic absorption . to provide a basis for comparison with spectral observations such as these
, we have calculated the emergent x ray spectra for our models . in figure
[ fig : spect_tot ] , the total integrated spectra from 0.110 kev are shown for a fairly homogeneous model ( @xmath84 , c300_8_01 ) for varying values of @xmath59 .
an absorption component due to foreground galactic material has not been included in this figure . as figure
[ fig : spect_tot ] demonstrates , the accumulated cooled material does produce significant absorption below 2 kev even for moderate values of @xmath59 . for spatially resolved spectroscopy , the spectral signatures of excess absorption can be even more apparent .
integrated spectra average over the entire range of absorbing column densities in the cooling flow resulting in a lower mean level of absorption .
however , a smaller aperture centered on the cluster will average over higher columns resulting in a greater mean absorption . figure [ fig : spect_cent ] shows several spectra for the same model extracted within a 10 kpc radius circular aperture centered on the core .
the degree of absorption is correspondingly greater than in figure [ fig : spect_tot ] .
in addition to the level of soft x ray absorption , it is also important to note that the accumulated cold material produces a distinctive signature in the _ shape _ of the emergent spectrum .
absorption due to foreground material produces an exponential fall off in the observed flux toward lower energies since , at a given photon energy @xmath67 , @xmath86 where @xmath87 is the optical depth .
however , for absorbing material which is intermixed with the emitting gas , the dependence of the absorption on optical depth is weaker .
roughly , we find that @xmath88 . consequently for a given absorbing column
, foreground absorption will produce deeper absorption edges and a stronger decrease in the spectrum toward lower photon energies than distributed absorption .
this point is illustrated in figure [ fig : spect_comp ] , which compares the spectrum for the @xmath84 ( c300_8_01 ) model with @xmath89 ( see figure [ fig : spect_tot ] ) with the spectrum of the same model with no internal absorption but foreground absorption corresponding to @xmath90 @xmath14 .
this amount of foreground absorption was selected to give roughly the same spectral shape as the internally absorbed model above 1.0 kev . as noted , the model with foreground absorption decreases much more steeply to lower photon energies , and has much deeper absorption edges ( e.g. , the o k - edge at 0.532 kev ) . in principle , this difference in the shape of the emergent spectrum may provide a means to distinguish between foreground and intrinsic absorption . for clusters with large foreground galactic absorption , however , making such a distinction may be problematic ( see [ sec : spect_fit ] ) .
centrally peaked x ray surface brightness profiles represent one of the primary pieces of evidence for the existence of cooling flows in clusters of galaxies . the x ray emission due to the cooling gas
is seen as an excess in the central surface brightness profile relative to that of the outer parts of the cluster ( jones & forman 1984 ) . in figure
[ fig : sbtot ] , the emergent x ray surface brightness profiles are shown for a fairly homogeneous model ( @xmath76 , c300_8_01 ) with values of @xmath59 ranging from 0 ( no absorption ) to 1.0 ( 100% of the accumulated cold material ) .
these profiles correspond to what would be observed in the _ rosat _ pspc bandpass .
clearly , the central peak in the cooling flow x ray emission is strongly depressed by the presence of distributed absorption .
for comparison , figure [ fig : sbtot ] also shows the x ray surface brightness profile in the pspc bandpass for the c300_8_fb model with @xmath48 and no internal absorption .
this model provides a reasonable fit to the observed x
ray surface brightness profiles of many cooling flow clusters ( e.g. , fabian et al . 1984 ) . as figure
[ fig : sbtot ] illustrates , even if only 10% of the cooling material collects in this cold form , substantial reductions in @xmath91 will occur .
the rough resemblance of the intrinsically absorbed , essentially homogeneous model profile to that of the optically thin , @xmath48 model highlights another important consequence of distributed absorption .
by suppressing the central peak of the surface brightness profile , a distributed absorber will make a given cluster cooling flow appear more inhomogeneous than it actually is .
_ rosat _ has provided high quality x
ray surface brightness profiles for many clusters . however , much of the initial work on the x ray surface brightness profiles of cluster cooling flows was based on observations with the _ einstein _ x ray observatory ( e.g. , fabian et al .
the _ einstein _
observatory had a substantially harder x ray bandpass , extending to approximately 4 kev . since the effects of absorption ( intrinsic or foreground ) are lessened at higher energies
, one might expect _ einstein _
surface brightness profiles to be less sensitive to the effects of distributed absorption . to test this intuition and provide standards for comparison to the _ einstein _ data , we have also calculated the x ray surface brightness for our models using the response of the _ einstein _ ipc ( imaging proportional counter ) .
figure [ fig : sbein ] confirms that internal absorption has a smaller effect on @xmath92 as seen by the _ einstein _
rosat_. roughly speaking , one must increase @xmath59 by about a factor of two to produce the same effect on _ einstein _ observations as with _
rosat_. the qualitative effect on the surface brightness profile is the same .
strong x ray line emission from cooling gas is one of the characteristic features of cluster cooling flows ( canizares , markert , & donahue 1988 ) .
many of the x ray lines produced by cooling plasma are non - resonance transitions having small oscillator strengths , @xmath93 , and correspondingly small optical depths due to resonant absorption .
such lines are affected by the internal absorption in the same way as the adjoining continuum discussed above .
consequently , the equivalent widths of optically thin lines are not strongly affected by internal absorption . however , many of the strongest lines arise from resonance transitions and can be moderately optically thick in cooling flows ( see paper ii ) . by itself , resonant line absorption ( or scattering ) merely redistributes photons spatially and spectrally .
when coupled with a significant source of continuum opacity , however , resonant scattering can increase the effects of the absorption on strong , resonance x ray lines .
this increase occurs for lines with significant optical depth because the effective pathlength through the absorbing medium is increased .
we consider some examples of these effects here . to illustrate some of the consequences of intrinsic absorption for emission lines , we consider the observed properties of the fe xvii @xmath94 resonance line which has a wavelength of 15.01 ( @xmath95 kev ) .
this line is produced by gas at temperatures in the range 13 @xmath96 k which is cooling rapidly out of the x ray emitting regime .
it has a large oscillator strength ( @xmath97 ) and an optical depth due to resonant absorption of @xmath980.32.0 depending on the assumed cooling flow model ( see paper ii ) .
since this fe xvii line has a low photon energy , it is also particularly susceptible to the effects of internal soft x ray absorption . in the simulations discussed here , we have assumed the line to be thermally broadened .
this assumption maximizes the optical depth in the line .
the gas in cluster cores , however , may be quite turbulent ( @xmath99 km s@xmath4 ) which would result in significantly broader emission line profiles and lower line optical depths ( see paper ii ) .
the reduction in the equivalent width , @xmath100 , of the fe xvii line as a function of @xmath59 is shown in figure [ fig : ew_fexvii ] for several models with varying degrees of inhomogeneity .
models with @xmath36 values of 0.1 , 0.3 , 1 , and 4 are shown ranging from fairly homogeneous to strongly inhomogeneous , respectively . the model with @xmath48
is also included for comparison . as figure
[ fig : ew_fexvii ] shows , resonant soft x
ray lines can be strongly attenuated by internal absorption in cooling flows . for a fixed value of @xmath59 , the equivalent width of the line decreases ( relative to the optically thin value ) as the deposition of the cooling material becomes more condensed , i.e. as @xmath36 decreases .
unlike the total luminosity of the flow , where the spatial coincidence of absorbing and emitting material serve to reduce model to model variations , the increase in opacity due to resonant scattering is more sensitive to the location of the absorbing material .
thus , models with differing mass deposition profiles can produce very different reductions in the line fluxes .
note that the reduction in the equivalent width of the line in figure [ fig : ew_fexvii ] is in addition to the overall reduction in the emission of continuum and optically thin lines at similar photon energies .
clearly , the fluxes of these lines can be greatly reduced . like their broadband counterparts ,
the presence of significant cold , absorbing material will suppress the central peaks in the surface brightness profiles of lower energy x ray emission lines .
profiles for resonance lines will be particularly affected due to the interplay between absorption from the accumulated cold material and the effects of resonant scattering .
figure [ fig : ewpro_fexvii ] shows the emergent equivalent width of the resonance fe xvii line as a function of projected radius for the fairly homogeneous @xmath76 ( c300_8_01 ) model assuming various values of @xmath59 . as the amount of internal absorption increases , the gradient in the equivalent width flattens dramatically inside of @xmath101 kpc .
again , since the equivalent width compares the flux in the line to that of the adjoining continuum , the reduction in equivalent width is in addition to the overall reduction in the x ray surface brightness shown in figure [ fig : sbtot ] . in the model without internal absorption ( @xmath102 )
, the surface brightness of the line is very strongly concentrated to the center of the cluster .
this concentration simply reflects the fact that , for relatively homogeneous models , most of the gas cools inside the sonic radius which occurs at rather small radii ( see paper i ) .
the fe xvii line is emitted by cooling gas at relatively low temperatures , so the emissivity for this line is very large near the center of the model .
similar results are found for other resonance fe l lines and for the resonance k - lines of low - z elements .
the k - lines of heavier elements ( si through ni ) are affected less by the opacity of accumulated cooled material due to their higher energy .
the combination of continuum opacity from accumulated cooled material and resonance scattering will also alter the spectral line profiles of low - energy resonance lines .
the left panel of figure [ fig : lprofile ] shows the spectral profile , @xmath103 , of the fe xvii line in the integrated spectrum of a cooling flow model for different values of the efficiency parameter @xmath59 .
these profiles were calculated for the fairly homogeneous @xmath84 model ( c300_8_01 ) and assume thermal line broadening .
the broad wings of the profile for the model without internal absorption are due to gas moving radially inwards at a fairly high velocity near the center of the model ( see paper i ) .
the right panel highlights the differences in the spectral profile due to optical depth effects and shows the ratio of the profiles with internal absorption to the profile with no accumulated cooled material .
overall , the combined effects of resonance scattering and internal absorption serve to reduce the strength of the line
. the broad wings of the line profile are produced by rapidly inflowing gas near the center of the model , where the opacity is largest .
consequently , the wings are reduced more than the line center , where there are contributions from larger radii . for lines produced by emitting material at larger radii or in models with smaller inflow velocities
, it is the line center which is most strongly reduced .
the spectral profiles depicted in figure [ fig : lprofile ] also exhibit asymmetries relative to the line center .
these asymmetries arise because blue - shifted photons ( due to material flowing into the cluster center on the far side of the cluster ) must traverse a greater absorbing column than red - shifted photons ( due to material flowing into the cluster center on the near side of the cluster ) .
similar results are found for other soft x
ray resonance lines .
emission lines from higher ionization stages such as the fe k@xmath104 at 6.7 kev are less effected by photoelectric absorption , although such lines can still be affected by resonance scattering ( paper ii ) .
these effects are reduced if the lines are significantly turbulently broadened , as this reduces the line scattering optical depth . in any case
, the detection of these effects will require very high spectral resolution observations ; as a result , it may be some time before opacity effects on line shapes can actually be observed .
in addition to its broadband effects ( see [ sec : xspect ] ) , accumulated cooled material in the cluster core will produce absorption edges in the emergent x ray spectra of cooling flows . at the columns of interest here ,
the strongest spectral feature associated with the absorption is the o k - edge at 0.532 kev ( e.g , figure [ fig : spect_comp ] ) .
if the intrinsic absorption is associated with the cooling flow , the o k - edge will be redshifted to lower energies such that @xmath105 kev .
thus , two o k - edges will appear in the spectrum : an unredshifted edge due to galactic gas ( and possibly oxygen in the detector and spacecraft environment ) , and a redshifted edge .
detection of such a redshifted edge feature at the redshift of the cluster would provide definitive evidence that the excess absorbing material was associated with the cooling flow and not due to anomalously high foreground absorption .
arnaud & mushotzky ( 1998 ) have reported detection of such a redshifted o k - edge in the perseus cluster based on analysis of bbxrt spectra .
we can measure the observed depth of the o k edge from our self - absorbed cooling flow models directly . in figure
[ fig : edge ] , the optical depth in the edge is plotted for a range of deposition efficiencies , @xmath59 , in the various models .
the observed optical depth was determined assuming the decrement in the observed flux at the edge energy was given by @xmath106 .
note that this assumption is equivalent to assuming a foreground absorption model . for @xmath107 , the measured depth in the edge ranges between @xmath1080.3 for all the models .
given an absorption cross section at o k of @xmath109 @xmath110 , these optical depths imply columns of @xmath111@xmath112 @xmath14 assuming the absorbing material is in the foreground of the cluster . as we show below in [ sec : spect_fit ] , this `` foreground '' assumption leads to significant underestimates of the true excess absorbing column .
no foreground galactic column was assumed in figure [ fig : edge ] .
the spatial variation of the o k - edge optical depth is shown in figure [ fig : edge_spatial ] .
for each model , the cooling region was divided into annular bins 5 kpc wide and the optical depth in the edge was then determined for the spectrum from each bin .
the different spatial profiles apparent in figure [ fig : edge_spatial ] simply trace the degree of inhomogeneity , and hence the degree to which the absorbing material is centrally condensed , in the underlying model . for a given model , @xmath113 increases steadily as one moves to smaller projected radii until the point where the cooling flow becomes optically thick in the edge . in the model with @xmath37 ,
this point occurs at a radius of @xmath114 kpc ; while in the slightly inhomogeneous @xmath76 model , @xmath113 increases steadily into a radius of @xmath115 kpc .
since this `` saturation '' point corresponds to the radius at which the actual optical depth in the edge is @xmath116 , figure [ fig : edge_spatial ] implies that the measured value of @xmath113 can be used to estimate the true absorbing column at this radius . from figure
[ fig : edge_spatial ] , the observed values for @xmath113 at these saturation radii range between @xmath1170.30 implying underestimates of the true columns by factors of @xmath1184 .
this values are consistent with the results presented below based on spectral fits to the integrated spectra assuming foreground absorption models ( see [ sec : spect_fit ] ) . in practice
, it is difficult to separate an intrinsic , redshifted o k - edge from a galactic edge with the _ asca _ sis ( solid imaging spectrometer ) or _ axaf _ acis ( _ axaf _ ccd imaging spectrometer ) because of their spectral resolution at these energies .
however , such observations may be possible with _ axaf _ for some higher redshift clusters ( @xmath119 ) , where the intrinsic edge is separated from the galactic feature by @xmath120 ev .
although the high energy transmission grating ( hetg ) spectrometer on _ axaf _ has high spectral resolution , its throughput is low and its resolution is degraded when observing extended sources .
astro - e , on the other hand , should easily be capable of detecting redshifted absorption edge features in cooling flow spectra . obviously , it is not possible to present results covering the full range of all cluster cooling flow parameters in this study .
thus , we have given a few characteristic examples to demonstrate the effects of internal absorption .
however , it is generally possible to crudely assess the effects of internal absorption for other parameter values by approximate scaling of the models presented here .
the steady - state models we have presented are primarily characterized by five parameters : the total cooling rate @xmath50 ; the temperature at the cooling radius @xmath25 ; the age of the cooling flow @xmath57 ; the gas loss efficiency parameter @xmath36 ; and the fraction of accumulated cooled gas which is present in an x - ray absorbing form @xmath59 . in addition , the models depend on the parameters characterizing the gravitational field of the cluster and central galaxy , and the abundances in the cooling gas . however , if the temperature at the cooling radius is sufficiently high and one is not concerned with the details of the innermost parts of the flow near the sonic radius , the hydrodynamical solutions do not depend strongly on the detailed form of the gravitational potential .
the atomic physics of the emission processes does depend in a complex fashion on the abundances and the outer temperature , so we will assume these are fixed in the scaling given below . for a fixed temperature @xmath25 and fixed abundances ,
the density @xmath121 at the cooling radius @xmath22 depends inversely on age , @xmath122 .
the total cooling rate in a model scales approximately as @xmath123 .
thus , the cooling radius in a model with fixed abundances and @xmath25 varies approximately as @xmath124 equation [ eq : rhocold ] implies that the total mass of the cold absorber in the model is @xmath125 consequently , the column density of the cold absorber @xmath126 and the implied absorption optical depth @xmath127 both scale as @xmath128 intervening absorption affects the observed properties in nonlinear way ; therefore , the value of @xmath127 must be the same for two models if they are to be scaled to one another .
on the other hand , most of the emission properties of cooling flow models nearly scale with the total cooling rate .
thus , the luminosity in any spectral feature ( and the flux in that feature in the integrated spectrum ) approximately scale as @xmath129 the surface brightness of most features will scale approximately as @xmath130 resonance emission lines present an exception to these scaling relations ( [ sec : xline ] ) .
the properties of these lines depend on both the optical depth due to absorption by cold material @xmath127 and due to resonance scattering @xmath131 . the latter scales approximately as @xmath132 requiring this quantity to be constant would generally make it difficult to scale the models . in what follows , we will consider the scaling of all properties of the models except those associated with resonance emission lines which are optically thick to resonance scattering .
let us assume one wishes to determine the properties of a model characterized by @xmath25 , @xmath36 , @xmath50 , @xmath57 , and @xmath59 which is not presented in this paper .
further , assume that a model with the desired @xmath25 and @xmath36 is presented , but with a different total cooling rate @xmath133 and age @xmath134 .
then , from the grid of models presented in this paper , one should select the model with a value of the deposition efficiency parameter , @xmath135 , given by @xmath136 we note that this scaling can break down for models with large values of @xmath59 and significantly different cooling rates or ages . strictly speaking ,
equation [ eq : eta_scale ] could produce values of @xmath135 which are much greater than unity .
this much absorption could not be produced by material cooling at the present rate .
once the value of @xmath135 has been fixed , the luminosity and integrated spectrum of the desired model can be scaled from the corresponding fiducial model using @xmath137 while the surface brightnesses scale as @xmath138 we have verified these relations for a number of cooling flow models , and find that they work to within 10 - 20% in essentially all cases .
with the exception of allen & fabian ( 1997 ) , most previous analyses of the x ray spectra from cluster cooling flows have not considered the effects of additional absorption due to accumulated cooled material . traditionally , any absorption in the cooling flow spectra has been assumed to come from foreground material and in many cases has been fixed at the galactic value as measured by stark et al .
neglecting the effects of this material , if present , can lead to serious errors in the derived physical properties of the gas in the cluster core . in the following discussion , we examine the effects of this intermixed internal absorption on a number of physical quantities commonly derived from x ray observations . assuming the x ray luminosity from a cooling flow is due to material cooling isobarically from some temperature @xmath71 , to first order @xmath139 is simply proportional to the enthalpy of the cooling gas : @xmath140 for a measured total luminosity , @xmath139 , this expression can be inverted to obtain an estimate of the total cooling rate of material . as we demonstrated in
[ sec : xlum ] , even moderate amounts of accumulated cooled material can produce significant reductions in the observed x ray luminosity .
therefore , estimates of the total cooling rate @xmath141 based on the _ observed _ value of @xmath139 will tend to _ underestimate _ the true value if the effects of internal absorption are neglected . based on the results in figure [ fig : flux_tot ] ,
the derived values of @xmath141 might be as much as a factor of @xmath832 too low . as we show below ( [ sec : spect_fit ] ) , estimates of the x ray cooling rate based on fits to x ray spectral observations which assume only foreground absorption will tend to underestimate the true absorbing column and therefore @xmath8 as well . individual x
ray emission lines can also be used to measure the total cooling rate of material in cluster cores . if the line is emitted primarily from material at temperatures below the ambient cluster temperature @xmath25 , then the luminosity in the line is given by @xmath142 where @xmath143 is the emissivity coefficient of the line . for lines emitted by gas at temperatures which are smaller than @xmath25 ,
the luminosity is nearly independent of @xmath25 , the details of the cooling , or even the abundances ( which affect both @xmath144 and @xmath145 in a similar manner ) .
this technique requires high spectral resolution observations and has been applied to a number of clusters based on fpcs data from the _ einstein _ x ray observatory ( canizares , markert , & donahue 1988 ) .
the high energy transmission grating ( hetg ) onboard _ axaf _ should also provide data from several cluster cooling cores with sufficient spectral resolution to apply this technique . for optically thin x
ray emission lines , the effects of intrinsic absorption will reduce the observed luminosity in the line , @xmath146 , to the same degree as the nearby continuum .
thus , lines at lower energies will suffer greater attenuation and consequently underestimate the total cooling rate , @xmath8 , to a greater degree relative to lines at higher energies .
for example , assuming a value of @xmath81 , optically thin lines with energies near 0.6 kev could suffer reductions as large as 60% whereas the luminosity of a line at 2.0 kev would be reduced by only 20% .
these reductions are in addition to any reduction due to the intervening galactic absorbing column .
the actual observed reductions will of course depend on the true underlying deposition profile of cooling material and the detector response .
if the line is a resonance line with an appreciable optical depth , an additional reduction of about a factor of two is possible ( see [ sec : xline ] and figure [ fig : ew_fexvii ] ) . as an aside
, we note that differing amounts of intrinsic absorption in cluster cooling cores may be partially responsible for the scatter observed in correlations between @xmath141 and other cooling flow properties .
given the many potential sources of scatter in observed cluster correlations , it seems unlikely that accumulated cold material could be solely responsible .
however , as these models illustrate , the magnitude of these effects is such that the presence of this material could represent a significant contribution to the observed scatter in such correlations .
centrally peaked x ray surface brightness profiles are one of the characteristic features of cluster cooling flows .
however , it has been known for some time that the observed profiles are less centrally peaked than would be expected if the gas in cluster cooling flows was homogeneous and all of the gas flowed into the inner regions ( @xmath147 kpc ) before cooling ( thomas et al .
1987 ; white & sarazin 1987 ) .
the conventional explanation of this discrepancy is that the gas is very inhomogeneous , and that denser lumps of gas cool more quickly .
thus , a significant portion of the gas cools below x ray emitting temperatures before it flows in very far , and does not contribute to the central x ray emission .
one way of quantifying this model is to calculate the amount of hot gas @xmath148 which is still flowing inward at each radius @xmath149 . by mass conservation
, this quantity is equal to the amount of gas which cools within this radius in steady - state .
thomas et al . (
1987 ) have argued that many clusters are reasonably well represented by a mass deposition profile of the form @xmath37 out to @xmath150 kpc .
one of our models ( c300_8_fb ) incorporates this spatial variation in the cooling rate .
the implied form of the mass deposition profile @xmath151 will depend on the observed x
ray surface surface brightness profile @xmath152
. however , as discussed in
[ sec : xsb ] , intrinsic absorption from accumulated cooled material has the effect of reducing the central surface brightness of cooling flows .
consequently , the gas in cooling flows may be less inhomogeneous than a simple interpretation of the observed surface brightness profiles might imply .
this effect is illustrated in figure [ fig : sbtot ] which shows the predicted _ rosat _ pspc x ray surface brightness profiles for a slightly inhomogeneous @xmath76 model with different values of @xmath59 ( expressed as percentages ) . as mentioned previously , cooling flow models with @xmath153 produce surface brightness profiles in reasonable agreement with x ray observations . for comparison , the optically thin surface brightness profile for the @xmath153 model
is indicated by the heavy solid line .
clearly , a value of @xmath154% is sufficient to bring the surface brightness of this nearly homogeneous @xmath76 model into rough agreement with the optically thin @xmath153 model . in [ sec : spect_fit ] below , we show that this level of internal absorption would correspond to an observed excess column of about @xmath155 @xmath14 if the spectrum were fit assuming only foreground absorption .
this value is similar to the excess columns which have been derived from x ray spectra of cooling flows under this assumption ( white et al .
1991 ) .
as noted above , the variation of the inflow or cooling rate @xmath148 as a function of radius has been used to characterize the degree of inhomogeneity of a cooling flow .
we have treated the predicted _ rosat _ x ray surface brightness data from our models as observed data , and deprojected the emission to determine @xmath148 .
figure [ fig : mdot ] shows the deprojected mass deposition profiles corresponding to the _ rosat _ x ray surface brightness profiles shown in figure [ fig : sbtot ] . for comparison ,
the heavy solid line shows the expected profile for a model with @xmath37 ( c300_8_fb ) . as the degree of internal absorption
is increased ( @xmath156 ) , the derived @xmath151 profiles steepen which would be interpreted as evidence that more of the gas is cooling below x ray emitting temperatures at larger radii .
again , a value of @xmath82 would cause the surface brightness of this fairly homogeneous @xmath76 model to have a mass deposition profile which resembled @xmath37 .
these qualitative results apply to all of the models .
we have illustrated the case for the fairly homogeneous @xmath76 model because the effects are particularly striking here .
however , we note that the specific values of @xmath59 required to turn a more centrally condensed mass deposition profile ( or equivalently , surface brightness profile ) into one which resembles the observed @xmath157 will depend on the specifics of the input model and the detector response . for example , due to its harder bandpass , if the above comparison were made with _
einstein _ ipc data , an @xmath59 value a factor of two larger would be required to produce the same effect .
similarly , a much smaller spectrally determined excess column of order @xmath158 @xmath14 would convert a moderately inhomogeneous @xmath159 model into a @xmath37 model . as a final point
, we note that the above effects do _ not _ apply to completely homogeneous models . in such models ,
the gas does not cool to low temperatures until it is inside the sonic radius which typically occurs at very small radii ( @xmath147 kpc ) .
therefore , any intrinsic absorption due to this accumulated cooled material would also be concentrated within a very small region . if present , this material would produce a large reduction in the surface brightness profile but only at the very center of the cluster
. it would not affect the surface brightness at larger radii since no absorbing material would have accumulated there .
the evidence for excess absorption in cluster cooling flows has come mainly from spectral fits such as those by white et al .
( 1991 ) . as we demonstrated in
[ sec : xspect ] , absorbing material intermixed with the emitting plasma produces a characteristic spectral shape in the emergent spectrum which is substantially different from the simple exponential fall - off created by foreground absorption .
consequently , fits to spectra from self - absorbed cooling flows assuming only foreground absorption will yield erroneous values for the derived columns .
specifically , such fits will tend to _ underestimate _ the true absorbing column . to quantify this effect
, we have fit foreground absorption models to the emergent x ray spectra of our models with internal absorption . in the following analysis
, we have used the effective area and spectral resolution of the _ rosat _ pspc due to its good soft x
ray response .
most observations of excess absorption in cluster cooling flows have been based on the integrated spectra for the entire cooling flow region .
for example , the original study by white et al .
( 1991 ) utilized the _
sss detector with a field of view of 6 arcmin .
most nearby cluster cooling flows are several arcmins in radius which means that the sss gives the integrated spectrum of the entire cooling flow .
similarly , _
has a resolution of several arcmins and typically yields integrated spectra for the entire cooling flow region .
the _ rosat _
pspc has a better angular resolution of about 0.5 arcmin , but still only gives a few separate spectra within the cooling flow regions of nearby clusters ( e.g. , allen et al .
1993 ) . for comparison to such integrated spectra
, we have determined the best - fit foreground column @xmath160 using the equivalent integrated spectra from our models . to perform the spectral fits ,
we have used the following procedure . for a given cooling flow model and choice of deposition efficiency , @xmath59 ,
we have calculated the integrated , emergent spectrum for the entire cooling flow .
this spectrum was then convolved with the instrument response of the _ rosat _ pspc to obtain the number of `` observed '' counts per energy bin , @xmath161 for this model and choice of @xmath59 .
the instrument response consists of the effective area , @xmath162 , and the redistribution function , @xmath163 , produced by the pspc s finite spectral resolution . as a fitting model
, we used the optically thin ( @xmath164 ) spectrum for the selected cooling flow multiplied by varying amounts of foreground absorption and convolved with the pspc instrument response . for a given choice of @xmath80 , this model yields the expected counts per energy bin , @xmath165
the value of @xmath80 which yielded the best fit was then obtained by minimizing the @xmath166 statistic @xmath167 ^ 2 } { f_i^{fit}(n_h ) } \ , , \label{eq : spec_fit}\ ] ] where @xmath168 is an index assigned to the spectral energy bins .
the @xmath169 in the denominator of equation [ eq : spec_fit ] assures the correct behavior if the errors in the observed spectra are dominated by poisson errors from photon noise . finally , because all clusters are observed through some amount of galactic material , both @xmath170 and @xmath169 were also multiplied by varying amounts of galactic foreground absorption prior to fitting .
fits assuming galactic columns of 0.1 , 0.2 , 1.0 , and @xmath171 @xmath14 were calculated and these columns were subtracted from the best - fit foreground absorption values after the fit .
figure [ fig : ncol_eta ] gives the excess column densities determined using the above procedure as a function of the deposition efficiency @xmath59 . using this figure
, we can compare the properties of our models with a given @xmath59 to the observed properties of clusters for which the excess absorption @xmath160 has been derived .
note that the values in figure [ fig : ncol_eta ] are specific to @xmath172 yr@xmath4 and @xmath24 yr .
the scaling to other values is given in equation [ eq : abs_scale ] .
the typical excess columns densities found by white et al .
( 1991 ) assuming foreground absorption models are of order @xmath173 @xmath14 .
these column values are in the regime where all of the model curves in figure [ fig : ncol_eta ] cross , and correspond to @xmath82 to 0.4 for all of the models .
observationally , values of @xmath174 have been used to derive limits on the total amount of cold , x ray absorbing material in cooling flows .
consequently , it is useful to compare the spectrally derived columns with those actually present .
the sense of the difference depends on whether one considers spatially resolved spectra or the integrated spectrum of the entire cooling flow region . for spatially resolved spectra ,
the spectrally derived column assuming only foreground absorption is always smaller than the actual column along that line of sight .
this underestimate occurs because , for intermixed absorption , a fraction of the absorbing material always lies behind a portion of the emitting gas , and thus is not fully effective in producing absorption . however , if one considers the integrated spectrum of an extended region where the absorbing column and emission vary with projected position , the difference can have either sign .
a given amount of absorber is more effective if it is concentrated where the emission is most intense , and such a concentration may not be resolved in the integrated spectrum .
the spectrally - derived excess columns in figure [ fig : ncol_eta ] are determined from the integrated spectrum of the entire cooling flow region .
following equation [ eq : abs_scale ] , the average column density of hydrogen in our models over the cooling flow region is given by @xmath175 figure [ fig : ncol_comp ] compares the actual average column densities determined using equation [ eq : dnh_real ] with those derived from fits to the integrated spectrum of the cooling flow region ( fig .
[ fig : ncol_eta ] ) .
the heavy solid line indicates @xmath176 . for low column in the inhomogeneous models ,
the spectrally - derived columns overestimate the real column . for large columns of intrinsic material , @xmath177 @xmath14 , and for the most inhomogeneous models , the spectrally derived column is smaller than the real average column for all of the models by a factor @xmath178 . for the more inhomogeneous models ( @xmath179 ) the measured columns underestimate the true values down to columns of @xmath180@xmath181 @xmath14 .
these results can be compared to a similar analysis performed by allen & fabian ( 1997 ) for a set of cooling flow clusters observed with the _ rosat _ pspc . using xspec simulations , these authors have simulated the errors in the apparent excess column for two distributions of internal absorption : a partial covering model and a `` multilayer '' model which assumes a homogeneous distribution of absorbing throughout the x ray emitting region .
these input models were then fit with a simple uniform foreground absorption model similar to our procedure .
of the two , this latter model is the closest approximation to the detailed distributions of accumulated cooled material modeled here . for values of @xmath182 of 1 and @xmath183 @xmath14 and a galactic column of @xmath184 of @xmath185 @xmath14
, these authors find that the fitted column densities underestimate the true `` multilayer '' columns by factors @xmath1865 .
these results are in good agreement with the factors of @xmath1184 obtained from our models .
allen & fabian ( 1997 ) also note that the degree of underestimation is essentially independent of the galactic column , a result we confirm . for the more homogeneous models @xmath187 , the spectrally - derived column can actually over - estimate the true average column for small amounts of absorption .
this behavior is seen in figure [ fig : ncol_comp ] for values of @xmath188 @xmath14 . in these models ,
both the emission and absorption are strongly concentrated to the center of the cooling flow .
when the emitting and absorbing material are concentrated in this way , a smaller amount of absorber is more effective . one way of understanding this effect is to note that in fairly homogeneous models , the effective radii containing most of the emission and absorption are both much smaller than the cooling radius .
thus , if one averages the absorption over the projected area of the entire cooling flow , as in equation [ eq : dnh_real ] , the resulting value of the average column density @xmath182 will be much smaller than the true value in the region where the absorbing material resides . with the exception of a few nearby clusters observed with _
, the cooling cores of clusters have not been well resolved spatially by instruments with good spectral resolution .
consequently , studies of excess absorption have depended primarily on analyses of integrated spectra .
however , _ axaf
_ s excellent spatial resolution ( fwhm @xmath189 arcsec ) will provide spatially resolved spectra for many cooling flows . with such future data in mind
, we have calculated the values of @xmath182 and @xmath160 as a function of projected radius in the cooling flow . to determine @xmath160
, we divided the cooling region into annular bins 5 kpc wide .
this choice gives roughly 20 bins over the cooling radius for each model . for a choice of @xmath59 ,
the emergent spectra within annular bins was calculated from the model surface brightness using @xmath190 where @xmath191 and @xmath192 represent the inner and outer radii of the bins .
similarly , emergent spectra were extracted from the corresponding optically thin model for the same set of annuli .
these spectra were then fit using the same procedure outlined above to calculate the best - fit values of @xmath160 . for comparison ,
the true variation of @xmath193 was calculated directly from the distribution of accumulated cooled material , @xmath194 ( see equation [ eq : rhocold ] ) , using @xmath195 where @xmath196 is the distance along the line of sight at a projected radius @xmath62 .
given @xmath193 , the average column density within an annular bin was then taken to be @xmath197 we note that equation [ eq : dnh_ave ] reduces to equation [ eq : dnh_real ] in the limit that @xmath198 and @xmath199 .
these expressions were used to derive the true average column densities for the same annular grid used in determining @xmath160 .
figure [ fig : ncol_spatial ] shows the results of this spatial analysis for two models : the slightly inhomogeneous @xmath84 model ( c300_8_01 ) and the model with @xmath157 ( c300_8_fb ) .
as expected , the spectrally derived column density underestimates the true column at all projected radii .
the magnitude of this error increases as one moves towards the center of the cluster . for the model with @xmath157 ,
the column determined from spectral fits varies over a range of @xmath11810 as one moves to smaller radii . in the @xmath84 model ,
more of the absorbing material is concentrated toward the center of the flow producing underestimates of the true absorbing column by factors of @xmath130 .
as before , the galactic column has been subtracted from the curves in figure [ fig : ncol_spatial ] .
_ axaf _ s spatial resolution should yield spatially resolved spectra , and consequently measurements of @xmath15 , on spatial scales comparable to this example for clusters out to redshifts of @xmath200 .
in this paper , we have calculated models for the x ray properties of cooling flow clusters including the effects of opacity due to accumulated cooled material . for consistency , we have assumed that the internal absorber has the same distribution as the gas which is cooling below x ray emitting temperatures in the models .
we have characterized the amount of accumulated absorbing material as a fraction @xmath59 of the amount of gas which would cool over the lifetime of the cluster ( @xmath24 yr ) at the present rates of cooling .
the emergent properties of the x ray emission were then calculated including the effects of radiative transfer through this accumulated material .
the presence of accumulated cooled material will reduce the observed x
ray fluxes and the implied cooling rates of cooling flow clusters below the actual values . depending on the efficiency with which cooling gas
is converted into absorbing material , the fluxes and cooling rates may be underestimated by as much as a factor of two over the 0.110 kev energy range . for instruments with softer bandpasses , such as _ rosat _ , the reduction can be higher .
although modeling the effects of the intrinsic absorption with foreground absorption models will reduce the magnitude of this underestimate , these models tend to underestimate the total amount of absorbing material and therefore do not completely compensate for this effect .
intrinsic absorption produces a characteristic spectral morphology below @xmath1 kev which is not well fit by foreground absorption models .
specifically , internal absorption does not increase in effectiveness as an exponential function of the optical depth , as foreground absorption does . as a result , the integrated x
ray spectra for the models described here do not decline as rapidly at softer energies and the absorption edges are less deep than for models with foreground absorption . for spatially
resolved spectra toward the core of the cooling flow , this effect is enhanced .
we find that these differences should be discernible in present _ asca _ and future astro - e and _ axaf _ spectra .
on the other hand , it is difficult to distinguish spectra of models with internal absorption from those with foreground absorption with a partial covering factor .
in addition to its effect on the broadband x ray spectrum , accumulated cooled material will produce redshifted absorption edges such as the o k edge at an energy of @xmath201 kev .
these redshifted edges provide a direct means of determining whether the absorbing material is intrinsic to the cooling flow and not due to anomalously high foreground absorption . resolving such redshifted o k edge features from edges due to foreground material and o in the ccd detectors of _ asca _ and _
axaf _ is difficult for nearby clusters given the spectral resolution of these instruments .
these features may be detectable in more distant clusters with _
axaf _ and _ xmm _ where the redshifted o k
edge is more cleanly separated or in nearby cooling flows with astro - e which has greater spectral resolution .
the effects of internal absorption can have important implications for the analysis of x ray surface brightness profiles from cluster cooling flows .
because the absorbing material in our models is concentrated toward the central regions of the cooling flow , it acts particularly to diminish the central peak in the surface brightness and flatten the observed profiles . starting with _ einstein _
observations , it has been noted that cluster cooling flows are less centrally peaked than would be expected if the cooling gas were homogeneous . this fact has led to the canonical view that the gas in cooling flows is rather inhomogeneous and that much of the gas cools at large radii . however , our models indicate that opacity due to accumulated cooled material can produce emergent surface brightness profiles much like those observed even for nearly homogeneous gas distributions ( @xmath84 ) .
consequently , much more of the gas may be cooling below x ray emitting temperature in the central regions of cooling flows ( @xmath2 kpc ) than one would infer from observed x ray surface brightness profiles assuming the gas was optically thin .
another way of presenting the same result is to note that internal absorption steepens the mass deposition profile , @xmath148 , one would derive from the observed surface brightness distribution , making a nearly homogeneous model resemble one with @xmath202 .
we note that the reduction in the central surface brightness due to internal absorption would also mean that central cooling times have been overestimated , and that central gas densities and pressures have been underestimated .
low energy resonance x ray emission lines are particularly sensitive to the presence of internal absorption .
these lines can have appreciable resonance scattering optical depths in cooling flows , and this scattering enhances the chances of absorption by accumulated cooled material . for lines such as the fe xvii line at 0.826 kev , the interplay between continuum opacity and resonant scattering can lead to reductions of 20%80% in the equivalent width of the line .
like their broadband counterparts , estimates of the cooling rate based on such lines will underestimate the true cooling rate .
line surface brightness profiles and spectral profiles are likewise effected . with sufficient spectra resolution
, one can resolve distortions in the spectral profiles of these low energy resonance lines due to absorption .
however , detection of such signatures requires spectral resolution well beyond that currently available .
our models assume thermal line broadening .
if significant turbulence is present in the cooling flow gas , the magnitude of these effects is reduced .
emission lines from high ionization stages , such as the fe k@xmath104 line at 6.7 kev , are relatively unaffected by internal absorption though they may still have appreciable resonance scattering optical depths .
previous studies of intrinsic absorption in cooling flows have primarily determined the amount of absorbing material by fitting foreground absorption models to the observed x ray spectra .
as our models indicate , the spectral signatures of accumulated cooled material are not well fit by such models . for spatially resolved spectra , the spectrally
derived columns always underestimate the actual columns along the line of sight , typically by a factor of @xmath11820 depending on the specifics of the cooling flow model . if the amount of excess absorption is derived from integrated x ray spectra for the entire cooling flow ( as has generally been the case to date due to the limited spatial resolution of the spectrometers used ) , the situation is more complicated .
for large columns or fairly inhomogeneous models , the absorption is always underestimated this way . however , for small columns in nearly homogeneous models , the degree of excess absorption can be overestimated .
this effect occurs because both the emission and absorption in nearly homogeneous models is strongly centrally concentrated , and a small amount of absorbing material is more effective in this case . c. l. sarazin would like to thank a. fabian for helpful conversations . c. l. sarazin was supported in part by nasa astrophysical theory program grant nag 5 - 3057 .
m. wise s research at m.i.t .
is supported in part by the _ axaf _ science center as part of smithsonian astrophysical observatory contract svi61010 under nasa marshall space flight center .
lcccccc + model & previous & @xmath203 & @xmath25 & gas loss & @xmath22 & @xmath204 + & model & ( @xmath205 yr@xmath4 ) & ( k ) & & ( kpc ) & ( pc ) + c300_8_01 & & 300&@xmath206&@xmath76 & 95.6 & 582 + c300_8_03 & & 300&@xmath206&@xmath207 & 93.5 & 414 + c300_8_10&c3 & 300&@xmath206&@xmath77 & 87.6 & 115 + c300_8_40&c4 & 300&@xmath206&@xmath78 & 72.3&0 + c300_8_fb&c5 & 300&@xmath206&@xmath48&83.7 & 11 + c100_8_10 & & 100&@xmath206&@xmath77 & 61.3 & 69 + | we have calculated the emergent x ray properties for models of cluster cooling flows including the effects of accumulated cooled material .
the opacity of this cooled gas can reduce the overall x ray luminosity of the cooling flow , and values of @xmath0 based on these luminosities can underestimate the true value by factors of @xmath1 .
we find that accumulated cooled material can produce emergent surface brightness profiles much like those observed even for nearly homogeneous gas distributions .
consequently , much more of the gas may be cooling below x ray emitting temperatures in the central regions of cooling flows ( @xmath2 kpc ) than one would infer from observed x ray surface brightness profiles assuming the gas was optically thin .
similarly , the central densities and pressures in cooling flows may have been underestimated .
we show that distributed absorption in cooling flows produces a number of observable effects in the spectrum which may allow it to be differentiated from absorption due to gas in our galaxy .
these include a characteristic suppression of the continuum below @xmath1 kev , absorption features such as a redshifted o k edge , and diminished intensity of resonance emission lines .
spectra including the effects of intrinsic absorption are not well fit by foreground absorbing models .
attempting to fit such models to the spatially resolved spectra can lead to underestimates of the true absorbing column by factors of 320 .
fits to integrated spectra of the entire cooling flow region can either underestimate or overestimate the mass of the absorbing gas depending on the specifics of the model .
we discuss the potential detection of these effects with _
axaf _ , _ xmm _ , and astro - e . |
section [ background ] provides our motivation for looking at torsional rigidity , and points to the relevant literature .
section [ torsrig ] introduces our main object of interest , the torsional rigidity of the complement of brownian motion on the unit torus .
section [ asympscal ] states our main theorems .
section [ disc ] places these theorems in their proper context and makes a link with the principal dirichlet eigenvalue of the complement .
section [ sketch ] gives a brief sketch of the main ingredients of the proofs and provides an outline of the rest of the paper .
let @xmath21 be a geodesically complete , smooth @xmath1-dimensional riemannian manifold without boundary , and let @xmath22 be the laplace - beltrami operator acting in @xmath23 .
we will in addition assume that @xmath24 is stochastically complete .
that is , brownian motion on @xmath24 , denoted by @xmath25 , with generator @xmath22 exists for all positive time .
the latter is guaranteed if for example the ricci curvature on @xmath24 is bounded from below .
see @xcite for further details . for an open , bounded subset @xmath26 , and @xmath27 we define the first exit time of brownian motion by @xmath28 it is well known that @xmath29\ ] ] is the unique solution of @xmath30 with initial condition @xmath31 .
the requirement @xmath32 represents the dirichlet boundary condition .
if we denote the expected lifetime of brownian motion in @xmath5 by @xmath33 , \qquad x\in \omega,\ ] ] where @xmath34 denotes expectation with respect to @xmath35 , then @xmath36 it is straightforward to verify that @xmath37 , the _ torsion function _ for @xmath5 , is the unique solution of @xmath38 the _ torsional rigidity _ of @xmath5 is the set function defined by @xmath39 the torsional rigidity of a cross section of a cylindrical beam found its origin in the computation of the angular change when a beam of a given length and a given modulus of rigidity is exposed to a twisting moment . see for example @xcite .
from a mathematical point of view both the torsion function @xmath37 and the torsional rigidity @xmath40 have been studied by analysts and probabilists .
below we just list a few key results . in analysis ,
the torsion function is an essential ingredient for the study of gamma - convergence of sequences of sets .
see chapter 4 in @xcite .
several isoperimetric inequalities have been obtained for the torsional rigidity when @xmath41 .
if @xmath42 has finite lebesgue measure @xmath43 , and @xmath44 is the ball with the same lebesgue measure , centred at @xmath45 , then @xmath46 .
the following stability result for torsional rigidity was obtained in @xcite : @xmath47 here , @xmath48 is the fraenkel asymmetry of @xmath5 , and @xmath49 is an @xmath1-dependent constant . the kohler - jobin isoperimetric inequality @xcite,@xcite states that @xmath50 stability results have also been obtained for the kohler - jobin inequality @xcite . a classical isoperimetric inequality @xcite states that @xmath51 in probability , the first exit time moments of brownian motion have been studied in for example @xcite and @xcite
these moments are riemannian invariants , and the @xmath52-norm of the first moment is the torsional rigidity .
the _ heat content _ of @xmath5 at time @xmath9 is defined as @xmath53 this quantity represents the amount of heat in @xmath54 at time @xmath9 , if @xmath54 is at initial temperature @xmath55 , while the boundary of @xmath54 is at temperature @xmath45 for all @xmath56 . by , @xmath57 , and so @xmath58 finally by , and we have that @xmath59 i.e. , _ the torsional rigidity is the integral of the heat content_. in the present paper we consider the flat unit torus @xmath0 .
let @xmath60 be a second independent brownian motion on @xmath0 .
our object of interest is the random set ( see fig .
[ bm2 ] ) @xmath61.\ ] ] in particular , we are interested in the _ expected torsional rigidity of _ @xmath62 : @xmath63 since @xmath64 and @xmath65|=0 $ ] , the torsional rigidity is the expected time needed by the first brownian motion @xmath66 to hit @xmath8 $ ] averaged over all starting points in @xmath0 . as @xmath67 ,
@xmath8 $ ] tends to fill @xmath0 .
hence we expect that @xmath68 . the results in this paper
identify the speed of convergence .
this speed provides information on the random geometry of @xmath62 . in earlier work @xcite we considered the inradius of @xmath62 .
$ ] for @xmath69 and @xmath11 .
the brownian path @xmath8 $ ] is black , its complement @xmath70 $ ] is white . ]
[ bm2 ] the case @xmath71 is uninteresting . for @xmath11 ,
as @xmath9 gets large the set @xmath62 decomposes into a large number of disjoint small components ( see fig . [ bm2 ] ) , while for @xmath72 it remains connected . as shown in @xcite , in the latter case @xmath62 consists of `` lakes '' connected by `` narrow channels '' , so that we may think of it as a _
porous medium_. below we identify the asymptotic behaviour of @xmath73 as @xmath67 when @xmath74 .
for @xmath15 we have @xmath75 for all @xmath76 because brownian motion is polar . to get a non - trivial scaling , the brownian path
must be thickened to a _ shrinking wiener sausage _
@xmath77 = \big\{x\in\t^m\colon\,d_t(x ) \leq r(t)\big\ } , \qquad t>0,\ ] ] where @xmath78 is such that @xmath79 .
this choice of shrinking is appropriate because for @xmath72 typical regions in @xmath62 have a size of order @xmath80 ( see @xcite and @xcite ) .
the object of interest is the random set @xmath81,\ ] ] in particular , the _ expected torsional rigidity of _ @xmath82 : @xmath83 below we identify the asymptotic behaviour of @xmath84 as @xmath67 for @xmath15 subject to a condition under which @xmath85 does not decay too fast .
theorems [ the1][the3 ] below are our main results for the scaling of @xmath73 and @xmath84 as @xmath67 . in what follows
we write @xmath86 when @xmath87 for @xmath9 large enough .
[ the1 ] if @xmath11 , then @xmath88 [ the2 ] if @xmath13 , then @xmath89\ , \frac{2}{t^2}\,\e_0\left(\frac{1}{\cp(\beta[0,1])^2}\right ) , \qquad t\to\infty,\ ] ] where @xmath90)$ ] is the newtonian capacity of @xmath91 $ ] in @xmath18 .
all inverse moments of @xmath90)$ ] are finite .
[ the3 ] if @xmath15 and @xmath92 m\geq 5\colon & \lim\limits_{t\to\infty } \frac{t}{\log^3 t}\,r(t)^{m-4 } = \infty , \end{array } \right.\ ] ] then @xmath93\ , \frac{1}{\kappa_m\,t^{2/(m-2)}}\ , \e_0\left(\frac{1}{\cp(w_{\eps(t)}[0,1])}\right ) , \qquad t \to \infty,\ ] ] where @xmath94 , @xmath95)$ ] is the newtonian capacity of @xmath96 $ ] in @xmath20 , and where @xmath97 is the newtonian capacity of the ball with radius @xmath55 in @xmath20 , @xmath98 .
all inverse moments of @xmath95)$ ] are finite for all @xmath99 .
we expect that similar results hold when @xmath0 is replaced by a smooth @xmath1-dimensional compact connected riemannian manifold without boundary .
we further expect that the torsional rigidity satisfies a strong law of large numbers for @xmath72 but not for @xmath11 .
a key ingredient in the proof of theorem [ the3 ] is the following scaling behaviour of the capacity of the wiener sausage for @xmath15 .
let @xmath100 ) , & m = 4,\\[0.2 cm ] { \displaystyle\frac{1}{t}}\,\cp(w_1[0,t ] ) , & m \geq 5 .
\end{array } \right.\ ] ] then there exist constants @xmath101 , @xmath15 , such that @xmath102\,c_m \quad \text { in $ \p_0$-probability as } t \to \infty.\ ] ] in section [ scalcap ] we prove for @xmath103 with the help of subadditivity . for @xmath104 ,
is proven in @xcite .
we refer the reader to @xcite and @xcite for an overview of what is known about the geometry of the complement of brownian motion on the unit torus .
* theorem [ the1 ] gives the leading order decay in time of torsional rigidity ( stiffness ) of a long cylindrical beam if a brownian crack develops in the cross section ( along the entire length of the beam ) .
we expect that our result for periodic boundary conditions will not change if these are replaced by the more physical relevant free ( neumann ) boundary conditions .
theorems [ the1 ] and [ the2 ] identify the scaling of the expected torsional rigidity in low dimensions .
this scaling may be viewed in the following context .
let @xmath105 denote the distance between @xmath106 .
the distance of @xmath6 to @xmath8 $ ] is denoted by @xmath107 } d(x , y).\ ] ] the _ inradius _ of @xmath62 is the random variable @xmath108 defined by @xmath109 a detailed analysis of @xmath108 and related quantities was given in @xcite , @xcite for @xmath11 and in @xcite , @xcite for @xmath72 .
in @xcite it was shown that for @xmath11 , @xmath110 } , \qquad t\to\infty,\ ] ] while for @xmath72 , @xmath111\,\left(\frac{m}{(m-2)\kappa_m}\,\frac{\log t}{t}\right)^{1/(m-2 ) } , \qquad t\to\infty.\ ] ] a ball of radius @xmath112 in @xmath0 with @xmath112 sufficiently small has a torsional rigidity proportional to @xmath113 .
theorem [ the1 ] and show that @xmath114\ , 4(\pi t)^{1/2 } = [ 1+o(1)]\ , \log \e_0(\rho_t)^4 $ ] for @xmath11 , while theorem [ the2 ] and show that @xmath115 for @xmath13 .
thus , for @xmath11 the main contribution to the asymptotic behaviour of @xmath116 comes from the components in @xmath62 that have a size of order @xmath108 ( which are atypical ; see @xcite and @xcite ) , while for @xmath13 the main contribution to the asymptotic behaviour of @xmath73 comes from regions in @xmath62 that have a size of order @xmath117 ( which are typical ; see @xcite and @xcite ) , i.e. , most of @xmath62 contributes .
* 2 . * for @xmath11 it is shown in @xcite that @xmath118 which is a considerable sharpening of .
the proof is long and difficult .
combining with what we found in theorem [ the1 ] , we get the relation @xmath119 provided also holds in mean ( which is expected but has not been proved ) . clearly , @xmath73 is not dominated by the largest component in @xmath62 alone : smaller components contribute too as long as they have a comparable size .
the scaling in suggests that the number of such components is of order @xmath120 . in order to settle this issue
, we would need to strengthen theorem [ the1 ] to a law of large numbers .
* theorem [ the3 ] identifies the scaling of the expected torsional rigidity in high dimensions . via the scaling relation in distribution @xmath121 ) = \cp(\eps w_1[0 , \eps^{-2 } ] ) = \eps^{m-2}\cp(w_1[0 , \eps^{-2 } ] ) , \qquad \eps>0,\ ] ] it follows from that @xmath95)=[1+o(1)]\,c_m\eps^{m-4}$ ] in @xmath122-probability as @xmath123 when @xmath103 . in that case theorem [ the3 ]
yields the asymptotics @xmath124\,\frac{1}{\kappa_m c_m\,t\ , r(t)^{m-4 } } , \qquad t \to \infty.\ ] ] it also follows from that @xmath95)=[1+o(1)]\ , c_4/2\log(1/\eps)$ ] in @xmath122-probability as @xmath123 when @xmath125 . in that case theorem [ the3 ]
yields the asymptotics @xmath126\,\frac{2\log(1/t^{1/2}r(t))}{\kappa_4 c_4\,t } , \qquad t \to \infty.\ ] ] by the second half of , both and correspond to the regime where @xmath127 .
we have not attempted to improve this to @xmath128 .
* we did not investigate the regime for @xmath15 where @xmath85 decays so fast that @xmath84 diverges as @xmath67 .
in that regime , the brownian motion @xmath66 in runs around @xmath0 many times before it hits @xmath16 $ ] , and the growth of @xmath84 depends on the global rather than the local properties of @xmath16 $ ] . * 5 . * we saw in section [ background ] that the torsional rigidity is closely related to the principal dirichlet eigenvalue . in section [ torsion ] we will see a relation with the square - integrated distance function and the largest inradius . in section
[ extra ] we will give a quick proof of the following inequality relating the torsional rigidity to @xmath129 the principal dirichlet eigenvalue of @xmath62 for @xmath74 and @xmath82 for @xmath15 .
[ the4 ] ( a ) if @xmath74 , then for @xmath9 large enough , @xmath130 ( b ) if @xmath15 and @xmath131 , then for @xmath9 large enough , @xmath132 combining the result for @xmath11 with what we found in theorem [ the1 ] , we obtain @xmath133 where @xmath134 means that @xmath135 for @xmath9 large enough .
in @xcite we conjectured that @xmath136 @xmath137\ , 2(\pi t)^{1/2}$ ] , which fits the lower bound in .
however , a better estimate than is possible .
namely , in section [ torsion ] we will see that @xmath138 , and so jensen s inequality gives the lower bound @xmath139 .
assuming that the scaling in also holds in mean ( which is expected but has not been proved ) , we get @xmath140 which is better than by a factor @xmath141
. presumably captures the correct scaling behaviour . for @xmath11 ,
@xmath62 consists of countably many connected component and the expected lifetime is sensitive to the starting point .
we make use of the hardy inequality to relate the time - integrated heat content to the space integral @xmath142)^2\,dx$ ] . because of the symmetry of @xmath143 , the problem boils down to studying the distribution of @xmath144)^2 $ ] with @xmath145 chosen uniformly at random .
this can be done by using a domain perturbation formula for the dirichlet laplacian eigenvalues . for @xmath72
, @xmath62 has only one connected component and the proof is probabilistic .
the starting point is the representation
@xmath146 \cap \tilde\beta[0,s ] = \emptyset\big).\ ] ] it is easy to see that @xmath66 hits @xmath8 $ ] within time @xmath147 ) with a very high probability .
for @xmath148 , the above integrand is the probability that @xmath149 avoids the small set @xmath150 $ ] for a long time @xmath9 .
we appeal to a recursive argument to evaluate this probability . roughly speaking , in each unit of time @xmath149 hits @xmath150 $ ] with probability @xmath151)$ ]
. * outline . *
the remainder of this paper is organised as follows . in section [ torsion ]
we recall some analytical facts about the torsional rigidity . in sections [ proofthe1][proofthe3 ]
we prove theorems [ the1][the3 ] , respectively .
the proof of theorem [ the4 ] is given in section [ extra ] , while the proof of the scaling in for @xmath103 is given in section [ scalcap ] .
let @xmath24 be an @xmath1-dimensional riemannian manifold without boundary that is both geodesically and stochastically complete . in most of this paper we focus on the case where @xmath24 is the @xmath1-dimensional unit torus @xmath0 . however , the results mentioned below hold in greater generality .
we derive certain a priori estimates on the torsional rigidity that will be needed later on . for an open set @xmath152 with boundary @xmath153 , and with finite lebesgue measure @xmath154
, we denote the dirichlet heat kernel by @xmath155 , @xmath156 , @xmath56 .
recall that the dirichlet heat kernel is non - negative , monotone in @xmath5 , symmetric .
thus , we have that @xmath157 since @xmath158 , there exists an @xmath159 eigenfunction expansion for the dirichlet heat kernel in terms of the dirichlet eigenvalues @xmath160 , and a corresponding orthonormal set of eigenfunctions @xmath161 in @xmath159 : @xmath162 since @xmath163 we have that @xmath164 and @xmath165 lemma [ lemlb * ] below provides an upper bound on the dirichlet eigenfunctions in terms of the dirichlet eigenvalues .
this bound will show that the eigenfunctions are in @xmath166 , which by hlder s inequality implies that they are in @xmath167 for all @xmath168 .
lemma [ lemlb ] below states upper and lower bounds on the torsional rigidity that will be needed later on .
[ lemlb * ] suppose that @xmath169 , @xmath158 , @xmath170 for all @xmath56 .
then @xmath171 by and the domain monotonicity of the dirichlet heat kernel ( @xcite ) , we have that @xmath172 taking first the supremum over @xmath173 in the right - hand side of and subsequently in the left - hand side of , we get .
let @xmath174 denote the distance of @xmath175 to @xmath176 .
[ lemlb ] ( a ) let @xmath24 be a riemannian manifold that is both geodesically and stochastically complete .
let @xmath54 be an open subset of @xmath24 with @xmath158
. then @xmath177 ( b ) suppose that @xmath24 and @xmath54 satisfy the hypotheses in ( a )
. then @xmath178 ( c ) let @xmath179 .
then @xmath180 ( d ) let @xmath181 be simply connected and @xmath182
. then @xmath183 ( e ) let @xmath184 .
then @xmath54 can be embedded in @xmath20 if and only if @xmath185 for all @xmath186 and @xmath187 .
if @xmath188 can be embedded in @xmath189 , then @xmath190 \(a ) since the eigenfunctions are in all @xmath191 , we have by , and parseval s identity that @xmath192 inequality goes back to @xcite . for a recent discussion and further improvements
we refer the reader to @xcite .
+ ( b ) by the first identity in and , we have that @xmath193 by lemma [ lemlb * ] , we have that @xmath194 , and so @xmath195 inequality follows from , , and the fact that @xmath196 does not change sign .
+ ( c ) for every @xmath175 the open ball @xmath197 with centre @xmath6 and radius @xmath198 is contained in @xmath54 .
therefore , by domain monotonicity , the expected life time satisfies @xmath199 .
hence @xmath200 choose @xmath201 , integrate over @xmath175 and use , to get the claim .
+ ( d ) it was shown in @xcite that the dirichlet laplacian on a simply connected proper subset of @xmath189 satisfies a strong hardy inequality : @xmath202 theorem 1.5 in @xcite implies .
+ ( e ) recall that the metric on @xmath0 is given by @xmath203 note that @xmath204 because @xmath205 . if @xmath206 for all @xmath207 , then @xmath208 .
next , suppose that @xmath208 .
then @xmath209 . let @xmath210 .
then @xmath211 . we therefore conclude that @xmath212 .
finally , follows from for @xmath11 and .
in section [ invdev ] we show that the inverse of the principal dirichlet eigenvalue of @xmath213 $ ] has a finite exponential moment . in section [ secproofthe1 ]
we use this result to prove theorem [ the1 ] .
[ lambdaest ] there exists @xmath214 such that @xmath215\right ) < \infty.\ ] ] let @xmath216 denote the logarithmic capacity of a measurable set @xmath217 .
it is well known ( see @xcite ) that if @xmath218 and @xmath219 is a homothety of @xmath220 by a factor @xmath221 , then @xmath222 , \qquad \eps\downarrow 0,\ ] ] and @xmath223 , \qquad \eps\downarrow 0.\ ] ] in particular , if @xmath224 is a straight line segment of length @xmath225 , then there exists a @xmath226 such that @xmath227 since @xmath90 ) \geq \cp(l_{|\beta(1)| } ) \geq \cp(l_{(\tfrac12 \wedge |\beta(1)|)})$ ] , we get @xmath228\right ) \leq \e_0\left ( ( \tfrac12 \wedge |\beta(1)|)^{-c / c ' } \right ) \leq ( \tfrac12)^{-c / c ' } + \e_0\left ( |\beta(1)|^{-c / c ' } \right)\nonumber \\ & = ( \tfrac12)^{-c / c ' } + \int_{\r^2 } |x|^{-c / c'}\,\frac{1}{4\pi}\,e^{-|x|^2/4}\,dx,\end{aligned}\ ] ] which is finite when @xmath229 .
the proof comes in 6 steps , and is based on lemmas [ pklem1][bbl ] below .
we use the following abbreviations ( recall and ):
@xmath230 * 1 . *
note that @xmath8 $ ] is a closed subset of @xmath143 a.s .
hence @xmath62 is open and its components are open and countable .
let @xmath231 enumerate these components .
let @xmath232 and abbreviate @xmath233 it follows from the proof of lemma [ lemlb](d ) that if @xmath234 , then @xmath235 can be isometrically embedded in @xmath189 . since @xmath8 $ ] is continuous a.s .
, each @xmath235 is simply connected . since the torsional rigidity is additive on disjoint sets we have that @xmath236 * 2 . * the first term in the right - hand side of
is estimated from above by lemma [ lemlb](d ) .
this gives ( recall ) @xmath237 the second term in the right - hand side of is estimated from above by lemma [ lemlb](a ) .
this gives @xmath238 by cauchy - schwarz , this term contributes to @xmath73 at most @xmath239 to bound the probability in the right - hand side of ( [ e5.61 ] ) from above , we let @xmath240 , @xmath241 , be any open disjoint collection of squares in @xmath242 each with area @xmath243 and not containing @xmath45 .
furthermore , we let @xmath244 be the open @xmath221-neighbourhood of the union of the boundaries of these squares with @xmath245 . then @xmath91 $ ] starting at @xmath45 has a positive probability @xmath246 of making a closed loop around each of these squares and staying inside @xmath244 . translating @xmath247 such that these squares do not contain @xmath248
, we find that @xmath249 $ ] starting at @xmath248 has a positive probability @xmath250 of making a closed loop around each of these translated squares and staying inside @xmath251 .
continuing this way , by induction we find that the probability of @xmath8 $ ] not making any of these closed translated loops is at most @xmath252 , where @xmath253 denotes the integer part .
hence @xmath254 , and so @xmath255 for some @xmath256 .
we conclude that @xmath257 since @xmath258 is non - decreasing , lemma [ lambdaest ] implies that the second term decays exponentially fast in @xmath259 and therefore is harmless for the upper bound in .
* 3 . * to derive a lower bound for @xmath73 , we note that by lemma [ lemlb](e ) we have @xmath260 where in the last inequality we use that @xmath261 and @xmath262 .
we conclude by that @xmath263 the second term is again harmless for the lower bound in .
* 4 . * the estimates in and show that @xmath264 up to exponentially small error terms . in order to obtain the leading order asymptotic behaviour of @xmath265 , we make a dyadic partition of @xmath143 into squares as follows .
partition @xmath143 into four @xmath55-squares of area @xmath266 each , and proceed by induction to partition each @xmath267-square into four @xmath268-squares .
so , @xmath143 is partitioned into @xmath269 @xmath267-squares .
we define a @xmath267-square to be _ good _ when the path @xmath8 $ ] does not hit this square , but does hit the unique @xmath270-square to which it belongs . clearly , if @xmath6 belongs to a good @xmath267-square , then @xmath271 ) \leq ( 2\sqrt{2})2^{-k}$ ] .
hence , as the area of each @xmath267-square is @xmath272 , we get @xmath273 where we write @xmath274 , which is the same as @xmath275 for the quantity under consideration , by translation invariance . to estimate the right - hand side of we need three lemmas .
[ pklem1 ] for @xmath276 , let @xmath277\cap s_k)=\emptyset)$ ] , where @xmath278 is any of the @xmath267-squares .
then @xmath279 let @xmath280 be the dirichlet heat kernel for @xmath281 . by the eigenfunction expansion in
, we have that @xmath282 where we use parseval s identity in the last equality .
[ lam1 ] there exists @xmath283 such that , for all @xmath276 , @xmath284 by ( * ? ? ?
* theorem 1 ) we have that , for any disc @xmath285 with radius @xmath221 , @xmath286^{-2}\big),\qquad \epsilon \downarrow 0.\ ] ] this implies , by monotonicity and continuity of @xmath287 , the existence of @xmath288 such that @xmath289^{-2},\qquad 0 < \epsilon \leq \tfrac12.\ ] ] for @xmath290 there exist two discs @xmath291 and @xmath292 , with the same centre and radii @xmath293 and @xmath294 , such that @xmath295 .
hence @xmath296 , and follows by applying with @xmath297 and @xmath298 , respectively .
[ pklem2 ] @xmath299 let @xmath300 be the event that @xmath278 is not hit . since @xmath278 is a good @xmath267-square if and only if the event @xmath301 occurs , the lemma follows because @xmath302 .
* we are now ready to estimate @xmath303 . by and lemma [ pklem2 ] , @xmath304 = 6 \sum_{k\in\n } 2^{-2k}p_k(t),\ ] ] where @xmath305 . in order to bound this sum from above
we consider the contributions coming from @xmath306 and @xmath307 and @xmath308 , respectively , where @xmath253 denotes the integer part , and we choose @xmath309 with @xmath310 the constant in .
since @xmath311 the first contribution is exponentially small in @xmath9 .
for @xmath312 we have @xmath313 , and hence by lemmas [ pklem1][lam1 ] , @xmath314 and so the second contribution is @xmath315 .
finally , for @xmath316 we have @xmath317 , and hence @xmath318 the summand is increasing for @xmath319 and decreasing for @xmath320 .
moreover , it is bounded from above by @xmath321 .
we conclude that for @xmath10 , @xmath322,\end{aligned}\ ] ] where we use formula 3.324.1 from @xcite and formula 9.7.2 from @xcite . putting the estimates in and together
, we obtain that @xmath323.\ ] ] this is the desired upper bound in .
* 6 . * to obtain a lower bound for @xmath303 , we consider a good @xmath267-square .
this square contains a square with the same centre , parallel sides and area @xmath324 .
the distance from this square to @xmath8 $ ] is bounded from below by @xmath325 . hence @xmath326 = \tfrac{3}{64 } \sum_{k\in\n } 2^{-2k}\,p_k(t ) , \end{aligned}\ ] ] since @xmath305 .
the following lemma provides a lower bound for the right - hand side of .
[ bbl ] there exists @xmath327 such that for all @xmath328 , @xmath329 by the eigenfunction expansion in we have that @xmath330 by the results of @xcite , @xmath331 as @xmath332 .
this implies that @xmath333 for @xmath267 sufficiently large . combining , , and lemma [ bbl ] , we have that @xmath334 now let @xmath9 be such that @xmath335 .
then @xmath336 because the summand is strictly decreasing in @xmath267 , we can replace the sum over @xmath267 by an integral with a minor correction .
this gives @xmath337 we have @xmath338,\end{aligned}\ ] ] where we use once more formulas 3.324.1 from @xcite and 9.7.2 from @xcite .
combining , and , we get @xmath339.\ ] ] this is the desired lower bound in .
it is well known that @xmath91 $ ] has a strictly positive newton capacity when @xmath13 . in section [ invcap ]
we show that the inverse of the capacity of @xmath91 $ ] on @xmath18 has a finite exponential moment . in section [ linkpdecap ]
we show that for every closed set @xmath340 that has a small enough diameter the principal dirichlet eigenvalue of @xmath341 is bounded from below by a constant times the capacity of @xmath342 .
( the same is true for @xmath15 , a fact that will be needed in section [ proofthe3 ] . ) in section [ secproofthe2 ] we use these results to prove theorem [ the2 ] .
[ lemcapexp ] let @xmath13
. then there exists @xmath214 such that @xmath343)}\right]\right ) < \infty.\ ] ] we use the fact that , for any compact set @xmath344 , @xmath345.\ ] ] as test probability measure we choose the sojourn measure of @xmath8 $ ] , that is @xmath346}(c ) = \int_0 ^ 1 1_c ( \beta(t))\,dt , \qquad c \subset \r^3,\ ] ] for which @xmath347}(dx)\mu_{\beta[0,1]}(dy)}{4\pi \left\vert x - y\right\vert } = \int_0 ^ 1 ds \int_0 ^ 1 dt\,\,\frac{1}{4\pi \left\vert \beta(s)-\beta(t)\right\vert}.\ ] ] it therefore suffices to prove that @xmath348\right ) < \infty\ ] ] for small enough @xmath349 . a proof of this fact is hidden in @xcite . for the convenience of the reader we write it out here .
by cauchy - schwarz and jensen , we have that @xmath350\right ) \leq
\e_0\left(\exp\left[2c \int_0 ^ 1 ds \int_s^1 dt\,\ , \frac{1}{\left\vert \beta(s)-\beta(t)\right\vert}\right]\right ) \nonumber \\ & \leq \e_0\left(\exp\left[2c \int_0 ^ 1 ds \int_s^{1+s } dt\,\ , \frac{1}{\left\vert \beta(s)-\beta(t)\right\vert}\right]\right ) \leq \int_0 ^ 1 ds\,\,\e_0\left(\exp\left[2c \int_s^{1+s } dt\,\ , \frac{1}{\left\vert \beta(s)-\beta(t)\right\vert}\right]\right ) \nonumber \\ & = \e_0\left(\exp\left[2c \int_0 ^ 1 dt\,\,\frac{1}{\left\vert\beta(t)\right\vert } \right]\right).\end{aligned}\ ] ] it therefore suffices to prove that the right - hand side is finite for small enough @xmath349 . expanding the exponent , we get @xmath351\right ) & = \sum_{k\in\n_0 } \frac{(2c)^k}{k!}\,\e_0\left(\left[\int_0 ^ 1 dt\,\ , \frac{1}{\left\vert\beta(t)\right\vert}\right]^k\right ) \nonumber \\ & = \sum_{k\in\n_0 } ( 2c)^k \int_{0 \leq t_1<\cdots < t_k \leq 1 } \e_0\left(\frac{1}{\left\vert\beta(t_1)\right\vert \times\cdots\times \left\vert\beta(t_k)\right\vert}\right ) dt_1 \times \cdots \times dt_k.\end{aligned}\ ] ] the integrand equals @xmath352 \right\vert}\right\vert \mathcal{f}_{t_{k-1}}\right)\right),\ ] ] where @xmath353 is the sigma - algebra of @xmath149 up to time @xmath9 . however , @xmath354 \right\vert}\right\vert \mathcal{f}_{t_{k-1}}\right ) = \left . \e_0\left(\frac{1}{\left\vert x+\sqrt{t_{k}-t_{k-1 } } \beta(1)\right\vert}\right ) \right\vert_{x=\beta(t_{k-1 } ) } \nonumber \\ & \leq \sup_{x\in\r^3 } \e_0\left(\frac{1}{\left\vert x+\sqrt{t_{k}-t_{k-1 } } \beta(1)\right\vert}\right ) \leq \e_0\left(\frac{1}{\left\vert\sqrt{t_{k}-t_{k-1 } } \beta(1)\right\vert } \right ) \leq \frac{\gamma}{\sqrt{t_{k}-t_{k-1 } } } \label{dec2}\end{aligned}\ ] ] with @xmath355 , where in the second inequality we use that @xmath356 is stochastically larger than @xmath357 for any @xmath358 . iterating , , we get @xmath359 where @xmath360 . hence @xmath361\right ) & = \sum_{k\in\n_0 } ( 2c)^k \gamma^k \int_{0 \leq t_1<\cdots < t_k \leq 1 } \frac{dt_1}{\sqrt{t_1 } } \times \cdots \times \frac{dt_k}{\sqrt{t_k - t_{k-1 } } } \nonumber\\ & \leq \sum_{k\in\n_0 } ( 2c)^k \gamma^k \left(\int_0 ^ 1 dt\,\,\frac{1}{\sqrt{t}}\right)^k = \sum_{k\in\n_0 } ( 4c)^k \gamma^k,\end{aligned}\ ] ] which is finite for @xmath362 .
[ eigcap ] let @xmath72 , and let @xmath342 be a closed subset of @xmath0 with @xmath363 .
then @xmath364 where @xmath365 and @xmath366 is the newtonian capacity of @xmath342 embedded in @xmath20 .
since @xmath367 , @xmath342 can be embedded in @xmath20 by lemma [ lemlb](e ) .
we let @xmath368 , identify @xmath369 with @xmath0 , and define @xmath370 by @xmath371 .
let @xmath196 be the first eigenfunction on @xmath372 with dirichlet boundary conditions on @xmath342 , and let @xmath373 be the corresponding first dirichlet eigenvalue . then @xmath374 integrating both sides of this identity over @xmath375 , we get @xmath376 where @xmath377 is the first hitting time of @xmath342 by brownian motion on @xmath0 .
it follows that for any @xmath56 , @xmath378 where we use the inequality @xmath379 , @xmath380 .
let @xmath381 be brownian motion on @xmath20 , and let @xmath382 be the first hitting time of @xmath383 by @xmath381 .
then @xmath384 where @xmath385 is the last exit time from @xmath342 by @xmath381 .
let @xmath386 denote the equilibrium measure on @xmath342 in @xmath20
. then ( see @xcite ) @xmath387 by , @xmath388 but @xmath389 for @xmath390 and @xmath391 .
hence the right - hand side of is bounded from below by @xmath392 .
we now get the claim by choosing @xmath393 in . we note that if @xmath13 and @xmath394 is a closed ball with radius @xmath221 , then @xmath395 $ ] as @xmath396 ( see @xcite ) . in that case , since @xmath397 , we see that the constant in is off by a large factor .
write , recalling , , , , and using fubini s theorem , @xmath398}\big ) = ( \e_0 \otimes \tilde\e)\left(\int_0^\infty ds\ , 1_{\{\tilde\tau_{\t^3\backslash \beta[0,t]}>s\}}\right ) = ( \e_0 \otimes \tilde\e)\left(\int_0^\infty ds\ , 1_{\{\tilde\beta[0,s ] \cap \beta[0,t ] = \emptyset\}}\right),\ ] ] where @xmath399 denotes expectation over @xmath66 with @xmath400 drawn uniformly from @xmath401 . by symmetry
, we may replace @xmath402 by @xmath403 .
the proof comes in 7 steps . in steps 1 - 2
we show that for a suitable @xmath404 , tending to zero as @xmath67 , @xmath405 \int_0^{\eta(t ) } ds\ , ( \p \otimes \tilde\p)(\beta[0,t ] \cap \tilde\beta[0,s ] = \emptyset).\ ] ] heuristically , the domain perturbation formula gives that @xmath406 \cap \tilde\beta[0,s ] = \emptyset ) & = \tilde\e\left[\exp\left\{-\lambda_1(\t^m \setminus \tilde\beta[0,s])[1+o(1)]\right\}\right]\\ & = \tilde \e\left[\exp\left\{-t\,\cp(\tilde\beta[0,s])[1+o(1)]\right\}\right]\\ & = \tilde\e\left[\exp\left\{-ts^{1/2}\,\cp(\tilde\beta[0,1])[1+o(1)]\right\}\right ] .
\end{aligned}\ ] ] substituting into and using the laplace principle , we get .
the details are made precise in steps 3 - 7 .
pick @xmath407 such that @xmath408 we begin by showing that the integral over @xmath409 decays faster than any negative power of @xmath9 and therefore is negligible . indeed , for any @xmath410 we have , by the spectral decomposition in , @xmath411 \cap \beta[0,t ] = \emptyset\}}\right ) \leq \tilde\e_0\left(\int_{\eta(t)}^{k(t ) } ds\,e^{-t \lambda_1(\t^3\backslash\tilde\beta[0,s])}\right).\ ] ] by lemma [ eigcap ] , @xmath412 for every closed set @xmath413 ( in the lower bound we interpret @xmath220 as a subset of @xmath18 ) .
hence the right - hand side of is bounded from above by @xmath414 \cap b_{1/4}(0))}\right),\ ] ] where we use that @xmath415 ) \geq \cp(\tilde\beta[0,\eta(t)])$ ] for @xmath416 . in step 2
we show that @xmath417 \subsetneq b_{1/4}(0))$ ] decays faster than any negative power of @xmath9 .
hence we may replace @xmath418 \cap b_{1/4}(0))$ ] by @xmath418)$ ] in at the cost of a negligible error term @xmath419 .
next , we note that @xmath418)$ ] is equal to @xmath420)$ ] in distribution . moreover , since @xmath421 for all @xmath422 , we have , for any @xmath349 , @xmath423 ) } & = e^{-c_3 t \sqrt{\eta(t)}\,\cp(\tilde\beta[0,1 ] ) - c\,\cp(\tilde\beta[0,1])^{-1 } } e^{c\,\cp(\tilde\beta[0,1])^{-1}}\\ & \leq e^{-2\sqrt{c_3ct\sqrt{\eta(t ) } } } e^{c\,\cp(\tilde\beta[0,1])^{-1}}. \end{aligned}\ ] ] by lemma [ lemcapexp ] , we therefore have @xmath424)}\right ) \leq c\,e^{-2\sqrt{c_3ct\sqrt{\eta(t ) } } } + o(t^{-2})\ ] ] for some @xmath283 and @xmath349 small enough . hence is @xmath425 when we pick @xmath426 the second half of ensures that @xmath427 grows faster than any positive power of @xmath9 , and
so we conclude that the integral in the left - hand side of is @xmath419 . to estimate @xmath428 \cap \beta[0,t ] = \emptyset\}}\right)\ ] ] we reverse the roles of @xmath149 and @xmath66 , and do the same estimate using that @xmath429 ) \geq \cp(\beta[0,\eta(t)])$ ] for @xmath430 .
this leads to @xmath431 \cap \beta[0,t ] = \emptyset\}}\right ) & \leq c \int_{k(t)}^\infty ds\ , e^{-2\sqrt{c_3cs\sqrt{\eta(t ) } } } + o(t^{-2})\\ & = [ 1+o(1)]\ , c\sqrt{\frac{k(t)}{c_3c\sqrt{\eta(t)}}}\,e^{-2\sqrt{c_3ck(t)\sqrt{\eta(t ) } } } + o(t^{-2 } ) , \end{aligned}\ ] ] in which the first term is even much smaller than @xmath419 .
* 2 . * the probability that @xmath66 leaves the ball of radius @xmath432 prior to time @xmath404 decays faster than any negative power of @xmath9 when @xmath433 . indeed , by lvy s maximal inequality ( theorem 3.6.5 in @xcite ) , @xmath434\colon\,\tilde\beta[0,s ] \notin b_{\tilde\eta(t)}(0)\big ) \leq 2\ , \tilde\p_0\big(\tilde\beta(\eta(t ) ) \notin b_{\tilde\eta(t)}(0)\big)\\ & = o(\exp\big[-\tfrac18\tilde\eta^2(t)/\eta(t)\big ] ) = o(\exp[-\tfrac18 m(t ) \log t ] ) = o(t^{-\tfrac18 m(t ) } ) = o(t^{-2 } ) .
\end{aligned}\ ] ] hence , with a negligible error we may restrict the expectation in the right - hand side of to the event @xmath435 \subset b_{\tilde\eta(t)}(0)\}.\ ] ] the first half of guarantees that @xmath436 for some choice of @xmath437 with @xmath438 .
fix @xmath439 , and consider the successive excursions of @xmath149 between the boundaries of the balls @xmath440 and @xmath441 , i.e. , put @xmath442 and , for @xmath443 , @xmath444 for @xmath443 , let @xmath445)$ ] denote the @xmath267-th excursion from @xmath446 to @xmath447 and back .
let @xmath448 denote the location where this excursion first hits @xmath447 . clearly , under the law @xmath449
, @xmath450 is a uniformly ergodic markov chain on @xmath451 .
let @xmath452 be the number of completed excursions prior to time @xmath9 . by the renewal theorem ,
we have @xmath453 moreover , for every @xmath454 there exists a @xmath455 such that @xmath456 where we have used the fact that @xmath457 and @xmath458 have finite exponential moments .
fix @xmath459 \subset b_{\tilde\eta(t)}(0)$ ] . for @xmath460 $ ] and @xmath461 , the probability that the first @xmath462 excursions do not hit @xmath150 $ ] equals @xmath463\big ) = \e\left(\prod_{k=1}^n 1_{\{\tilde\beta[0,s ] \cap \beta_k = \emptyset\}}\right ) = \e\left(\e\left ( \prod_{k=1}^n \big[1-p\big(\bar x_k,\bar x_{k+1};\tilde\beta[0,s]\big)\big ] ~\bigg|~ \cf_{n+1 } \right)\right),\ ] ] where @xmath464 is the sigma - algebra generated by @xmath465 , @xmath466 , and @xmath467\big ) = \p_x^y\left(\sigma_{\tilde\beta[0,s]}<\infty\right )
, \qquad x , y \in \partial b_\delta(0),\ ] ] is the probability that a brownian motion , starting from @xmath468 and conditioned to re - enter @xmath441 at @xmath469 after it has exited @xmath440 , hits @xmath150 $ ] .
the following lemma gives a sharp estimate of @xmath470)$ ] .
[ hitest ] if @xmath471 and @xmath459 \subset b_{\tilde\eta(t)}(0)$ ] , then @xmath472\big ) = [ 1+o(\delta ) ] \left\{(\kappa_3\delta)^{-1 } \cp(\tilde\beta[0,s ] ) + o(\delta^{-2})\,\tilde\eta^2(t)\right\ } , \qquad \delta\downarrow 0,\ ] ] for all @xmath473 and @xmath460 $ ] .
we begin by showing that if @xmath471 , then @xmath474}<\infty\big ) - ( \kappa_3\delta)^{-1}\cp(\tilde\beta[0,s])\big| \leq 2\delta^{-2 } \tilde\eta^2(t)\ ] ] for all @xmath475 and @xmath150 \subset b_{\tilde\eta(t)}(0)$ ] .
indeed , for any compact set @xmath476 , we have @xmath477 where @xmath386 is the equilibrium measure on @xmath342 ( see @xcite , @xcite , @xcite ) . if @xmath478 and @xmath479 , then @xmath480 .
hence yields the estimate @xmath481 , provided @xmath482 . in that case @xmath483 , and the claim in follows .
furthermore , since @xmath484 for all @xmath485 , we have @xmath486}<\infty\big ) - \inf_{y \in \partial b_\delta(0)}p\big(x , y;\tilde\beta[0,s]\big ) \leq \kappa_3 ( 4\delta ) \sup_{y , z
\in \partial b_\delta(0 ) } p\big(y , z;\tilde\beta[0,s]\big ) , \qquad \forall\,x \in \partial b_\delta(0).\ ] ] hence implies . *
* recalling , we have @xmath487 \cap \beta[0,\sigma_0 ] = \emptyset\ } } \,\prod_{k=1}^{n_\delta(t)+1 } 1_{\{\tilde\beta[0,s ] \cap \beta_k = \emptyset\ } } \leq 1_{\{\tilde\beta[0,s ] \cap \beta[0,t ] = \emptyset\ } } \leq \prod_{k=1}^{n_\delta(t ) } 1_{\{\tilde\beta[0,s ] \cap \beta_k = \emptyset\}}.\ ] ] in terms of the probability defined in , and with the help of the large deviation estimate in , this sandwich gives us , on the event @xmath488 , @xmath489 \cap \beta[0,t ] = \emptyset\}}\right ) = o\left(\eta(t)\,e^{-c_\delta(\delta')t}\right ) + [ 1+o_t(1)]\ , \pi\big([1+o_t(1)](e_\delta+e'_\delta)^{-1}t;\tilde\beta[0,s]\big),\ ] ] where the error terms @xmath490 tend to zero as @xmath67 ( here we use that @xmath491 = \emptyset)=1 $ ] ) .
combining the estimates in steps 15 , and using that @xmath415)$ ] equals @xmath492)\sqrt{s}$ ] in distribution under @xmath493 , we get @xmath494\,\tilde\e_0 \left(\int_0^{\eta(t ) } ds\,e^{-a_\delta(t)\sqrt{s}}\right)\\ & = o(t^{-2 } ) + [ 1+o_t(1)]\,\tilde\e_0\left(\frac{2}{a_\delta(t)^2}\ , \left\{1+e^{-a_\delta(t)\sqrt{\eta(t)}}\,[a_\delta(t)\sqrt{\eta(t)}-1]\right\}\right ) , \end{aligned}\ ] ] with @xmath495\,[1+o_t(1)]\,(e_\delta+e'_\delta)^{-1}\,(\kappa_3\delta)^{-1}\cp(\tilde\beta[0,1 ] ) , \qquad t \to \infty.\ ] ] the term between braces in is bounded , and tends to @xmath55 in @xmath493-probability as @xmath67 because of the first half of . therefore , lead us , for fixed @xmath496 , to @xmath497\,2(\kappa_3\delta)^2(e_\delta+e'_\delta)^2\,\tilde\e_0\left(\frac{1}{\cp(\tilde\beta[0,1])^2}\right),\ ] ] where we have used lemma [ lemcapexp ] .
the latter also implies that the expectation in the right - hand side is finite . *
7 . * finally ,
letting @xmath498 and using that @xmath499 we arrive at @xmath500)^2}\right).\ ] ] this proves the claim in .
the same estimates as in the proof of theorem [ the2 ] for @xmath13 in section [ secproofthe2 ] can be used to prove theorem [ the3 ] for @xmath15 after we replace @xmath150 $ ] by @xmath501 $ ] .
the details are explained in sections [ proofthe35ormore][proofthe34 ] . in the proof
we assume that @xmath502 * 1 - 2 . * the estimates in steps 12 are sharp enough to produce a negligible error term @xmath503 when is replaced by @xmath504 where we note that by the second half of there exists a choice of @xmath404 satisfying .
indeed , the analogues of give ( recall that lemma [ eigcap ] also holds for @xmath15 ) @xmath505 \cap \beta[0,t ] = \emptyset\}}\right ) \leq k(t)\ , \tilde\e_0\left(e^{-c_mt\,\cp(\tilde{w}_{r(t)}[0,\eta(t ) ] \cap b_{1/4}(0))}\right),\ ] ] where we use that @xmath506 ) \geq \cp(\tilde{w}_{r(t)}[0,\eta(t)])$ ] for @xmath507 .
the estimate in step 2 shows that , because of the first half of , @xmath508 with @xmath488 defined in decays faster than any negative power of @xmath9 , so that we can remove the intersection with @xmath440 at the expense of a negligible error term . since @xmath509)$ ] equals @xmath510)$ ] in distribution under @xmath493 , we obtain that @xmath511)}\right ) = \tilde\e_0\left(e^{-c_m\,t\,r(t)^{m-2}\,\cp(\tilde{w}_1[0,\eta(t)/r(t)^2])}\right).\ ] ] via an estimate similar as in with @xmath512 replaced by @xmath513 , we obtain , with the help of lemma [ lemcapest * ] below ( which is the analogue of lemma [ lemcapexp ] and is proved in section [ invcapt ] ) , @xmath514)}\right ) \leq c\,e^{-2\sqrt{c_mc\ , t\,r(t)^{m-2}\,\eta(t)/r(t)^2 } } + o(t^{-2/(m-2)}).\ ] ] hence the right - hand side of is @xmath425 when we pick @xmath515 the second half of ensures that @xmath427 grows faster than any positive power of @xmath9 , and so is negligible .
the contribution @xmath516 \cap \beta[0,t ] = \emptyset\}}\right)\ ] ] can again be estimated in a similar way by reversing the roles of @xmath149 and @xmath66 .
this leads to a term that is even much smaller . *
* step 3 is unaltered . in step 4 the term
@xmath517 is to be replaced by @xmath518 , because in the term @xmath519 is to be replaced by @xmath520)$ ] equals @xmath521)$ ] in distribution under @xmath493 .
this gives @xmath522 where @xmath523 with @xmath524\,[1+o_t(1)]\,(e_\delta+e'_\delta)^{-1}\,t\,\delta^{-(m-2)}\ , \cp(\tilde w_{r(t)/\sqrt{s}}[0,1]).\ ] ] with the change of variable @xmath525 , the integral becomes @xmath526 where @xmath527 with ( recall ) @xmath528\,[1+o_t(1)]\,(e_\delta+e'_\delta)^{-1}\,\delta^{-(m-2)}\ , \cp(\tilde w_{\epsilon(t)/u}[0,1]),\ ] ] and where @xmath529 .
now , tells us that @xmath530 ) = [ 1+o(1)]\,u^{-(m-4)}\cp(\tilde w_{\epsilon(t)}[0,1 ] ) \quad \text{in $ \p_0$-probability as $ t\to\infty$}\ ] ] for every @xmath531 and @xmath103 , where we use that @xmath532 by the first half of .
therefore with the help of and dominated convergence , we find that @xmath533\ , \tilde\e_0\left(2 \int_0^\infty du\,u\,e^{-a''_\delta(t)\,u^2}\right ) = [ 1+o(1)]\,\tilde\e_0\left(\frac{1}{a''_\delta(t)}\right ) , \qquad t \to \infty,\ ] ] with @xmath534\,[1+o_t(1)]\,(e_\delta+e'_\delta)^{-1}\,\delta^{-(m-2)}\ , \cp(\tilde w_{\epsilon(t)}[0,1]).\ ] ] in step 7 the first line in is replaced by the statement that @xmath535 .
combining , and , and letting @xmath536 , we get the scaling in . in the proof
we assume that @xmath537 * 1 - 2 . * the estimates in steps 12 are sharp enough to produce a negligible error term @xmath503 when is replaced by @xmath538 where we note that by the second half of there exists a choice of @xmath404 satisfying .
the estimate uses with @xmath512 replaced by @xmath539 , and also lemma [ lemcapest * ] below ( which is the analogue of lemma [ lemcapexp ] and is proved in section [ invcapt ] ) . *
these steps are unaltered .
* these steps are unaltered : , [ capscalm4 ] tell us that @xmath530)= [ 1+o(1)]\,\cp(\tilde w_{\epsilon(t)}[0,1 ] ) \quad \text{in $ \p_0$-probability as $ t\to\infty$}\ ] ] for every @xmath531 , where we use that @xmath540 by the first half of .
this is used in , to get , with @xmath104 .
by a direct calculation via the fourier transform , we have that the dirichlet heat kernel on @xmath0 is given by ( recall the notation in section [ background ] ) @xmath541 where @xmath542 .
it follows that @xmath543 by translation invariance , @xmath544 is independent of @xmath6 , and we will denote it by @xmath545 . by the eigenfunction expansion in with @xmath546 and @xmath547 $ ] , and by the monotonicity of the dirichlet heat kernel , we have for @xmath548 , @xmath549 where we abbreviate @xmath550 as in . taking the supremum over @xmath6
, we obtain @xmath551 by lemma [ lemlb](b ) we have , for @xmath548 , @xmath552 since @xmath553 is convex for every @xmath548 , jensen gives that @xmath554 for @xmath555 this reads @xmath556 since the right - hand side of increases to infinity as @xmath67 , there exists @xmath557 such that @xmath558 for @xmath559 .
we now put @xmath560 and note that @xmath561 for @xmath559 . by and
, we find that , for @xmath559 , @xmath562 we conclude that , for @xmath559 , @xmath563
in section [ invcapt ] we derive the analogue of lemma [ lemcapexp ] , showing that the inverse of @xmath564 for @xmath15 defined in has a finite exponential moment uniformly in @xmath565 . in section [ proofcapscal ] we prove for @xmath103 .
the proof is similar to that of lemma [ lemcapexp ] . for any compact set @xmath567
, we use the representation ( compare with ) @xmath568.\ ] ] as test probability measure we choose the sojourn measure of @xmath19 $ ] , namely , @xmath569 } = \frac{1}{t } \int_0^t \nu_{\beta(s)}\,ds \quad \text { with } \quad \nu_z(dx ) = \frac{1}{\omega_m}\,1_{b_1(z)}(x)\,dx , \quad z \in \r^m,\ ] ] where @xmath570 .
since @xmath571 has support in @xmath19 $ ] , we have @xmath572 ) } \leq \frac{1}{\kappa_m\omega_m^2 t^2 } \int_0^t du \int_0^t dv\ , \int_{b_1(0 ) } dx \int_{b_1(0 ) } dy\,\ , \frac{1}{\left\vert\beta(u)+x-\beta(v)-y\right\vert^{m-2}}.\ ] ] moreover , there exists @xmath573 such that for all @xmath574 and @xmath575 , @xmath576 let @xmath577 .
we have that @xmath578 & \leq \exp\left[\frac{\bar{c}}{t } \int_0^t du \int_0^t dv\,\,\frac{1}{\left\vert\beta(u)-\beta(v)\right\vert^{m-2 } \vee 1}\right]\\ & \leq \frac{1}{t } \int_0^t du\,\exp\left[\bar{c } \int_0^t dv\,\,\frac{1}{\left\vert\beta(u)-\beta(v)\right\vert^{m-2 } \vee 1}\right]\\ & \leq \frac{1}{t } \int_0^t du\,\exp\left[\bar{c } \int_\r dv\,\,\frac{1}{\left\vert\beta(u)-\beta(v)\right\vert^{m-2 } \vee 1}\right ] .
\end{aligned}\ ] ] taking the expectation and using the translation invariance of brownian motion , we obtain the @xmath9-independent bound @xmath579\right ) & \leq \e_0\left(\exp\left[\bar{c } \int_\r dv\,\,\frac{1}{\left\vert\beta(v)\right\vert^{m-2 } \vee 1}\right]\right)\\ & \leq \e_0\left(\exp\left[2\bar{c } \int_0^\infty dv\,\ , \frac{1}{\left\vert\beta(v)\right\vert^{m-2 } \vee 1}\right]\right ) , \end{aligned}\ ] ] and so it remains to show that the right - hand side is finite for @xmath512 small enough .
arguing in the same way as in the proof of lemma [ lemcapexp ] , we obtain @xmath580\right ) \\ & \leq \sum_{k \in \n_0 } ( 2\bar{c})^k \,\e_0\left(\int_{0 \leq v_1 < \cdots < v_k < \infty } \prod_{i=1}^k \frac{dv_i}{\left\vert\beta(v_i)\right\vert^{m-2 } \vee 1}\right ) \\ & \leq \sum_{k \in \n_0 } ( 2\bar{c})^k \left[\int_0^\infty dv\,\ , \e_0\left(\frac{1}{\left\vert\beta(v)\right\vert^{m-2 } \vee 1}\right)\right]^k .
\end{aligned}\ ] ] therefore it remains to prove the finiteness of the integral .
that that end , we estimate @xmath581 where the last inequality holds because @xmath582 .
let @xmath583 , and replace by @xmath584 \leq \frac{1}{t } \int_0^t du\,\exp\left[\frac{\bar{c}}{\log t } \int_{u - t}^{u+t } dv\,\,\frac{1}{\left\vert\beta(u)-\beta(v)\right\vert^2 \vee 1}\right],\ ] ] and by @xmath585\right ) \leq \e_0\left(\exp\left[\frac{2\bar{c}}{\log t } \int_0^t dv\,\ , \frac{1}{\left\vert\beta(v)\right\vert^2 \vee 1}\right]\right),\ ] ] and by @xmath586\right ) \\
\leq \sum_{k \in \n_0 } \left(\frac{2\bar{c}}{\log t}\right)^k \left[\int_0^t dv\,\ , \e_0\left(\frac{1}{\left\vert\beta(v)\right\vert^2 \vee 1}\right)\right]^k,\ ] ] and by @xmath587 for some @xmath226 . note that capacity is subadditive : @xmath589 ) \leq \cp(w_1[0,s ] ) + \cp(w_1[s , s+t])$ ] for all @xmath590 . hence , kingman s subadditive ergodic theorem yields that @xmath591 ) = \bar{c}_m \quad \beta - a.s.\ ] ] for some @xmath592 .
we therefore get the claim with @xmath593 , provided we show that @xmath594 . in view of , we can get a lower bound on capacity by choosing a test probability measure .
we again choose the sojourn measure of @xmath19 $ ] in .
this gives @xmath595 ) } \leq t \int_{\r^m } \int_{\r^m } \frac{\mu_{w_1[0,t]}(dx ) \mu_{w_1[0,t]}(dy)}{\kappa_m|x - y|^{m-2 } } = \frac{1}{t } \int_0^t du \int_0^t dv \int_{\r^m } \int_{\r^m } \frac{\nu_{\beta(u)}(dx ) \nu_{\beta(v)}(dy)}{\kappa_m|x - y|^{m-2}}.\ ] ] now , there exists a @xmath283 such that @xmath596 hence @xmath595 ) } \leq \frac{1}{t } \int_0^t du \int_0^t dv \,\ , \frac{c}{|\beta(u)-\beta(v)|^{m-2 } \vee 1}.\ ] ] to prove that @xmath594 it suffices to show that the right - hand side has a finite expectation .
to that end , we estimate @xmath597 and note that , as shown in , the integral converges as @xmath67 when @xmath103 . m. t. kohler - jobin , une mthode de comparaison isoprimtrique de fonctionnelles de domaines de la physique mathmatique .
cas inhomogne : une ingalit isoprimtrique entre la frquence fondamentale dune membrane et lnergie dquilibre dun problme de poisson .
( french ) z. angew . math .
phys . 29 ( 1978 ) 767776 . | let @xmath0 be the @xmath1-dimensional unit torus , @xmath2 .
the torsional rigidity of an open set @xmath3 is the integral with respect to lebesgue measure over all starting points @xmath4 of the expected lifetime in @xmath5 of a brownian motion starting at @xmath6 . in this paper
we consider @xmath7 $ ] , the complement of the path @xmath8 $ ] of an independent brownian motion up to time @xmath9 .
we compute the leading order asymptotic behaviour of the expectation of the torsional rigidity in the limit as @xmath10 . for @xmath11
the main contribution comes from the components in @xmath12 $ ] whose inradius is comparable to the largest inradius , while for @xmath13 most of @xmath14 $ ] contributes .
a similar result holds for @xmath15 after the brownian path is replaced by a shrinking wiener sausage @xmath16 $ ] of radius @xmath17 , provided the shrinking is slow enough to ensure that the torsional rigidity tends to zero .
asymptotic properties of the capacity of @xmath8 $ ] in @xmath18 and @xmath19 $ ] in @xmath20 , @xmath15 , play a central role throughout the paper .
our results contribute to a better understanding of the geometry of the complement of brownian motion on @xmath0 , which has received a lot of attention in the literature in past years .
0.5truecm _ ams _ 2000 _ subject classifications .
_ 35j20 , 60g50 .
+ _ key words and phrases . _
torus , laplacian , brownian motion , torsional rigidity , inradius , capacity , spectrum , heat kernel . _ acknowledgment . _
the authors acknowledge support by the leverhulme trust through international network grant _ laplacians , random walks , bose gas , quantum spin systems_. eb is supported by snsf - grant 20 - 100536/1 .
fdh is supported by erc advanced grant 267356-varis and nwo gravitation grant 024.002.003-networks .
the authors thank greg lawler for helpful discussions about capacity of the wiener sausage . |
understanding correlation effects is one major goal of condensed matter physics .
strong correlations manifest themselves in various forms . the paramagnetic mott - hubbard metal - insulator - transition ( pmit)@xcite is a well - known and interesting example . with increasing interaction strength the fermi liquid state breaks down at a critical value and an insulator
is formed .
another fundamental example is magnetism , where electrons reduce the energetic cost of the coulomb - interaction by ordering .
both effects can of course occur simultaneously and are the heart of the extremely rich phase diagram of e.g. transition metal compounds like for example v@xmath3o@xmath4 or latio@xmath4@xcite . besides strong correlations , another major ingredient for the understanding of the phase diagram of compounds like v@xmath3o@xmath4 is frustration .
v@xmath3o@xmath4 crystallizes in the corundum structure with the v - ions located on a honeycomb lattice in the ab - plane , while along the c - axis a more complicated coordination is observed , which induces frustration of the magnetic interactions @xcite .
nevertheless does the phase diagram of v@xmath3o@xmath4 show an antiferromagnetic phase at temperatures below @xmath5 . upon doping with ti one
may suppress this order .
such a doping with a smaller ion can be viewed as internal pressure,@xcite hence the suppression of the magnetic order is commonly interpreted as happening through an increase of the bandwidth respectively a decrease of the correlation effects .
consequently , the critical ti doping is conventionally related to the existence of a lower critical value of the electronic interaction parameter . at higher temperatures
the antiferromagnetic state becomes unstable towards a paramagnet and one can eventually observe a paramagnetic metal - insulator - transition up to temperatures @xmath6 .
frustration is a quite common feature in real materials .
very interesting examples for frustrated systems are layered organic compounds like @xmath7-(bedt - ttf)2x@xcite .
they have a similar phase diagram as the high - temperature superconductors ( htsc)@xcite .
the phases of these organic systems are controlled by pressure and frustration rather than by doping as in htsc@xcite .
they are usually described by an anisotropic triangular lattice , and changing the anion ( x ) in these systems modifies the frustration of the lattice .
besides superconductivity also magnetic ordering and a pmit can be found .
these two examples by no means exhaust the zoo of materials showing such interplay or competition between pmit and ordered phases@xcite .
for example , rare - earth compounds like ce(rh , ir)@xmath8(co , ir)@xmath9in@xmath10 do show a similarly bizarre phase diagram@xcite .
besides their usually complicated lattice structure another challenge for a theoretical description of such compounds is that the presence of elements with partially filled d- or f - shells in principle requires a multi - orbital description to account for effects like hund s or spin - orbit coupling properly .
furthermore the residual degeneracies in the solid state crystalline environment lead to degenerate multiplets which in turn can give rise to even more complex structures like orbital order or polaron formation ( see e.g. @xcite for an overview and references ) .
although all these ingredients play an important role for a quantitative theoretical description of transition - metal or rare - earth compounds , we here want to focus on the one - orbital situation , in particular on the relation between pmit and antiferromagnetism .
this restriction to a simpler but by no means trivial situation will enable us to investigate the relation between these two paradigms of correlation effects with a small and controllable set of parameters and thus obtain some hint to how both phases interact .
a model suitable for analyzing this kind of physics is provided through the hubbard model@xcite @xmath11 where @xmath12 creates ( annihilates ) an electron with spin @xmath13 at site @xmath14 and @xmath15 is the density operator for spin up(down ) at site i. the parameters @xmath16 represent the hopping amplitude from @xmath14 to @xmath17 and @xmath18 is the interaction strength . in this paper
we will measure the interaction relative to the bandwidth , which is related to the hopping amplitude .
although at first sight very simplistic , this model is highly nontrivial .
besides other methods , especially in one dimension , progress in understanding its physics was achieved by the development of the dynamical mean field theory ( dmft)@xcite . the dmft is a very powerful tool for analyzing strongly correlated lattice systems , mapping the lattice problem onto a quantum impurity problem , which has to be solved self consistently
. for solving this impurity problem for arbitrary interaction strengths and temperatures we here use wilson s numerical renormalization group@xcite .
an interesting fact is that the only information about the lattice structure , which enters a dmft self consistency calculation , is the local density of states ( dos ) of the non - interacting system .
we performed our calculations for a bethe lattice with nearest - neighbor ( nn ) and next - nearest - neighbor ( nnn ) hopping @xmath0 and @xmath1 , respectively . .
observe the van - hove singularity at the lower band edge . @xmath19 and @xmath20 are scaled with the bandwidth @xmath21.,scaledwidth=45.0% ] the dos in this case can be calculated using a topological ansatz@xcite .
starting from a particle - hole symmetric dos at @xmath22 the density of states becomes now asymmetric with increasing @xmath1 ( see fig .
[ frustdos ] ) and develops a van - hove singularity at the lower band edge for positive and increasing nnn - hopping @xmath1 .
in contrast to @xmath22 , where the particle - hole symmetry can be employed to fix the filling at @xmath23 precisely , the asymmetry present for @xmath24 makes it more difficult to perform calculations with the filling kept at @xmath23 with sufficient accuracy .
thus dmft calculations typically take much longer here due to the necessary adjustment of the chemical potential .
of course the bethe lattice does not represent a lattice realized in real materials .
however , in contrast to the hypercubic lattice with infinite coordination number the bethe lattice has a compact support and thus possesses band edges , which provides a more realistic scenario . since the early days of dmft
, there have been many contributions by different groups to the subject of the pmit and antiferromagnetism.@xcite however , frustration effects up to now where introduced in dmft typically within the so - called two - sublattice fully frustrated model@xcite , which results in a particle - hole symmetric dos even with frustration . as side effect , this way of introducing frustration leaves the paramagnetic phase unchanged . for the non - frustrated system
the pmit is then completely covered by the antiferromagnetic phase , which exists for half - filling for all finite values of @xmath18@xcite . for the frustrated system ,
on the other hand , there exists a lower critical value for the interaction @xmath18 , which increases with increasing frustration .
it was furthermore found that the nel - temperature decreases with increasing frustration such that the pmit outgrows the antiferromagnetic phase@xcite . in early calculations using this way of introducing frustration based on exact diagonalization studies of the two - sublattice
fully frustrated model@xcite , the authors also found parameter regions in the phase diagram where an antiferromagnetic metal appeared to be stable .
however , this antiferromagnetic metal phase was later traced back to numerical subtleties in the exact diagonalization procedure and shown to be actually absent from the phase diagram@xcite .
the first attempt to study the hubbard model on the bethe lattice with correct inclusion of nn- and nnn - hopping has been performed rather recently@xcite . in this work
the authors concentrated on the paramagnetic pmit and found phase - separation between the insulating and metallic phase . in this paper
we investigate the pmit as well as antiferromagnetism and concentrate on the competition between the paramagnetic phase including the pmit and the antiferromagnetic phase at intermediate and high grades of frustration .
we especially look at the case @xmath25 and raise the question , if the scenario of the outgrowing pmit , proposed in @xcite , still holds for the correct asymmetric density of states .
the paper is arranged as following .
after this introduction we start with a brief look at the pmit , followed by a discussion of the phase diagram at half - filling including antiferromagnetism and the pmit .
the next paragraph addresses especially the case of very strong frustration and the question how the magnetic order is realized there . the paper will be closed by a summary of our results and an outlook .
the metal insulator transition for the bethe lattice with nnn - hopping has been analyzed by @xcite within the self - energy functional approach@xcite .
they particularly focused on @xmath26 and discussed an unexpected occurrence of phase separation in the paramagnetic state between a mott - hubbard insulator and a correlated metal at and near half - filling . here
we want to investigate the behavior of the system as function of increasing frustration . due to symmetry
there is no difference between @xmath1 and @xmath27 .
the calculations were done using wilson s nrg as impurity solver for the dmft , with @xmath28 , @xmath29 states kept per nrg step and a logarithmic broadening @xmath30 to obtain spectral functions .
we want to note at this point that the choice of nrg numerical parameters does not influence the qualitative nature of the results .
we observe , however , small dependencies on @xmath31 and @xmath32 , which tend to become more pronounced close to phase transitions and may result in systematic errors in numerical values for critical parameters of the order of @xmath33.@xcite figure [ pmit ] shows the paramagnetic metal insulator transition for various values @xmath34 . as energy - scale
we choose the bandwidth @xmath35 \mbox{$4\vert t_2\vert + 2t_1+t_1 ^ 2/(4\vert t_2\vert)$ for $ \vert t_2\vert > t_1/4 $ } \end{array}\right.\ ] ] of the non - interacting system .
note that these results are obtained by artificially suppressing an antiferromagnetic instability .
we will come back to this point later .
the occupation was kept fixed at @xmath36 by adjusting the chemical potential .
note that in contrast to the case with @xmath22 it is not possible to achieve @xmath37 here within numerical precision due to the asymmetric dos ( see fig .
[ frustdos ] ) . for increasing @xmath2 the pmit
is shifted towards lower interaction strengths and also lower temperatures .
while the shift in the interaction strength is rather moderate , we notice a large difference in the temperature of the critical endpoint between the unfrustrated and highly frustrated system .
this observation of course renews our interest in the question , to what extent long - range hopping can help to push the paramagnetic mit out of the expected antiferromagnetic phase for reasonable magnitudes of @xmath1 to create a phase diagram similar to the one found for v@xmath3o@xmath4 .
the scenario proposed by @xcite relied on the fact that the paramagnetic phase largely remains unaltered with increasing @xmath1 .
as the nel - temperature for the antiferromagnet is reduced at the same time , the pmit can eventually outgrow the antiferromagnetic phase .
we now allow for antiferromagnetic ordering in our calculations . to this end
we reformulate the dmft for an ab lattice structure@xcite to accommodate the nel ordering and initialize the calculation with a small staggered field , which is turned off after one dmft iteration .
the system then either evolves into a paramagnetic or antiferromagnetic state with increasing number of dmft iterations .
figure [ phase1 ] shows the resulting phase diagrams for @xmath38 ( upper panel ) and @xmath39 ( lower panel ) for different temperatures and interaction strengths .
phase diagram for @xmath38 @xmath40 .
the shaded area represents the antiferromagnetic phase , while the white area represents the paramagnetic phase .
the lines show , where the pmit in the paramagnetic phase would occur .
the points denote the parameter values , where dmft calculations were done . from these points
the shaded area was constructed as guide to the eye .
additional calculations were performed to find the pmit - lines.,title="fig:",scaledwidth=45.0% ] phase diagram for @xmath38 @xmath40 .
the shaded area represents the antiferromagnetic phase , while the white area represents the paramagnetic phase .
the lines show , where the pmit in the paramagnetic phase would occur .
the points denote the parameter values , where dmft calculations were done . from these points
the shaded area was constructed as guide to the eye .
additional calculations were performed to find the pmit - lines.,title="fig:",scaledwidth=45.0% ] the small black points show the locations , where calculations have actually been performed . from these data
the shaded areas were constructed representing the antiferromagnetic phases .
this of course means that the phase boundaries shown here must be considered as guess only .
however , as we do not expect any strange structures to appear , this guess will presumably represent the true phase boundary within a few percent .
the full lines in fig .
[ phase1 ] are the pmit transitions .
note that for both diagrams the same division of axes was chosen .
in contrast to the hubbard model on a bipartite lattice with @xmath22 , there now exists a finite critical value @xmath41 , below which no antiferromagnetism can be stabilized even for temperature @xmath42 . with increasing frustration the paramagnetic - antiferromagnetic transition is shifted towards higher interaction strengths and lower temperatures , while the pmit is shifted towards lower interactions strengths .
so obviously the pmit is shifted towards the phase boundaries of the antiferromagnetic dome .
so far this is the expected effect of the nnn - hopping which introduces frustration to the antiferromagnetic exchange .
however , note that although @xmath39 represents already a very strongly frustrated system , the pmit still lies well covered within the antiferromagnetic phase .
let us now have a closer look at the paramagnetic - antiferromagnetic transition . here
, @xcite made the prediction that one has to expect a first order transition close to the critical @xmath41 at low temperatures ; while at larger values of @xmath18 again a second order transition was found .
for two different temperatures and @xmath39 . in the upper panel
there are for each temperature two transition lines , representing either increasing or decreasing interaction strength .
the region between both lines embodies a hysteresis region .
the lower panel shows the transition for large interaction strengths
. here no hysteresis region could be found , but a smooth transition.,scaledwidth=45.0% ] figure [ hysteresis2 ] shows the staggered magnetization for different temperatures and interaction strengths at fixed @xmath39 .
the upper panel collects data for the transition at low temperatures at the lower edge of the antiferromagnetic phase .
the full lines represent the transition from the paramagnetic to the antiferromagnetic state with increasing interaction strength for two different temperatures , while the dashed lines represent the transitions from the antiferromagnetic to the paramagnetic state with decreasing interaction strength . in the upper panel ( small @xmath18 )
one can clearly see a hysteresis of the antiferromagnetic transition .
this hysteresis as well as the jump in the magnetization are clear signs for a first order transition .
this antiferromagnetic hysteresis is very pronounced for strong frustration but numerically not resolvable for example for @xmath43 .
we believe that the hysteresis region shrinks with decreasing @xmath1 and eventually can not be resolved anymore with numerical techniques . the whole temperature depending hysteresis region can be seen in fig .
[ hysteresis ] for the case @xmath39 .
one can see clearly the shrinking of the hysteresis region with increasing temperature .
note that such a hysteresis is also found in the two - sublattice fully frustrated model,@xcite which means that this quite likely is a generic effect in frustrated systems at intermediate coupling strengths . the lower panel in fig .
[ hysteresis2 ] shows the staggered magnetization for temperatures just below the corresponding nel - temperatures and at large interaction strengths . here
the magnetization vanishes smoothly , which is the behavior expected for a second order phase transition . in summary
we thus find a first order transition at the critical interaction @xmath41 where antiferromagnetism sets in , and a second order transition for the large coulomb parameter @xmath44 .
including the temperature depending hysteresis region ( between dashed and full line ) .
the meaning of symbols is as in fig .
[ phase1 ] .
the shaded area is as before meant as guide to the eye .
, scaledwidth=45.0% ] the merging from both transition lines is an interesting point in itself .
there must be a critical point where the first order transition changes into a second order transition .
it is however not possible to resolve this merging within dmft / nrg .
first , the logarithmic discretization of the temperatures within the nrg does not allow to resolve this merging - region with arbitrary precision .
second , the magnetization of the system becomes very small in this region , so it is not possible to distinguish between a ( tiny ) jump and numerical artifacts of a smoothly vanishing order parameter .
consequently , we can not decide anymore of what order the transition will be . antiferromagnetic metallic phases at half - filling were reported in earlier publications . in our calculations we saw no evidence for an antiferromagnetic metallic state at half - filling . especially for strong frustration @xmath45
the system directly jumps from a paramagnetic metallic solution into an antiferromagnetic insulating solution with high magnetization . in the papers cited , the region showing an antiferromagnetic metallic solution broadens with increasing @xmath1 .
this prediction we clearly can not confirm , as discussed above . only in systems with small to intermediate frustration there are narrow interaction regimes where we observe a small finite weight at the fermi level .
one must however consider that the occupation number is not exactly one but only within @xmath46 .
also it was sometimes difficult to stabilize a dmft solution in these regions . in summary , we can not see any clear signs for an antiferromagnetic metallic state at half - filling in our calculations . if any exists , then only for rather low frustration in a very small regime about the critical interaction . to what extent
these rather special conditions can then be considered as realistic for real materials is yet another question .
in this last paragraph we want to study the situation , in which @xmath0 and @xmath1 are comparable in strength .
interestingly , there has been no attempt to calculate the phase diagram on a mean - field level in the strongly frustrated model @xmath47 .
therefore , before discussing the results of the dmft calculations for strongly frustrated systems @xmath48 let us try to gain some insight into the physics we must expect , by inspecting classical spins on a bethe lattice with nn - interaction @xmath49 and nnn - interaction @xmath50 . allowing that nearest - neighbor spins enclose an angle @xmath51
one ends up with the energy functional @xmath52 performing the same limits and scaling as in dmft one finds ( see appendix a ) @xmath53 thus , the nel - state with @xmath54 is the stable ground state for @xmath55 , while one finds a spin wave with @xmath56 for @xmath57 . for the dmft calculations we can allow only for solutions commensurate with the lattice
this however will possibly be inconsistent with the spin structure favored by the system .
if , for example , we perform a calculation focusing on the ferromagnetic solution within a parameter regime , where the system wants to order antiferromagnetically , dmft will not converge . to investigate spin wave states with periodicities with more than two lattice sites
, one has to set up the correct dmft self - consistency equations respecting the lattice structure . while for a system on an infinite bethe lattice with nn hopping
only it is straightforward to extend the dmft to commensurate magnetic structures with periodicities of more than two lattice sites , we did not succeed in devising a scheme that allows for such calculations for systems with nnn hopping .
the reason is that one has to partition the lattice into an @xmath58 structure .
however , the nnn hopping makes it impossible to uniquely identify the connectivity of the respective sublattices .
a method proposed by @xcite for the two - dimensional cubic lattice is not applicable in our case .
we thus only allowed for paramagnetic , ferromagnetic and antiferromagnetic solutions in our calculations .
the resulting phase diagrams for @xmath59 are shown in figs .
[ hight2_u ] and [ hight2_t ] .
figure [ hight2_u ] displays the ground states for different grades of frustration and interaction strengths . for @xmath60
the phase diagram has the same structure as for small and intermediate @xmath1 .
the critical interaction strength @xmath41 necessary to stabilize the nel - state increases and for all values above @xmath41 we find an antiferromagnetic phase with a hysteresis region at the phase boundary . ) phase diagram for different strengths of frustration as function of the interaction .
the brindled regions are hysteresis regions for increasing or decreasing interaction .
the phase boundaries of the incommensurate ( ic ) phase are only qualitative ( explanation see text).,scaledwidth=45.0% ] for @xmath61 including the pmit .
the brindled area represents again the hysteresis region.,scaledwidth=45.0% ] for @xmath62 the critical value @xmath41 one needs to stabilize the nel - state increases dramatically . for @xmath63
finally we do not find an antiferromagnetic nel - state at all for any interaction strength @xmath18 .
our dmft calculations however indicate that in this range of @xmath34 there actually does exists another magnetic phase .
namely , for sufficiently small temperatures one obtains a finite spin polarization in every dmft iteration .
however , the dmft does not converge to a unique state as function of dmft iterations ( see also fig .
[ noncon ] ) . ,
@xmath64 and @xmath65 .
the lines are meant as guide to the eye.,scaledwidth=45.0% ] in the phase diagrams in figs .
[ hight2_u ] and [ hight2_t ] we have named this regime the incommensurate phase ( ic phase ) . in this parameter regime the nel - state becomes unstable towards the behaviour shown in fig .
[ noncon ] . here
one can switch between a conventional nel - state and the ic phase by only a small change of the interaction strength .
note , that the phase boundaries shown in the figure must be taken with some care as we can not compare the energies of the nel - state and this ic phase to properly determine the phase boundaries .
as we observe precisely the same behavior for all investigated values @xmath66 we are convinced that the ground state in this region is an incommensurate state , as to be expected from our results for @xmath67 .
similar observations also hold for finite temperatures as shown in fig .
[ hight2_t ] , where the @xmath68 phase diagram for fixed @xmath61 is displayed . for increasing interactions and @xmath69 there
first is a transition from a paramagnetic metal to the ic phase and for @xmath70 from the ic phase to the nel - state .
for increasing temperature the ic phase eventually becomes unstable towards the nel - state . in fig .
[ hight2_t ] one can also see the pmit lines . as one
can see it lies within the hysteresis region of the magnetic phases but clearly outgrows both magnetic phases .
this is the scenario described in @xcite .
we studied the dmft phase diagram of the hubbard model at half - filling in the presence of nn- and nnn - hopping .
in contrast to previous investigations we did our calculations for a bethe lattice with proper nnn hopping @xmath1 , introducing a highly asymmetric dos already for the non - interacting system .
the first important observation concerns the paramagnetic metal - insulator transition , which is suppressed by increasing @xmath1 , but at the same time shifted to lower values of the coulomb interaction .
the at @xmath22 ubiquitous antiferromagnetic phase on the other hand is suppressed up to a critical value @xmath71 with increasing @xmath1 , as expected .
furthermore , a hysteresis region between the paramagnetic metal at small u and the antiferromagnetic insulator at large u develops , showing that the transition is of first order . note that we did not observe any evidence for an antiferromagnetic metal close to the phase boundary , nor did the pmit reach out of the antiferromagnetic insulator up to values @xmath72 .
thus far the observations are similar to the results found by @xcite for the two - sublattice fully frustrated bethe lattice@xcite .
the shift of the pmit to lower values of @xmath18 together with a moderate suppression of the critical temperature for larger @xmath1 however motivated a more detailed investigation of the region of larger @xmath1 . a simple argument based on classical spins with competing interactions showed that one has to expect an additional incommensurate phase here .
in fact , as already anticipated qualitatively by @xcite , for frustrations @xmath73 we eventually found that the pmit lies within the hysteresis region of the antiferromagnetic phase for @xmath69 , but outgrows it in temperature . for such strong frustration
we also found evidence for another magnetic phase besides ferromagnetism or antiferromagnetism .
unfortunately this phase could not be stabilized within our dmft calculations , so its real nature remains open . in connection with our argument based on classical spins
we believe that we can interpret the observed structure as an incommensurate phase .
this conjecture is further supported by the fact , that for @xmath74 we found no antiferromagnetic solution of the nel - type , but only this frustrated magnetic phase .
especially the latter findings make it highly desirable to set up a scheme that allows to study commensurable structures with period beyond nel - type for arbitrary lattice structures including longer - ranged hopping .
we want to thank martin eckstein and dr .
markus kollar for many helpful discussions on the bethe lattice with nnn hopping , and dr .
timo aspelmeier for his help with the vector - spins .
this work was supported by the dfg through pr298/10 .
computer support was provided by the gesellschaft fr wissenschaftliche datenverarbeitung in gttingen and the norddeutsche verbund fr hoch- und hchstleistungsrechnen .
here we want to present the calculation for 3-dimensional vector spins on a bethe lattice with antiferromagnetic coupling between nn- and nnn - lattice sites . with sites numbered according to
the vector spins on the right .
the nearest neighbors of site 1 must lie on a circle so there is an angle @xmath51 between @xmath75 and @xmath76 .
similarly the nearest neighbors of one of the @xmath76-spins must lie on a circle including the @xmath77-spins and the @xmath75-spin.,scaledwidth=45.0% ] we take @xmath78 nearest neighbors and interaction strengths @xmath49 between nn - sites and @xmath50 between nnn - sites .
the last parameter entering this calculation is the angle @xmath51 between nn - spins .
although the initial assumption that two neighboring spins form an angle theta may seem somewhat restrictive , we are not aware of other configurations with lower total energy @xcite .
we want to minimize the energy with respect to this angle . according to fig .
[ vectorspins ] the nn - spins of one spin , must lie on a circle .
the spins ending on the circle are all nnn - spins . due to the antiferromagnetic interaction @xmath50
, we assume that they want to maximize the angle between them .
since there are @xmath78 spins on each circle , we assume they will have angle @xmath79 projected on the circle . using now simple trigonometry , the angle between nnn - spins is given by @xmath80 where @xmath14 runs from @xmath81 to @xmath82 , giving the different positions on one circle .
inserting this into the hamiltonian @xmath83 one finds for the energy @xmath84 performing now the limit @xmath85 and scaling @xmath86 and @xmath87 one finally obtains for the energy per lattice site @xmath88 one can now see that the nel - state @xmath54 is stable for @xmath89 , because @xmath90 . | we study the interplay between nel - antiferromagnetism and the paramagnetic metal - insulator - transition ( pmit ) on a bethe lattice with nearest and next - nearest neighbor hopping @xmath0 and @xmath1 .
we concentrate in this paper on the situation at half - filling . for @xmath2
the pmit outgrows the antiferromagnetic phase and shows a scenario similar to v@xmath3o@xmath4 . in this parameter regime
we also observe a novel magnetic phase . |
gamma ray bursts ( grbs ) are the most energetic explosions that take place in our universe , with a typical energy of @xmath6 released on a timescale of between a millisecond and a few thousand seconds .
the release of such a considerable amount of energy over such a short period requires an outflow that is relativistic @xcite and is likely to be anisotropic @xcite .
the energy within the outflow is released primarily though shocks .
the grb is thought to be produced through internal shocks from interactions between successive shells of ejecta @xcite .
subsequently an afterglow is emitted as the outflow is decelerated through collisionless shocks with the external medium @xcite .
a forward shock propagates into the external medium and emits from x - ray to radio wavelengths , and a reverse shock travels back through the ejecta and peaks at longer wavelengths than the forward shock @xcite .
the afterglow can reveal many properties of the progenitor and its surroundings .
currently , the most up to date model @xcite allows the use of the temporal and spectral indices to indicate the nature of the surrounding medium , whether it is a uniform density medium , or a medium with a density that is decreasing radially as expected for a stellar wind e.g @xmath7 , where @xmath8 is the particle density and @xmath9 is the radius from the grb .
the indices also indicate the location of the observing band relative to the synchrotron self absorption frequency @xmath10 , the peak frequency @xmath11 and cooling frequency @xmath12 .
some grb afterglows have a period of slow decline in their lightcurves ( @xmath13 , where @xmath14 ) @xcite , which is generally accepted to be due to continued energy injection @xcite .
this energy could be due to a central engine that is long lasting @xcite , later shells catching up and colliding with slower shells that were emitted earlier @xcite , or the slow release of energy stored in the form of poynting flux @xcite .
the end of energy injection is signalled by an increase in decay rates of the afterglow lightcurves to @xmath13 , where @xmath15 .
currently the most effective observatory for studying grbs and their early afterglows is _ swift _ , which has now been in operation for over two years .
it has the ability to observe emission ranging from @xmath16-rays to optical with the three on - board telescopes , namely the burst alert telescope ( bat ; @xcite 2005 ) , the x - ray telescope ( xrt ; @xcite 2005 ) and the ultra violet and optical telescope ( uvot ; @xcite 2005 ) .
the bat covers a large area of the sky allowing of order 100 bursts to be detected per year .
when the bat has been triggered by the grb , the satellite slews automatically allowing the xrt and the uvot to begin observing as soon as possible , usually within @xmath17 of the burst trigger .
_ swift _ is thus able to observe both the initial @xmath16-ray explosion and the following early afterglow . this paper looks at grb 050802 , a burst that appears to challenge the standard picture . at the end of the shallow decay
the x - ray afterglow breaks to a steeper decay , while the optical afterglow continues to decay as a power - law without a break .
if both the x - ray and optical arise from the same component and if the x - ray break were due to the end of energy injection then the optical lightcurve would be expected to break at the same time . such behaviour has been noted recently in 6 _ swift _ grb afterglows @xcite of which grb 050802 is currently the best example .
this burst has a x - ray lightcurve with two distinct breaks and it was observed for 1.2@xmath18s .
it also had an optically bright afterglow ( @xmath1916 magnitude at early times ) that was well sampled in 6 filters of the uvot up to 1 @xmath20 s and thereafter observed with the white uvot filter . in this paper
we discuss the possible models that could explain this behaviour .
we will use the convention flux @xmath21 with @xmath22 and @xmath23 being the temporal and spectral indices respectively .
we assume the hubble parameter @xmath24@xmath25 and density parameters @xmath26=0.7 and @xmath27=0.3 .
uncertainties are quoted at 1@xmath28 unless otherwise stated .
the bat was triggered by grb 050802 at 10:08:02 ut on the @xmath29 august 2005 @xcite .
the lightcurve rises for 5s to the first of three peaks and has a @xmath30s ( 90 % confidence level ) .
the fluence in the @xmath31 band is @xmath32 to @xmath33 confidence level @xcite .
observations with the xrt and uvot began 289s and 286s respectively , after the bat trigger@xcite .
both the xrt and uvot continued to observe until @xmath34 after the burst trigger .
the xrt began observations by locating the burst with image mode ( i m ) .
after the burst was located , data were taken in windowed timing ( wt ) mode for 163s .
a fading uncatalogued source was found within @xmath35 of the bat position @xcite and was confirmed as the x - ray counterpart of grb 050802 .
480s after the burst trigger , the xrt changed modes and continued observations in photon counting ( pc ) mode .
uvot observations showed a fading , uncatalogued source at ra=@xmath36@xmath37@xmath38 , dec=@xmath39@xmath40@xmath41 @xcite . following the trigger ,
a series of automated exposures were taken in the three optical and three ultra - violet filters .
a log of the observations is given in table [ tab : uvotpoints ] .
the observations consisted of an initial 100s ` finding chart ' exposure in the v - band , 10s exposures in each passband for 7 rotations of the filter wheel , followed by a sequence of @xmath42s and @xmath43s exposures .
later observations ( after @xmath44s ) , were taken in the uvot white filter .
the afterglow was also imaged and detected with the 2.6 m shain telescope , 8hrs after the burst , with r and i band magnitudes of 20.6 and 20.2 respectively @xcite .
spectroscopic observations were carried out with alfosc on the nordic optical telescope .
several absorption features were detected providing a redshift measurement of @xmath45 ( @xcite 2005a @xmath46 2005b ) .
@xcite discovered an extended source within 1 of the uvot afterglow location , which they propose could be the host galaxy .
the xrt data were reduced using the xrt pipeline software v2.0 .
source and background counts were taken from the cleaned event files using extraction regions , in order to construct spectra and lightcurves .
events of grade 0 - 2 were used for the wt mode and 0 - 12 were used for the pc mode . for the wt mode data we used a 40 pixel strip for the source extraction region and a 40 pixel strip for the background extraction region . in the pc mode
data the first 2.5ks of data were found to be piled - up and so required the use of an annular extraction region .
the size of the region that was affected by pile up was determined by comparing the radial profile of the afterglow with a model of the xrt point spread function ( psf ) .
the radial profile and the model psf are inconsistent at radii less than 2.5 pixels ( @xmath47 ) , so we used an inner radius of 2.5 pixels and an outer radius of 30 pixels ( @xmath48 ) .
the size of the source extraction region for the rest of the pc data was 30 pixel ( @xmath48 ) radius . for all pc data
, background counts were extracted from a circular region of radius 80 pixels ( @xmath49 ) .
appropriate response matrices ( rms ) were taken from the swift calibration database , caldb 20060424 and effective area files were constructed using the standard xrt software . a correction factor was calculated and applied to the piled - up section of the xrt lightcurve to account for the excluded , piled up pixels .
fortunately , the source was not located near the bad pixel columns and so no correction was required .
the uvot event files were screened for bad times ( e.g. south atlantic anomaly passage , earth - limb avoidance ) and the images were corrected for mod-8 noise using the standard uvot software .
the images were transformed to sky coordinates and then corrected for the @xmath50 uncertainty in the aspect of the spacecraft pointing using bespoke software .
counts for the afterglow were extracted using an aperture of radius @xmath51 for the optical filters and @xmath52 for the ultra - violet filters .
background subtraction was performed using counts extracted from a larger region offset from the source position .
the measured count rates were aperture - corrected to radii of @xmath47 for the v , b and u filters and 12@xmath53 for the uv filters .
these were then translated to magnitudes using the standard uvot zero - points ( see table [ tab : uvotpoints ] ) .
the optical lightcurve is shown in fig .
[ fig : lateopticallc ] .
the lightcurve from each filter of the uvot was normalized to the v filter .
the normalization factor for each filter ( see table [ tab : uvotnorm ] ) was calculated by taking the average count rate through the filter in the 400s-1000s time range , and dividing this value by the average count rate in the v filter over the same time interval .
later observations were obtained with the uvot white filter ; these were normalized to the equivalent v count rates as follows .
the optical - uv spectral energy distribution ( sed ) was modelled using the average count rates from the v , b , u , uvw1 , uvm2 and uvw2 filters in the 400s-1000s time range . the
optical / uv response matrices were then used to predict the ratio of the v to white count rates .
seds were produced spanning the optical to x - ray range for early ( 400s-1000s ) and late ( 35ks-55ks ) times .
for each period , the average count rates of the exposures in each uvot filter were used to produce the optical spectral values . for the x - ray part of each sed a spectrum
was extracted in the relevant time range . [ cols="<,^,^,^,^,<,<,^,^,^,^",options="header " , ]
in summary , we have determined that the x - ray lightcurve can be divided into 3 segments with two breaks at 420s and 5000s . throughout the afterglow
, there is no evidence for x - ray spectral evolution .
we also find that the optical lightcurve decays as a single power - law , which has a slightly steeper decay than the second segment of the x - ray lightcurve .
we have determined the earliest possible time that the optical lightcurve could have changed its decay rate to match that of the third x - ray segment to be 19ks , which is 14ks after the last x - ray break . in the following subsections
, we will look in detail at the x - ray afterglow to examine the origin of the x - ray breaks and to determine the best fitting closure relations between the temporal and spectral slopes .
then we shall discuss the question of why the x - ray lightcurve breaks at 5000s while the optical lightcurve continues to decay as a power - law .
several mechanisms will then be investigated to find a compatible model for the production of the x - ray and optical afterglow .
we begin by looking at the first break to see what we may learn .
the break occurs at @xmath54s , which is preceded by a rise from 286s with a slope of -0.80@xmath55 .
the break is then followed by a gradual decay with a slope of @xmath56 .
the rise and the successive slow decay can be regarded as a broad peak in the lightcurve and it may be attributed to ` flaring ' activity @xcite or could be the early phase of the jet interacting with the external medium , giving rise to the forward shock @xcite . however , it is not possible to discriminate between the two mechanisms .
flares often have different spectra to the afterglow , but if this is a flare it may be so small we would not expect to see any spectral variability . on the other hand , we can not tell if this is the early rise of the afterglow as we have too little data prior to the peak . an additional contribution to the second segment from a large flare
may also be excluded as a power - law fit to the x - ray lightcurve between 1000s and 3000s results in a slope of @xmath57 with @xmath58/d.o.f=21/38 .
typically , x - ray afterglows are analysed by applying the closure relations given by @xcite to the x - ray spectral and temporal indices .
these relations are an aid in determining the location of the observed x - ray band relative to the synchrotron frequencies ( @xmath59 ) and the environment in which the the burst occurs .
we skip the first segment of the x - ray lightcurve as the closure relations are not applicable to a lightcurve that is rising .
we begin by applying the simplest non - injected relations to the second segment , in which @xmath60 and @xmath61 , the only closure relation that agrees without the requirement of energy injection at 3@xmath28 confidence level , is @xmath62 this relation applies to a wind or ism medium with @xmath63 , but , this relation is not appropriate for grb 050802 as the value of the energy index @xmath64 , determined through @xmath23=@xmath64/2 is @[email protected] .
values of @xmath64@xmath662 require a different set of closure relations provided by @xcite , of which none satisfies the second decay of grb 050802 .
more complex closure relations , which include a third parameter to account for energy injection , are provided by @xcite . in these models
the luminosity is assumed to evolve as l(t)=@xmath67/@xmath68 where @xmath69 is the luminosity index and @xmath70 is the time for the formation of a self - similar solution , which is approximately the deceleration time .
the luminosity index @xmath69 must be @xmath661 for the energy injection to affect the afterglow dynamics .
we found that there are two injected closure relations that are consistent with @xmath2 and @xmath71 .
the first relation is @xmath72 this closure relation is for a wind medium with slow cooling electrons with either @xmath73 or @xmath74 and where @xmath69=@xmath75 .
the second possible energy injected relation is @xmath76 this relation is for a uniform medium with slow cooling electrons with either @xmath77 or @xmath78 and where @xmath69=@xmath79 . for the closure relations given in equations [ equ:3 ] and [ equ:4 ] , @xmath80 and so @xmath81 . when the closure relations are applied with the indices of the third segment , where @xmath82 and @xmath83 , we find that the most consistent closure relation without energy injection is @xmath84 which is consistent at 3.5@xmath28 confidence .
all other non - injected closure relations are ruled out at a minimum of 4@xmath28 .
equation [ equ:1 ] is for electrons that are slow cooling within the range @xmath11@xmath66@xmath85@xmath66@xmath12 in a wind medium . again
a value for the energy index may be determined though @xmath80 with @[email protected] .
generally , the energy index is expected to remain constant throughout the x - ray afterglow , unless it is changed by energy injection @xcite .
therefore , the application of the closure relations to the x - ray temporal and spectral indices suggests that the second segment is energy injected and the third segment is not , and therefore that the break between the second and third segments at @xmath865000s is due to the discontinuation of energy injection . at this point
it is worth pointing out that the second and third segments of grb 050802 are consistent with the second and third segments of the canonical lightcurve presented by @xcite .
furthermore , from the application of the closure relations above we would come to the same physical explanation as @xcite and @xcite for this temporal behaviour : that the break between the two segments corresponds to the end of energy injection .
having looked at the x - ray afterglow in isolation , we will now investigate whether this physical picture is compatible with the observed optical emission .
the standard afterglow is based on the assumption of a single relativistic outflow @xcite and we will begin by considering models of this type before considering multi - component outflows .
we start by considering an afterglow in which the emission in both the x - ray and optical bands is produced predominantly by the forward shock .
if we compare the decay of the optical lightcurve with the segments of the x - ray lightcurve we find that the optical decay is similar , although not identical to , the slow ` energy injected ' decay of the second segment .
this would suggest that the optical lightcurve is also characteristic of an energy - injected afterglow .
since the x - ray lightcurve breaks at 5000s , suggesting the end of energy injection , and the optical lightcurve continues with a shallow decay indicative of energy injection , we first consider the possibility that the electrons responsible for producing the optical emission continue to receive energy while the x - ray emitting electrons do not .
there are two possible mechanisms for continued energy injection .
these are the arrival of additional shells of material at the shock region or the release of energy stored in the form of poynting flux . in the first case , the injected energy changes the fireball dynamics , but can only change the balance of x - ray to optical emission and produce a break in one of the lightcurves if @xmath11 or @xmath12 pass through one of these bands .
however , we can rule out the passage of one of these breaks through the x - ray or optical afterglow because there is no break in the optical lightcurve and there is no change in the x - ray spectrum .
this therefore rules out continued energy in the form of blast waves and so we consider continued energy injection in the form of poynting flux .
energy is supplied in the form of poynting flux when there is a rotating compact object that has a magnetic field , @xmath87 .
the flux is converted to kinetic energy when kink instabilities cause the magnetic field lines to reconnect in the outflow @xcite .
the energy density of poynting flux is @xmath88 , and within the reconnection region the energy available per particle for particle acceleration is @xmath89 @xcite . to decrease the photon energy
so that further emission is below the x - ray band , requires a decrease of the average particle energy .
the xrt is sensitive over more than a decade in photon energy , therefore a decrease in the average photon energy from above the xrt energy band to below the xrt energy band requires a factor @xmath9010 change in @xmath87 and therefore a factor of @xmath91 change in the energy density of poynting flux @xcite .
this implies a large change in the overall rate of energy injection , which is inconsistent with the continued power - law decay in the optical .
a reverse shock is an alternative means for producing optical emission without x - ray emission and so we now examine the possibility that this could sustain the optical emission of grb 050802 after the x - ray lightcurve has broken .
the reverse shock will travel back through the ejected matter and will emit as long as it is passing through ejecta .
the reverse shock emission will cease when it passes over the last ejected shell @xcite .
the shock crossing time is expected to be approximately the duration of the grb , but if energy continues to be injected into the afterglow by the arrival of further shells the reverse shock may continue for as long as the energy injection phase . if the end of energy injection occurs at 5000s , as suggested by the x - ray lightcurve , then we should expect that the reverse shock must cease at approximately 5000s .
therefore , this is not the required mechanism as the optical lightcurve has no break until at least 19ks .
@xcite have proposed that it is possible to sustain a power - law decay in the optical , while the x - ray lightcurve shows a break due to the end of energy injection , if the microphysical parameters are changing with time .
the microphysical parameters that must change to keep the optical temporal decay as a power - law are : the fraction of post - shock energy in the magnetic fields @xmath92 , the fraction of post - shock energy that is given to electrons @xmath93 and the blast - wave kinetic energy e @xcite .
these parameters are expected to evolve with lorentz factor @xmath94 , if they evolve at all . in the scenario proposed by @xcite
, the cooling frequency @xmath12 lies between the optical and x - ray bands . in this case
, there should be a slope change of @xmath95 between the x - ray and optical bands and @xmath12 moves to lower frequencies with time @xcite .
for grb 050802 , through the fitting of a broken power - law with a fixed @xmath95 to the 400s-1000s sed ( see section 4.2 ) , the maximum energy , at which a break could be present , was determined to be 0.02kev .
we examined the optical to x - ray seds to see if the x - ray and optical lightcurves can be explained by @xmath12 moving to lower frequencies with time as expected in the model of @xcite . in this model the largest change in the ratio @xmath96 , where @xmath97 is the x - ray count rate and @xmath98 is the v band count rate corresponds to the motion of @xmath12 from its highest allowed value to below the optical pass band , at which point the optical emission lies on the same power - law spectral segment as the x - ray emission . in the broken power - law fit to the 400s-1000s sed , the highest allowed value of @xmath12 at 3@xmath28 confidence is 0.06kev .
therefore we refited the 400s-1000s sed with @xmath12 fixed at 0.06kev , then reduced the value of @xmath12 to below the uvot spectral range and so determined that the smallest @xmath96 value allowed by this model is @xmath99 .
the observed range of @xmath96 extends to much lower values than this ( fig .
[ fig : opticalxrayratio ] ) .
we find that the weighted mean of the data points more than @xmath100ks after the bat trigger is @xmath101 , which is @xmath102 below the lowest allowed value of @xmath99 , thus allowing the model of @xcite to be ruled out .
we now look in to the optical to x - ray sed further and ask : to what extent can single component models be ruled out completely ? as discussed in section 4.2 , a power - law fit to the late time ( 35ks-55ks ) sed provides a poor fit and the best fit value of @xmath23 is softer than the x - ray spectral index for segment 3 , which includes the 35ks-55ks time interval . in this respect ,
we note that there is no evidence for any evolution in the x - ray spectral index at any point in the afterglow and the weighted mean of the indices found in the 3 segments of the x - ray lightcurve gives @[email protected] .
to see why the fit to the late time sed is poor we show the x - ray spectrum extrapolated to lower energies using this spectral slope ( @xmath23=0.88 ) as a dashed line in both panels of fig [ fig : sedearly ] .
the dashed line lies above the optical and uv data points as expected for an afterglow with significant extinction . in section 4.2
, we determined that at @xmath103 confidence , the host galaxy extinction is greater than 0.8 and 1.1 in the b and u bands respectively , no matter which models we choose for the extinction and continuum .
in addition , the galactic extinction is 0.1 for both the b and the u bands .
after correction for extinction , the b and u band fluxes in the late time sed lie significantly above the dashed line .
as described in section 4.2 , this power - law can be ruled out with @xmath9099 per cent confidence , implying at 99 per cent confidence that the region responsible for the x - ray emission can not produce all of the optical emission at late times , no matter where @xmath12 and @xmath11 lie . since single component outflows are unable to represent the late time sed satisfactorily , we now consider multi - component outflows to see which of these models are able to reproduce the observed lightcurves and seds .
a multi - component outflow is one consisting of two or more components that have different bulk lorentz factors .
the simplest model is one of two components : a narrow jet and a second wider , but slower jet that surrounds the narrow component .
this geometry may be generated by the narrow jet giving rise to a wider and slower component @xcite , where the bulk @xmath94 decreases over time and produces a jet with angular structure .
alternatively , the two components may be formed at the same time for example when neutrons and protons decouple in a neutron - rich , hydroydynamically accelerated jet from a neutron star or from the neutron - rich accretion disk of a collapsed massive star @xcite .
@xcite show that within such a neutron - rich jet , the wide component has greater energy than the narrow component so that the wider component is able to dominate at late times . in both cases , the narrow component will produce the x - ray emission , whereas the optical emission will be produced by the wider component as it travels at lower lorentz factors than the narrow component .
this picture can be used to explain bursts such as grb 021004 and grb 030329 @xcite .
the wider jet is expected to reach the undisturbed medium at a slightly later time than the narrow jet .
thus , the wide component may be observed as a later start for the optical lightcurve when compared to the x - ray lightcurve , or on an established optical lightcurve , the arrival of the wide jet to the external medium may be shown as a late rise / bump or a change in the decay slope .
there are three options within a two - component jet we will investigate to see if one could have produced the observed lightcurves of grb 050802 .
all of these options involve energy injection , which is introduced through shells of material .
the options are : energy injection with the ejected shells wide enough to refresh both components of the jet with a lorentz factor that decreases over time ( so that after 5000s the shells no longer reach the shock in the narrow component ) , the cessation of energy injection at 5000s in one component only and , finally , the continuation of energy injection for the duration of the afterglow , but with the narrow component producing a jet break at 5000s .
the first option is that energy injection is distributed uniformly throughout both jets and continues for a sustained period of time .
initially , the shells have a wide enough angular distribution , and travel at a @xmath94 large enough , to reach the forward shock produced by both components .
both the x - ray and the optical afterglows will appear energy injected .
as time goes by , the lorentz factor of the freshly ejected shells decreases .
eventually , the ejected shells will not travel at a lorentz factor large enough to reach the shocked region in the narrow component , but will still be able to reach the wide component .
this will cause a break to be seen only in the x - ray afterglow .
however , the freshly ejected shells will still travel at @xmath94 large enough to reach the wide component and therefore the optical lightcurve will continue to appear energy injected .
this much appears to be consistent with the data because , as discussed in section 5.2 , the optical lightcurve appears to be energy injected for its entire observed duration and does not break with the x - ray lightcurve .
however , as the wide component is slower than the narrow component , the wide component will absorb shells ( i.e receive energy ) at a faster rate than the narrow component .
therefore , the optical emission is expected to decay less rapidly than the x - ray emission . for grb 050802 , the opposite is actually observed : in the optical the decay index @xmath104 , while the x - ray second segment has a decay index of @xmath60 , thus this model is excluded .
the second option calls for the x - ray break at 5000s to have been caused by the cessation of energy injection in the narrow component , while injection continues in the wide component .
however , this implies that the energy is injected as a hollow jet at late times and we consider this to be an unattractive solution on physical grounds , and lacking in testable predictions .
the third option is that energy is continuously injected into both components and the x - ray break at 5000s results from a jet break of the narrow component @xcite .
the spectral indices of the x - ray segments support this model as there is no spectral change over the break , as is expected for a jet break .
a jet break may be produced within a laterally expanding or a non - expanding jet .
the closure relations for these cases are provided by @xcite . in order for this scenario to be viable we have to find a post jet - break closure relation that has consistent energy injection , external medium , synchrotron frequencies and electron spectral index to that of the pre jet - break closure relation . the closure relation that satisfies this uses eq 35 of @xcite , and is : @xmath105 this relation is independent of the grb host medium and is for a laterally expanding jet with @xmath106 .
the relation provides a value of @xmath69=@xmath107 , which is consistent at the 99 per cent confidence level with the value of @xmath69 obtained through the closure relation in equation [ equ:4 ] .
the consistency with equation [ equ:4 ] implies that this burst originated in a uniform ism and not a wind medium as suggested in section 5.1 .
furthermore , we can use the second x - ray break time to infer the opening angle of the narrow beam and to put a lower limit on the angle of the wide component .
we calculated the angles using the following expression from @xcite : @xmath108 where @xmath109 is the isotropic energy of the burst , @xmath8 is the density of the medium and @xmath110 is the efficiency of converting energy in the ejecta into @xmath16-rays .
we use values of @xmath111=0.2@xmath112ergs ( in the range 15 - 350 kev ) and grb efficiency @xmath110=0.2 . approximating the circumburst medium as a spherical , uniform gas cloud of radius @xmath113 , the particle density @xmath8 ,
may be estimated simply as @xmath114 . the cloud radius for gas and dust to survive the grb ,
may be estimated using fig .
8 of @xcite .
this shows that the obscuring medium must extend to @xmath115 cm , implying a number density of @xmath11630 @xmath117 .
thus , the value for the narrow component is estimated to be @xmath118 and for the wide component the opening angle is @xmath119 .
we recognise that these angles are only approximate because they depend weakly on the bolometric energy , the efficiency of the burst and on the particle density of the external medium .
this model requires that energy be continually injected into both components of the jet for the observed duration of the burst .
the time scale over which this energy injection takes place is more than an order of magnitude larger than @xmath120s , which is the duration of energy injection required to explain the canonical x - ray afterglow @xcite .
this situation may seem unusual , but grb 050802 is not the first burst that has required discrete or continuous energy injection for an extended period .
a few examples of bursts that have required long duration energy injection are grb 021004 , xrf 050406 and grb 060729 @xcite .
the continuation of energy injection well beyond the jet break time will produce shallower post jet break slopes than the non - injected case and these may be mistaken for the discontinuation of energy injection .
this would lead to the surprisingly small number of reported jet breaks in the _ swift _ era @xcite .
however , the mechanism that could be responsible for long duration ( @xmath121s ) energy injection is uncertain .
currently , the most favoured scenario is one in which the duration of the central engine is short , but where the ejecta are emitted with shells of varying velocities @xcite .
the shells will order themselves before reaching the external medium and will form a continuous flow of shells . in this case , small time scale fluctuations need not be observed in the power - law lightcurves and the central engine is not required to be particularly long lived .
this model would allow the break to occur only in the x - ray lightcurve because the narrow and wide components travel at different @xmath94 , and the narrow component is beamed into a smaller angle .
therefore , the two component outflow with continued energy injection and a jet break in the narrow component is our favoured hypothesis for producing the observed lightcurve of grb 050802 .
the result from this paper has wider implications for our understanding of grbs .
segments two and three of the x - ray lightcurve of grb 050802 have the same characteristics as the segments two and three of the canonical x - ray afterglow lightcurve @xcite .
if we had looked at the x - ray data alone , then our explanation for the second and third x - ray segments of grb 050802 would have been equivalent to the explanation proposed by @xcite and @xcite for the canonical lightcurve : an energy injected decay followed by a ` normal ' decay . however , when the optical observations are taken into account , this explanation is no longer attractive . instead , a more plausible explanation is a two component outflow with the break in the x - ray lightcurve likely to be due to a jet break .
our results suggest that without the optical lightcurve the interpretation of grb afterglows may not be correct .
this paper has investigated grb 050802 , an unusual burst because the optical decays as a power - law , while the x - ray breaks twice .
we have analysed optical and x - ray data gathered from the xrt and uvot instruments onboard _
swift_. observations continued with both the xrt and the uvot up until 1.2@xmath122s .
the analysis of the afterglow began by looking at the origin of the three power - law segments in the x - ray lightcurve .
the x - ray lightcurve commences with a rise until 420s .
the second segment decays slowly with @[email protected] until 5000s .
the third and last segment decays faster with a slope of @[email protected] .
the optical lightcurve decays as a single power - law with @[email protected] . through modelling of the 400s-1000s x - ray to optical sed
, it was determined that the best fit is a power - law with milky way extinction . however ,
when extinction in the host galaxy is accounted for , the optical points lie above the x - ray power - law at late times , indicating that the optical afterglow could not be produced by the same component as the x - ray emission .
this result rules out single component afterglows with 99 per cent confidence .
the next logical step was to examine a multi - component outflow .
several variations of a two component jet , the simplest multi - component outflow , were investigated .
of all the possibilities , we find the most physically self - consistent model is one in which both a narrow and wide component are energy injected for the duration of the observed afterglow and the x - ray break at 5000s is due to a jet break in the narrow component .
this paper has wide implications for the grb community .
we have found that the explanation obtained by examining only the x - ray data of grb 050802 differs remarkably from the answer obtained by examining the optical and x - ray emission together .
our results suggest that without the optical lightcurve the correct interpretation of grb afterglows may not be possible .
sro acknowledges the support of a pparc studentship .
sz thanks pparc for its support through a pparc advanced fellowship .
we also thank the referee for useful comments and suggestions .
band d. et al . , 2005 ,
3734 bathelmy s.d .
et al . , 2005 , space science review , 120 , 143 burrows d.n .
et al . , 2005a , space science review , 120 , 165 burrows d.n .
et al . , 2005b , science , 309 , 1833 bj@xmath124rnsson g. , gudmundsson e. h. , jhannesson g. , 2004 , apj , 615 , l77 capalbi m et al . , 2006 , accepted a&a , astro - ph/0610845 dai z.g . ,
lu t. , 1998a , a&a , 333 , l87 dickey , j.m . , lockman , f.j .
, 1990 , annual review a&a , 28 , 215 drenkhahn , g. , 2002 , a&a , 386 , 714 frail , d.a . , 2001 ,
apj , 562 , l55 fynbo j.u .
et al . , 2005a , gcn circ .
3749 fynbo j.u .
et al . , 2005b , gcn circ .
3756 giannios d. & spruit h.c . , 2006 ,
a&a , 450 , 887 giannios d. , 2006 , a&a , 455 , l5 grupe d. et al . , 2006 ,
apj , submitted , astro - ph/0611240 krimm h , a. et al . , 2006 ,
apj , 648 , 1117 kumar p. , piran t. , 2000a , apj , 532 , 286 kumar p. , granot j. , 2001 , apj , 591 , 1075 mcgowan k. , band d. , brown p. , gronwall c. , huckle h. , hancock b. , 2005 , gcn circ .
3739 mszros p. , rees m.j .
, 1997 , apj , 476 , 232 nousek j. a. et al , 2006 , apj , 642 , 389 palmer d. et al . , 2005 ,
3737 panaitescu a. & kumar p. , 2004 ,
mnras , 350 , 213 panaitescu a. et al . , 2006a , apj , mnras , 369 , 2059 panaitescu a. et al . ,
2006b , mnras , 366 , 1357 pavlenko e. et al .
, 2005 , gcn circ .
3744 pei c. , 1992 , apj , 394 , 130 peng f. et al .
, 2005 , apj , 626 , 966 perna r. , lazzati d. , 2002 , apj , 580 , 261 rees m.j .
, mszros p. , 1998 ,
apj , 496 , l1 romano , p. et al . , 2006 ,
a&a , 450 , 59 roming p. et al . , 2005 , space science reviews , 120 , 95 sari r. , 1997 , apj , 489 , 37 sari r. , piran t. , 1997 , apj , 585 , 270 sari r. , piran t.,helpen j.p . , 1999 , apj , 519l , 17s schady p. , et al .
, 2007 , mnras , doi:10.1111/j.1365 - 2966.2007.11592.x starling r.l.c . , et al . , 2006 ,
apj , accepted , astro - ph/0610899 tashiro m , s. et al . , 2006 , accepted pasj suzaku special issue ( vol .
58 ) , astro - ph/0609437 vlahakis n. et al . , 2003 ,
apj , 594 , l23 willingale r. et al . , 2006 ,
apj , submitted , astro - ph/0612031 zhang b. , fan y.z . , dyks j. et al . , 2006 , apj , 642 , 354 zhang b. , kobayashi s. , 2005 , apj , 628 , 315 zhang b. , mszros p. , 2001
, apj , 552 , l35 zhang b. , mszros p. , 2002 , apj , 566 , 712 zhang b. , mszros p. , 2004
, ijmpa , 19 , 2385 | this paper investigates grb 050802 , one of the best examples of a _ swift _ gamma - ray burst afterglow that shows a break in the x - ray lightcurve , while the optical counterpart decays as a single power - law .
this burst has an optically bright afterglow of 16.5 magnitude , detected throughout the 170 - 650 nm spectral range of the uvot on - board _
swift_. observations began with the xrt and uvot telescopes @xmath0 after the initial trigger and continued for @xmath1 .
the x - ray lightcurve consists of three power - law segments : a rise until 420s , followed by a slow decay with @[email protected] until 5000s , after which , the lightcurve decays faster with a slope of @[email protected] .
the optical lightcurve decays as a single power - law with @[email protected] throughout the observation .
the x - ray data on their own are consistent with the break at 5000s being due to the end of energy injection . modelling the optical to x - ray spectral energy distribution , we find that the optical afterglow can not be produced by the same component as the x - ray emission at late times , ruling out a single component afterglow .
we therefore considered two - component jet models and find that the x - ray and optical emission is best reproduced by a model in which both components are energy injected for the duration of the observed afterglow and the x - ray break at 5000s is due to a jet break in the narrow component .
this bright , well - observed burst is likely a guide for interpreting the surprising finding of _ swift _ that bursts seldom display achromatic jet breaks .
[ firstpage ] gamma - rays : bursts |
market impact is the expected price change conditioned on initiating a trade of a given size and a given sign .
understanding market impact is important for several reasons .
one motivation is practical : to know whether a trade will be profitable it is essential to be able to estimate transaction costs , and in order to optimize a trading strategy to minimize such costs , it is necessary to understand the functional form of market impact .
another motivation is ecological : impact exerts selection pressure against a fund becoming too large , and therefore is potentially important in determining the size distribution of funds . finally , an important motivation is theoretical : market impact reflects the shape of excess demand , the understanding of which has been a central problem in economics since the time of alfred marshall . in this paper
we present a theory for the market impact of large trading orders that are split into pieces and executed incrementally .
we call these _ metaorders_. the true size of metaorders is typically not public information , a fact that plays a central role in our theory .
the strategic reasons for incremental execution of metaorders were originally analyzed by kyle ( @xcite ) , who developed a model for an inside trader with monopolistic information about future prices .
kyle showed that the optimal strategy for such a trader is to break her metaorder into pieces and execute it incrementally at a uniform rate , gradually incorporating her information into the price . in kyle s theory the price increases linearly with time as the trading takes place , and all else being equal , the total impact is a linear function of size .
the prediction of linearity is reinforced by huberman and stanzl ( @xcite ) who show that , providing liquidity is constant , to prevent arbitrage permanent impact must be linear .
real data contradict these predictions : metaorders do not show linear impact .
empirical studies consistently find concave impact , i.e. impact per share decreases with size .
it is in principle possible to reconcile the kyle model with concave dependence on size by making the additional hypothesis that larger metaorders contain less information per share than smaller ones , for example because more informed traders issue smaller metaorders .
a drawback of this hypothesis is that it is neither parsimonious nor easily testable , and as we will argue here , under the assumptions of our model it violates market efficiency .
huberman and stanzl are careful to specify that linearity only applies when liquidity is constant .
in fact , liquidity fluctuates by orders of magnitude and has a large effect on price fluctuations .
empirical studies find that order flow is extremely persistent , in the sense that the autocorrelation of order signs is positive and decays very slowly .
no arbitrage arguments imply either fluctuating asymmetric liquidity as postulated by lillo and farmer ( @xcite ) , or no permanent impact , as discussed by bouchaud et al .
( @xcite ) .
the central goal of our model is to understand how order splitting affects market impact . whereas kyle assumed a single , monopolistic informed trader ,
our informed traders are competitive .
they submit their orders to an algorithmic execution service that bundles them together as one large metaorder and executes them incrementally .
we show that this leads to a symmetric nash equilibrium satisfying the condition that the final price after a metaorder is executed equals its average transaction price .
we call this condition _ fair pricing _ , to emphasize the fact that under this assumption trading a metaorder is a breakeven deal neither party makes a profit as a result of trading . our equilibrium is less general than kyle s in that it assumes uniform execution , but it is more general in that it allows an arbitrary information distribution .
this is key because , as we show , there is an equilibrium between information and metaorder size , making it possible to match the metaorder size distribution to empirical data .
combining the fair pricing condition with a martingale condition makes it possible to derive the price impact of metaorders as a function of the metaorder size distribution .
this allows us to make several strong predictions based on a simple set of hypotheses . for a given metaorder size distribution
it predicts the average impact as a function of time both during and after execution .
we thus predict the relationship between the functional form of two observable quantities with no a priori relationship , making our theory falsifiable in a strong sense .
this is in contrast to theories that make assumptions about the functional form of utility and/or behavioral or institutional assumptions about the informativeness of trades , which typically leave room for interpretation and require auxiliary assumptions to make empirical tests . for example , gabaix et al .
( @xcite ) have also argued that the distribution of trading volume plays a central role in determining impact , and have derived a formula for impact that is concave under some circumstances . however , in contrast to our model , their prediction for market impact depends sensitively on the functional form for risk aversion , where @xmath0 is the standard deviation of profits , the impact will increase with the size @xmath1 of the metaorder as @xmath2 .
thus the impact is concave if @xmath3 , linear if @xmath4 ( i.e. if risk is proportional to variance ) , and convex otherwise .
for another theory that also predicts concave impact see toth et al .
( @xcite ) . ] . our theory , in contrast , is based entirely on market efficiency and does not depend on the functional form of utility .
our work here is related to several papers that study market design .
viswanathan and wang ( @xcite ) , glosten ( @xcite ) , and back and baruch ( @xcite ) derive and compare the equilibrium transaction prices of orders submitted to markets with uniform vs. discriminatory pricing . depending on the setup of the model , these prices can be different so that investors will prefer one pricing structure to the other and can potentially be cream - skimmed " by a competing exchange .
the fair pricing condition we introduce here forces the average transaction price of a metaorder ( which transacts at discriminatory prices ) to be equal to the price that would be set under uniform pricing .
fair pricing , therefore , means investors have no preference between the two pricing structures , and they have no incentive to search out arrangements for better execution . on the surface
, this result is similar to the equivalence of uniform and discriminatory pricing in back and baruch ( @xcite ) .
however , in their paper , this equivalence results because orders are always allowed to be split , whereas ours is a true equivalence between the pricing of a split vs. unsplit order . in section ii
we give a description of the model and discuss its interpretation . in section iii
we develop the consequences of the martingale condition and show how this leads to zero overall profits and asymmetric price responses when order flow is persistent .
in section iv we show that any nash equilibrium must satisfy the fair pricing condition . in section
v we derive in general terms what this implies about market impact . in section
vi we introduce specific functional forms for the metaorder size distribution and explicitly compute the impact for these cases .
finally , in section vii we discuss the empirical implications of the model and make some concluding remarks .
we study a stylized model of an algorithmic trading service combining and executing orders of long - term traders .
this can be thought of as a broker - dealer receiving multiple orders on the same security and executing them algorithmically at the same time , or as an institutional trading desk of a large asset manager combining the orders from multiple portfolio managers into one large metaorder .
our goal is to model the price impact of a metaorder during a trading period in which it may or may not be present .
we set up the model in a stylized manner as a game in which trading takes place across multiple periods , after which final prices are revealed . while this is somewhat artificial in comparison to a real market ( which has no such thing as a final " price ) , the framework is simple enough to allow us to find a solution , and the basic conclusions should apply more broadly .
the structure of the model is in many respects similar to the classic framework of kyle ( @xcite ) , but with several important differences .
( a point - by - point comparison is made in the conclusions ) .
we begin with an overview .
there is a single asset which is traded in each of @xmath1 periods , which can be regarded as a game .
there are three kinds of agents , long - term traders , market makers and day traders .
the long - term traders have a common information signal that is received before the game starts ; based on this information they formulate their orders and submit them to an algorithmic trading service .
the algorithmic service operates mechanically , dividing the bundle of orders ( which we call a _ metaorder _ ) into equal sized lots which are submitted as anonymous market orders in each successive period until the metaorder is fully executed .
the day traders , in contrast , receive a new information signal and submit orders based on this signal in every period of the game . in every period
each market maker observes the netted order of the long - term and day traders and submits a quote accordingly ; all orders are filled at the best price .
the game ends with a final liquidation period at time step @xmath5 in which prices are set exogenously to reflect the accumulated information .
the length @xmath1 of the game varies based on the amount of information received , and is unknown to the market makers .
we now discuss the setup of the game in more detail , beginning with a description of each of the agents .
the _ long - term traders _ receive a common information signal @xmath6 before the game begins , which only they observe . in order to model situations in which there may or may not be a metaorder present , we allow a nonzero probability that @xmath7 . thus with probability @xmath8 the signal @xmath6 is drawn from a distribution @xmath9 , which has nonzero support over a continuous interval @xmath10 , where @xmath11 , and with probability @xmath12
there is no information , i.e. @xmath7 . for simplicity we discuss the case where the draw of @xmath6 happens to be positive , i.e. it causes the final price to increase , but the results apply equally well if @xmath6 is negative .
after @xmath6 is revealed the orders of the long - term traders are aggregated together into a _ metaorder _ and executed in a package .
the bundling and execution process can be thought of as representing an algorithmic trading firm , broker , or institutional trading desk .
there are @xmath13 long - term traders , labeled by an index @xmath14 , where @xmath13 is a large number . after the common information signal
is received each long - term trader submits an order of size @xmath15 , where @xmath16 is a large positive integer .
each long - term trader decides @xmath17 independently .
the individual orders are bundled together into a metaorder of size @xmath18 shares . .
for this figure we assume the metaorder is present and show only expected price paths , averaged over the day trader s information .
the price is initially @xmath19 ; after the first lot is executed it is @xmath20 .
if @xmath21 it is finished and the price reverts to @xmath22 , but if @xmath23 another lot is executed and it rises to @xmath24 .
this proceeds similarly until the execution of the metaorder is completed . at any given point
the probability that the metaorder has size @xmath25 , i.e. that the order continues , is @xmath26 . if had we followed a typical price path under circumstances when the day trader s noisy information signal is large , rather than the expected price paths shown here , the sequence of prices would be a random walk with a time - varying drift caused by the metaorder s impact . ]
the algorithmic trading firm operates purely mechanically , chopping the metaorder into equal pieces and submitting market orders at successive times @xmath27 .
these are executed at transaction prices @xmath28 , where @xmath29 , as illustrated in figure [ fig1 ] .
we assume the metaorder is executed in lots , which for mathematical convenience are chosen to be of @xmath13 shares .
the imposition of a maximum trade size @xmath30 is a technical detail , which induces a bound on @xmath30 , to avoid mathematical problems that occur in the limit @xmath31 .
this is explained in the appendix .
for the typical situations we have in mind @xmath30 , @xmath13 , and @xmath32 are all large numbers .
the choice of lots of size @xmath13 shares is purely a matter of convenience ; the assumption that the lots have a constant size is something we hope can be relaxed in the future . ]
; we assume that @xmath30 is large .
if the metaorder is present ( i.e. if @xmath33 ) , trading ends when the metaorder is fully executed , i.e. when @xmath34 .
the equilibrium distribution of metaorder lengths is @xmath35 . if @xmath7 we randomly choose a value of @xmath1 from an arbitrary distribution @xmath36 ( which in general differs from @xmath35 ) .
the _ day traders _ can be treated as a single representative agent who receives a private information signal @xmath37 at the beginning of each period @xmath38 of the game , and submits a market order ( either to buy or sell ) of size @xmath39 , where @xmath37 is a zero mean iid noise process with an otherwise arbitrary distribution @xmath40 , and @xmath41 is an increasing function is not important here .
we assume the day traders do not engage in order splitting , i.e. they trade on the information they receive in given time step only in that time step .
this guarantees that in the absence of a metaorder the day trader s order flow provides a sufficient signal for a market maker to infer the new information . ] .
there is no restriction on the size of @xmath42 , and in particular we allow the possibility that @xmath37 is large and negative while @xmath6 is positive , so that during the course of execution of the metaorder the combined order flow may change sign , and that at different times the market makers may buy or sell . at each time
step the market orders of the long - term traders and the day trader are netted and the single combined order is submitted to the _ market makers _ , who are competitive and profit maximizing .
we are not assuming any special institutional privileges , such as those of the specialists in the nyse ; our market makers are simply competitive liquidity providers . at each time step
each market maker observes the combined order and submits a quote .
the combined order is fully executed by the market maker(s ) offering the best price .
the market makers are able to take past order flow and prices into account in setting their quotes . given that @xmath37 is iid , the typical transaction price sequence @xmath43 will look like a random walk , as the market maker responds to the day trader s order flow , with a possible superimposed drift if a metaorder is present .
in the final period @xmath44 the combination of the long - term information @xmath6 and the accumulated short term information are revealed . since the long - term traders information signals are independent of those of the day trader , information is additive and the final price is @xmath45 where @xmath19 is the initial price .
the average final price is @xmath46 = s_0 + \alpha$ ] , where @xmath47 implies an average over the noise @xmath37 , @xmath48 = \int x \hat{p}(\eta ) d\eta.\ ] ] similarly the average transaction price at time @xmath38 is @xmath49 $ ] .
the goal of the paper is to compute the _ average immediate impact _ @xmath50 and the _ average permanent impact _
@xmath51 of the metaorder .
the corresponding incremental average impacts are @xmath52 and @xmath53 .
as we will show , at the equilibrium the average impacts do not depend on @xmath8 , @xmath54 or @xmath55 .
the number of long - term traders @xmath13 is fixed and is common knowledge .
the market makers know the initial price @xmath19 , the information distributions @xmath9 and @xmath40 , they can deduce the function @xmath41 relating the day trader s information to their order size , and they observe the net combined order in each period and remember previous order flow .
however they do not know the information signals @xmath6 or @xmath37 , and thus they do not know how much of the order flow to ascribe to the long - term trader vs. the day trader . they do not know whether a metaorder is present , and if it is , they do not know its size .
perhaps the strongest assumption we have made concerns knowledge of the timing of metaorders . the market makers know the period @xmath38 , and as a result if a metaorder is present , they know when it started . in typical market settings order flow is anonymous and the starting time of a metaorder is uncertain .
nonetheless , for long meta - orders and sufficiently high participation rates the starting time can be inferred from the imbalance in order flow , where @xmath56 is the desired number of standard deviations of statistical significance .
the accumulated imbalance after @xmath38 steps is @xmath57 , where @xmath58 is the participation rate .
equating these gives @xmath59 .
hiding an order of size @xmath1 requires @xmath60 .
thus larger metaorders need to be executed more slowly to avoid detection .
since the time needed to complete execution is inversely proportional @xmath58 , for a large metaorder this can become prohibitive it is impossible to escape detection . ] . assuming a typical participation rate of @xmath61 and two standard deviations to reject the null hypothesis of balanced order flow means that a metaorder can be detected after about @xmath62 timesteps .
large metaorders are frequently executed in @xmath63 or even @xmath64 lots share of ibm . ] , implying an error in inferring the starting time of @xmath65 to @xmath66 .
thus , it is not unrealistic to imagine that market makers can infer the presence of large metaorders and estimate their starting times .
we are not concerned here with the question of whether or not the algorithmic execution service s strategy of splitting the order into equal pieces is an optimal strategy .
our goal is instead to assume that this is what they do , and to derive the implications for price impact .
our main result is to derive the equilibrium relationship between the metaorder size distribution @xmath35 and the average immediate and permanent impacts @xmath67 and @xmath68 .
in addition to taking expectations over the day trader s noise @xmath42 , which we denote by @xmath47 , we must compute expectations about the length of the game .
the crux of our argument hinges around the market makers ignorance of @xmath6 ; when @xmath33 this translates into uncertainty about metaorder size .
we will use the notation @xmath69 to represent an average over all metaorders of size @xmath70 , and as described above , @xmath47 for averages over @xmath37 . for a generic function
@xmath71 the average over metaorder sizes is @xmath72 = \sum_{n = t}^\infty q_n f_n= \frac{\sum_{n = t}^\infty p_n f_n}{\sum_{n = t}^\infty p_{n}}= \frac{\sum_{i=0}^\infty p_{it } f_{it}}{\sum_{i=0}^\infty p_{it } } , e_t[f_n ] = \frac{\sum_{n = t}^m p_n f_n}{\sum_{n = t}^m p_{n}}= \frac{\sum_{i=0}^{m - t } p_{t+i } f_{t+i}}{\sum_{i=0}^{m - t } p_{t+i } } , \label{averages}\ ] ] where in the last term we made the substitution @xmath73 . assuming that a metaorder is present , the likelihood that it will persist depends on the distribution @xmath35 and the number of executions @xmath38 that it has already experienced
let @xmath74 be the probability that the metaorder will continue given that it is still active at timestep @xmath38 .
this is equivalent to the probability that it is at least of size @xmath38 , i.e. @xmath75 this makes precise how order splitting can make order flow positively autocorrelated . in particular ,
if @xmath35 has heavy tails than an exponential @xmath26 will increase with time and induce persistence in order flow .
in the case where no metaorder is present one can similarly define the probability that the game continues as @xmath76
in this section we introduce a martingale condition and discuss its implications for liquidity and overall profitability .
market makers must set prices given only past and present order flow information , without knowing whether the order flow originated from a metaorder or from day traders .
their decision function is of the form @xmath77 where @xmath78 $ ] .
as we show below , we are able to finesse this difficult problem by imposing a martingale condition and averaging over the day traders noise , which is sufficient for the main goal of this paper of deriving the equilibrium between the impacts and @xmath35 . during the game
the market makers may also use information from order flow and prices to update their prior probability @xmath8 for the presence of a metaorder to a more accurate value @xmath79 .
we assume that transaction prices are a martingale , so that the current transaction price @xmath28 is equal to the expected future price .
define an indicator variable @xmath80 with @xmath81 is the metaorder is present and @xmath82 if it is absent . for the price in the next period it is necessary to average over four possibilities : 1 .
with probability @xmath83 the metaorder is present and with probability @xmath26 trading continues . in this case
@xmath84 = \tilde{s}'_t + \tilde{r}'_t$ ] , where @xmath85 $ ] .
2 . with probability @xmath83
the metaorder is present and with probability @xmath86 time @xmath38 is the last trading period . in this case
@xmath87 = \tilde{s}'_t - { r}'_t$ ] , where @xmath88 $ ] .
3 . with probability @xmath89
the metaorder is not present and with probability @xmath90 trading continues .
since the day trader s information @xmath37 is zero mean and iid , the expected average transaction price on the next time step conditioned on the current transaction price must satisfy @xmath91 = \tilde{s}'_t$ ] , i.e. the average transaction price is unchanged .
4 . with probability @xmath89
the metaorder is not present and with probability @xmath92 time @xmath38 is the last trading period . for similar reasons @xmath93
= \tilde{s}'_t$ ] , i.e. the average final price conditioned on the current transaction price is equal to the current transaction price . thus the martingale condition can be written @xmath94 + ( 1 - \mathcal{p}_t)\hat{e } [ { s}'_{t + 1 } | \tilde{s}'_t , m = t ] \right)\\ + ~(1 - \mu'_t ) & \left ( \hat{\mathcal{p}}_t \hat{e } [ \tilde{s}'_{t + 1 } | \tilde{s}'_t ,
m = f ] + ( 1 - \hat{\mathcal{p}}_t ) \hat{e } [ { s}'_{t + 1 } | \tilde{s}'_t ,
m = f ] \right)\\\end{aligned}\ ] ] substituting for the average next price conditioned on the current price for cases 1 - 4 gives : @xmath95 the current transaction price @xmath28 and @xmath83 both cancel and this reduces to @xmath96 since @xmath37 is mean zero and iid , @xmath97 = \tilde{r}_t = \tilde{s}_{t+1}- \tilde{s}_t$ ] and similarly for @xmath98 .
taking averages over @xmath37 gives @xmath99 equation ( [ shortterm ] ) holds for @xmath100 . if the metaorder has maximal length at the end of the @xmath30th interval by definition @xmath101 , which implies that @xmath102 .
let us pause for a moment to digest this result .
we started with a martingale condition for realized prices , including fluctuations caused by day traders , and then reduced it to a martingale condition for the average impact due to the presence of a metaorder .
the reduced martingale no longer depends on @xmath103 , @xmath36 , or the day trader s information .
the fact that we assume a martingale for realized prices implies that arbitrage of the impact is impossible .
the ability to average away the day traders is a consequence of our assumption that @xmath6 and @xmath37 are independent , which implies additivity of information .
this separates the problem of the metaorder s impact from that of the day trader s impact the metaorder s impact effectively rides on top of the day trader s impact . as we will see , the virtue of this approach is that it allows us to infer quite a lot without needing to solve for the market makers optimal quote setting function @xmath104 .
equation ( [ shortterm ] ) can be trivially rewritten in the form @xmath105 where @xmath52 and @xmath53 .
thus the martingale condition fixes the ratio of the price responses @xmath106 and @xmath107 , but does not fix their scale . if @xmath74 is large , corresponding to a metaorder that is likely to continue , then @xmath108 small .
this means that the price response if the order continues is much less it is than if it stops . to complete the calculation
we need another condition to set the scale of the price responses @xmath106 and @xmath107 , which may change as @xmath38 varies .
such a condition is introduced in section [ sizeindependencesec ] .
even without such a condition , one can already see intuitively that all else equal " , for a heavy - tailed metaorder distribution @xmath35 , the impact will be concave .
( recall that heavy tails in @xmath35 imply that @xmath26 increases with @xmath38 ) .
the martingale condition implies that the market makers break even overall , i.e. that their total profits summing over metaorders of all sizes is zero .
this is stated more precisely in proposition one .
* proposition 1 . *
the market makers transact @xmath1 lots at average prices @xmath109 , which later are all valued at a final price @xmath110 .
the martingale condition implies zero overall profits , i.e. @xmath111\equiv\sum_{n=1}^\infty p_n \pi_n=0 .
\pi = e_1[\pi_n]\equiv\sum_{n=1}^m n \pi_n p_n = 0 , \label{breakevenonaverage}\ ] ] where @xmath112 is the profit per lot transacted .
the proof of proposition 1 is given in appendix a. the phrase overall profits " emphasizes that the martingale condition only implies zero profits when averaged over metaorders of all sizes .
it allows for the possibility that the market makers may make profits on metaorders in a given size range , as long as they take corresponding losses in other size ranges .
surprisingly , proposition 1 is not necessarily true when @xmath30 is infinite .
the basic problem is similar to the st .
petersburg paradox : as the metaorder size becomes infinite it is possible to have infinitely rare but infinitely large losses . the conditions under which this holds are more complicated , as discussed in appendix a.
we now derive the _ fair pricing condition _ , which states that for any @xmath1 @xmath113 under fair pricing the average execution price is equal to the final price .
we call this fair pricing for the obvious reason that both parties would naturally regard this as fair " .
fair pricing implies that the market makers break even on metaorders of any size , as opposed to the martingale condition , which only implies they break even when averaging over metaorders of all sizes . .
* proposition 2 .
* _ if the immediate impact @xmath114 has a second derivative bounded below zero , in the limit where the number of informed traders @xmath115 , any nash equilibrium must satisfy the fair pricing condition @xmath116 for @xmath29 . on average market makers profit from orders of length one and take ( equal and opposite ) losses from orders of length @xmath30 .
_ this result is driven by competition between informed traders .
all informed traders receive the same information signal @xmath6 , and the strategy of informed trader @xmath117 is the choice of the order size @xmath17 .
the orders are then bundled together to determine the combined metaorder size @xmath118 .
the decision of each informed trader is made without knowing the decisions of others .
the derivation has two steps : first we examine the case @xmath119 for @xmath29 , and show that if others hold their strategies constant , providing the impact is concave and @xmath13 is sufficiently large , traders can increase profits by changing strategy . secondly we show that if @xmath116 there is no incentive to change strategy
. then we return to examine the cases @xmath21 and @xmath120 , which must be treated separately .
the derivation is given in the appendix .
in contrast to the martingale condition , which only implies that immediate profits are zero when averaged over size , fair pricing means that they are identically zero for every size .
it implies that no one pays any costs or makes any profits simply by trading in any particular size range .
the nash equilibrium is symmetric , i.e. all agents make the same decision .
( this must be in any case since there is nothing to distinguish them ) .
this means that there is a unique order size @xmath121 for any @xmath6 , and the distribution of information @xmath9 implies the distribution of metaorder size @xmath35 .
we can use @xmath35 as a proxy for @xmath9 , which is the key fact allowing us to state our results in terms of the observable quantity @xmath35 rather than @xmath9 , which is much more difficult to observe .
although this derivation is based on rationality , the fair pricing condition potentially stands on its own , even if other aspects of rationality and efficiency are violated . in modern markets portfolio managers routinely receive trade cost analysis reports
that compare their execution prices relative to the close , which for a trade that takes place over one day is a good proxy for @xmath110 .
such reports are typically broken down into size bins , making persistent inconsistencies across sizes clear .
execution times for metaorders range from less than a day to several months ( vaglica et al . , @xcite ) , and are much shorter than typical holding times , which for mutual funds are on average a year and are often much more ( schwartzkopf and farmer , @xcite ) .
thus the statistical fluctuations for assessing whether fair pricing holds are much smaller than those for assessing informational efficiency .
since @xmath122 is fairly well determined , portfolio managers will exert pressure on their brokers to provide them with good execution . as a result
we expect the fair pricing condition to be obeyed to a higher degree of accuracy than the informational efficiency condition .
although we have derived the nash equilibrium only in the case where the immediate impact is concave , for the remainder of the paper we will simply assume that the martingale condition holds for all @xmath1 and the fair pricing condition holds for @xmath123 .
this allows us to derive both the immediate impact @xmath124 and the permanent impact @xmath68 for any given metaorder size distribution @xmath35 .
we later argue that for realistic situations the metaorder size distribution gives rise to a concave impact function , consistent with the nash equilibrium .
the martingale condition ( eq . [ shortterm ] ) and the fair pricing condition ( eq . [ fairpricing ] ) define a system of linear equations for @xmath125 and @xmath126 at each value of @xmath38 , which we can alternatively express in terms of the price differences @xmath127 and @xmath52 , where @xmath128 .
the martingale condition holds for @xmath129 and the fair pricing condition holds for @xmath130 .
there are thus @xmath131 homogeneous linear equations with @xmath132 unknowns does not exist , so @xmath133 is not needed .
this reduces the number of unknowns by one . ] . because the number of unknowns is one greater than the number of conditions there is necessarily an undetermined constant , which we choose to be @xmath134 .
* proposition 3 .
* _ the system of martingale conditions ( eq . [ shortterm ] ) and fair pricing conditions ( eq . [ fairpricing ] ) has solution _ @xmath135 the proof is given in appendix a. an important property of the solution is the equivalence of the impact @xmath114 as a function of either time or size .
this is in contrast to the prediction of an extended " kyle model under the assumption that traders of different sizes are differently informed , which yields linear impact as a function of time , but allows the slope to vary nonlinearly with size .
summing eq .
[ solution ] implies that for @xmath136 the immediate impact is @xmath137 for @xmath138 the immediate impact is @xmath139 and for @xmath140 it is @xmath141 .
( the meaning of the undetermined constants @xmath142 and @xmath143 is discussed in a moment ) . the permanent impact @xmath144 is easily obtained .
making some simple algebraic manipulations @xmath145 by combining eqs .
( [ final ] ) and ( [ calpdef ] ) we get @xmath146 we have expressed both the permanent and immediate impact purely in terms of @xmath35 and the undetermined constants @xmath143 and @xmath142 . the undetermined constants can in principle be fixed based on the information at the equilibrium . at the equilibrium information signals in the range @xmath147 $ ] will be assigned to metaorders of size one , with an average size @xmath148 , signals in the range @xmath149 $ ] will be assigned to metaorders of size two , with an average size @xmath150 , and so on .
the scale of the impact is set by the relations @xmath151 we have used the words in principle " because , unlike metaorder sizes , information is not easily observed . barring the ability to independently measure information , the constants @xmath143 and @xmath142 remain undetermined parameters .
@xmath152 plays the important role of setting the scale of the impact .
the constant @xmath143 , in contrast , is unimportant it is simply the impact of the first trade , before the metaorder has been detected .
the power of the theory developed here is that the impact is predicted in terms of @xmath35 , which is directly measurable ( at least with the proper data ) .
the calculation to set the scale shows how the continuous variable @xmath6 maps onto the discrete variable @xmath1 .
@xmath153 is a discrete function whose inverse can be written @xmath154 .
if there is a continuum limit for large @xmath1 , the two distributions are related by conservation of probability as @xmath155 so for example , if the empirical metaorder size is @xmath35 is asymptotically pareto distributed , as argued in the next section , @xmath156 , @xmath157 , and @xmath158 .
based on the empirically observed value @xmath159 , this gives @xmath160 , which means that the cumulative scales as @xmath161 .
this is what is typically observed for price returns in american stock markets ( plerou et al . , @xcite ) .
we have so far left the metaorder size distribution @xmath35 unspecified . in this section
we compute the impact for two examples .
the first of these is the pareto distribution , @xmath162 which we argue is well - supported by empirical data means that there exists a constant @xmath163 such that in the limit @xmath164 , @xmath165 .
we use it to indicate that this relationship is only valid in the limit of large metaorder size @xmath1 . ] .
the second is the stretched exponential distribution , which is not supported by data , but provides a useful point of comparison .
there is now considerable accumulated evidence that in the large size limit in most major equity markets the metaorder size @xmath166 is distributed as @xmath167 , with @xmath159 .
* _ trade size .
_ in many different equity markets for large trades the volume @xmath166 has been observed by several groups to be distributed as a power law ( gopikrishnan et al .
( @xcite ) ; gabaix et al .
( @xcite ) is somewhat controversial , however : eisler and kertesz ( @xcite ) and racz et al .
( @xcite ) have argued that the correct value of @xmath168 . ] . this relationship becomes sharper if only block trades are considered ( lillo et al . , @xcite ) .
* _ long - memory in order flow . _ the signs of order flow in many equity markets are observed to have long - memory .
we use the term long - memory in its more general sense to mean any process whose autocorrelation function is non - integrable ( beran , @xcite ) . this can include processes with structure breaks , such as that studied by ding , engle and granger ( @xcite ) . ]
this means that the transaction sign autocorrelation function @xmath169 decays in time as @xmath170 , where @xmath171 . under a simple theory of order splitting the exponent @xmath172 , which is in good agreement with the data [ ( lillo et al . ,
@xcite ) ; ( gerig , @xcite ) ; bouchaud , farmer , and lillo ( @xcite ) ] . *
_ reconstruction of large metaorders from brokerage data .
_ vaglica et al .
( @xcite ) reconstructed metaorders spanish stock exchange using data with brokerage codes and found that @xmath1 is distributed as a power law for large @xmath1 with @xmath173 .
there is thus good evidence that metaorders have a power law distribution , though more study is of course needed . in this section
we derive the functional form of the impact for pareto metaorder size .
we do this by using eq .
[ final ] in the limit as @xmath31 , and return in section [ finitesize ] and in appendix b to discuss how this is modified when @xmath30 is finite . while we only care about the asymptotic form for large @xmath1 , for convenience we assume an exact pareto distribution for all @xmath1 , i.e. @xmath174 where the normalization constant @xmath175 is the riemann zeta function . for the pareto distribution
the probability @xmath74 that an order of size @xmath38 will continue is @xmath176 where @xmath177 is the generalized riemann zeta function ( also called the hurwitz zeta function ) . the approximations are valid in the large @xmath38 limit . the immediate impact can be easily calculated from eq .
( [ final ] ) .
for pareto distributed metaorder sizes , using eqs .
( [ solution2 ] ) and ( [ pareto2eq ] ) , @xmath107 is @xmath178 thus the immediate impact @xmath179 behaves asymptotically for large @xmath38 as @xmath180 the exponent @xmath181 has a dramatic effect on the shape of the impact .
for lorenzian distributed metaorder size ( @xmath182 ) the impact is logarithmic , for @xmath183 it increases as a square root , for @xmath184 it is linear , and for @xmath185 it is superlinear .
thus as we vary @xmath181 the impact goes from concave to convex , with @xmath184 as the borderline case is special is that for @xmath186 the second moment of the pareto distribution is undefined . under the theory of lillo et al .
( @xcite ) , long - memory requires @xmath186 . ] .
( black circles ) and permanent impact @xmath187 ( red squares ) for @xmath183 .
the dashed line is the price profile of a metaorder of size @xmath188 , demonstrating how the price reverts from immediate to permanent impact when metaorder execution is completed .
the inset shows a similar plot in double logarithmic scale for a wider range of sizes ( from @xmath189 to @xmath190 ) .
the blue dashed line is a comparison to the asymptotic square root scaling . _
bottom panel _ : expected immediate impact @xmath114 as a function of time @xmath38 for tail exponents @xmath191 and @xmath192 , illustrating how the impact goes from concave to convex as @xmath181 increases . , title="fig : " ] ( black circles ) and permanent impact @xmath187 ( red squares ) for @xmath183 . the dashed line is the price profile of a metaorder of size @xmath188 , demonstrating how the price reverts from immediate to permanent impact when metaorder execution is completed .
the inset shows a similar plot in double logarithmic scale for a wider range of sizes ( from @xmath189 to @xmath190 ) .
the blue dashed line is a comparison to the asymptotic square root scaling .
_ bottom panel _ : expected immediate impact @xmath114 as a function of time @xmath38 for tail exponents @xmath191 and @xmath192 , illustrating how the impact goes from concave to convex as @xmath181 increases . , title="fig : " ] figure [ numerical ] illustrates the reversion process for @xmath183 and shows how the shape of the impact varies with @xmath181 . the permanent impact under the pareto assumption is easily computed using eq .
( [ generalpermanentimpact ] ) .
a direct calculation shows that @xmath193 eqs .
( [ permanentimpact ] ) and ( [ temporaryimpact ] ) imply that the ratio of the permanent to the immediate impact is @xmath194 for example if @xmath183 the model predicts that on average the permanent impact is equal to two thirds of the maximum immediate impact , i.e. , following the completion of a metaorder the price should revert by one third from its peak value .
in the previous section we have assumed that @xmath195 , so that we can treat the problem as if @xmath30 were infinite . in real markets
the maximum order size is probably quite large , a significant fraction of the market capitalization of the asset .
thus we doubt that the finite support of @xmath1 has much practical importance , except perhaps for extremely large metaorders . from a conceptual point of view
, however , having an upper bound on metaorder size creates some interesting effects .
as we have already mentioned , proposition 1 fails to hold when @xmath196 , so this must be handled with some care . in appendix c
we illustrate how the results change when @xmath197 .
what we observe is that when @xmath198 the impact is roughly unchanged from its behavior in the limit @xmath31 , but when @xmath199 the impact becomes highly convex .
this is caused by the fact that the market maker knows the metaorder must end when @xmath120 . since by definition @xmath200 , the martingale condition requires that @xmath201 , i.e. there is no reversion when the metaorder is completed .
this propagates backward and when @xmath197 it significantly alters the impact , as seen in figure [ figimpact ] .
nonetheless , from a practical point of view we do not think this is an important issue , which is why we have relegated the details of the discussion to appendix c. changing the metaorder distribution has a dramatic effect on the impact . while we believe that the pareto distribution is empirically the correct functional form for metaorder size , to get more insight into the role of @xmath35 we compute the impact for an alternative functional form .
for this purpose we choose the stretched exponential , which can be tuned from thin tailed to heavy tailed behavior and contains the exponential distribution as a special case .
there is no simple expression for the normalization factor needed for a discrete stretched exponential distribution , so we make a continuous approximation , in which the metaorder size distribution is @xmath202 the normalization factor @xmath203 is the incomplete gamma function .
the shape parameter @xmath204 specifies whether the distribution decays faster or slower than an exponential .
( @xmath205 implies faster decay and @xmath206 implies slower decay . ) for short data sets , when @xmath207 is small this functional form is easily confused with a power law .
it can be shown that the stretched exponential leads to an immediate impact function that for large @xmath38 asymptotically behaves as @xmath208 this is the product of a power law and an exponential ; for large @xmath38 the exponential dominates .
the permanent impact is @xmath209 where @xmath210 is the exponential integral function and in the last approximation we have used its asymptotic expansion .
the ratio of the permanent to the immediate impact of the last transaction is @xmath211 in contrast to the pareto metaorder size distribution this is not constant . instead the ratio between permanent and immediate impact decreases with size , going to zero in the limit as @xmath212 . the fixed ratio of permanent and immediate impact is a rather special property of the pareto distribution .
the three types of agents in our model are similar to kyle s ; his informed trader is replaced by our long - term traders , and his noise traders are replaced by our day traders . in both cases we assume a final liquidation .
there are also several key differences . in our model : * our long - term traders do not have a monopoly , but rather have common information and compete in setting the size of their orders .
their orders are bundled together and executed as a package by an algorithmic execution service .
these two facts are essential to show that the fair pricing condition is a nash equilibrium . *
the number of periods @xmath1 for execution is uncertain , and depends on the information of the long - term traders .
this is important because the martingale condition is based on the market makers uncertainty about when the metaorder will terminate . *
the distribution @xmath9 is arbitrary ( whereas kyle assumed a normal distribution ) .
our key result is that information is almost fully reflected in metaorder size , i.e. that @xmath35 and @xmath9 are closely related .
this means our results apply to any empirical size distribution @xmath35 . *
our day traders respond to ongoing information signals that are permanent in the sense that they affect the final price , in contrast to kyle s noise traders , who are completely uninformed .
this means that the information in the final liquidation price is incrementally revealed .
* the most important difference is that our model has a different purpose .
kyle assumed an information signal and solved for the optimal strategy to exploit it .
we assume metaorder execution may or may not be going on in the background , and solve for its impact on prices .
the theory presented here makes several predictions with clear empirical implications . in this section
we summarize what these are and outline a few of the problems that are likely to be encountered in empirical testing . 1 . the fair pricing condition , eq .
[ fairpricing ] , is directly testable , although it requires a somewhat arbitrary choice about when enough time has elapsed since the metaorder has completed for reversion to occur .
( one wants to minimize this time because of the diffusive nature of prices , but one wants to allow enough time to make sure that reversion is complete ) .
2 . the asymmetric price response predicted by eq .
[ returnratio ] is testable .
however , this only tests the martingale condition , which is the less controversial part of our model .
the equivalence of impact as a function of time and size is directly testable . under our theory , for @xmath25
the immediate impact from the first @xmath38 steps is the same , regardless of @xmath1 .
this is in contrast to the kyle model which predicts linear impact as a function of time , but can explain concavity in size only by postulating variable informativeness of trades vs. metaorder size .
the prediction of immediate and permanent impact based on @xmath35 is directly testable through equations [ final ] and [ generalpermanentimpact ] .
if the metaorder distribution is a power law ( pareto distribution ) , then for large @xmath1 the immediate impact scales as @xmath213 and the ratio of the permanent to the immediate impact of the last transaction is @xmath214 .
see section [ paretosec ] .
prediction ( 2 ) has been tested and confirmed by lillo and farmer ( @xcite ) , farmer et al .
( @xcite ) and gerig ( @xcite ) .
preliminary results seem to support , or at least not contradict , prediction ( 5 ) . the only studies of which we are aware that attempted to fit functional form to the impact of metaorders are by torre ( @xcite ) , almgren et al .
@xcite , and moro et al .
( @xcite ) ; they find immediate impact roughly consistent with a square root functional form .
moro et al . also tested the ratio of permanent to immediate impact and found @xmath215 for the spanish stock market and @xmath216 for the london stock market , with large error bars . to our knowledge
the other predictions remain to be empirically tested . though the market makers in our model are uncertain whether or not a metaorder is present , if it is present , they know when its execution begins and ends . the ability to detect metaorders from imbalances in order flow using brokerage codes has been demonstrated [ ( vaglica et al .
@xcite ) , ( toth et al @xcite ) ] .
a recent study of metaorders based on brokerage code information found average participation rates of @xmath217 for the spanish stock market ( bme ) and @xmath218 for the london stock market , for metaorders whose average size was just under @xmath62 in both markets , making such metaorders difficult to hide .
the detection problem introduces uncertainties in starting and stopping times that may affect shape of the price impact .
the traditional view in finance is that market impact is just a reflection of information .
this point of view often goes a step further and postulates that the functional form of impact is determined by behavioral and institutional factors , such as how informed the agents are who trade with a given volume .
this hypothesis is difficult to test because it is inherently complicated and information is difficult to measure independently of impact . within the framework developed here
, such anomalies would violate the fair pricing condition .
in this paper we embrace the view that impact reflects information , but we show how at equilibrium the trading volume reflects the underlying information and makes it possible to compute the impact .
the metaorder size distribution determines the shape of the impact but does not set its scale .
metaorder size has the important advantage of being a measurable quantity , and thus predictions based on it are much more testable than those based directly on information .
the fair pricing condition that we have derived here may well hold on its own , even without informational efficiency .
this could be true for purely behavioral reasons : the fair pricing condition holds because it can be measured reliably , and both parties view it as fair .
thus while the main results here are consistent with rationality , they do not necessarily depend on it .
we provide an example solution for the pareto distribution for metaorder size because we believe that the evidence supports this hypothesis .
this gives the simple result that the impact is a power law of the form @xmath219 , and the ratio of permanent impact to the temporary impact of the last transaction is @xmath214 .
however , the bulk of our results do not depend on this assumption .
thus the reader who is skeptical about power laws may simply view the results for the pareto distribution as a worked example .
the strength of our approach is its empirical predictions . because these involve explicit functional relationships between observable variables they are strongly falsifiable in the popperian sense .
a preliminary empirical analysis seems to support the theory , but the statistical analysis so far remains inconclusive .
we look forward to more rigorous empirical tests .
we would like to acknowledge conversations with jean - philippe bouchaud and a very helpful discussion by ionid rosu .
this work was supported by prin project 2007tkltsr computational markets design and agent - based models of trading behavior `` and national science foundation grant 0624351 .
fl acknowledges financial support from the grant ' ' progetto interno lillo 2010 " by scuola normale superiore di pisa .
almgren , r.f . , 2003 .
optimal execution with nonlinear impact functions and trading - enhanced risk .
applied mathematical finance , 10 , 118 .
almgren , r.f . and chriss , n. , 1999 .
value under liquidation .
risk , 12 , 6163 .
almgren , r.f . and chriss , n. , 2000
. optimal execution of portfolio transactions .
journal of risk , 3 , 539 .
almgren , r.f . ,
thum , c. and hauptmann , h. l. , 2005 .
direct estimation of equity market impact . risk . back , kerry and baruch , shmuel , 2007 .
working orders in limit order markets and floor exchanges .
journal of finance , 62 , 15891621 .
beran , j. , 1994 .
statistics for long - memory processes .
new york : chapman & hall .
berk , j.b . and
green , r.c , 2004 . mutual fund flows and performance in rational markets .
journal of political economy , 112 , 12691295 .
bernhardt , dan , hughson , eric and naganathan , girish , 2002 .
cream - skimming and payment for order flow . technical report .
university of illinois .
bertismas , d and lo , a. , 1998 . optimal control of execution costs . journal of financial markets , 1 , 150 .
bouchaud , j - p , farmer , j. doyne and lillo , f. , 2009 .
how markets slowly digest changes in supply and demand , in : hens , t. and schenk - hoppe , k. ( eds . ) , handbook of financial markets : dynamics and evolution .
elsevier , pp .
57156 .
bouchaud , j - p . ,
gefen , y. , potters , m. and wyart , m. , 2004 .
fluctuations and response in financial markets : the subtle nature of random " price changes . quantitative finance , 4 , 176190 .
bouchaud , j - p . ,
kockelkoren , j. and potters , m. , 2006 .
random walks , liquidity molasses and critical response in financial markets . quantitative finance , 6 , 115123 .
challet , 2007 .
the demise of constant price impact functions and single - time step models of speculation .
physica a. , 382 , 2935 .
chan , l. k.c . and lakonishok , j. , 1993 .
institutional trades and intraday stock price behavior .
journal of financial economics , 33 , 173199 .
chan , l. k.c . and lakonishok , j. , 1995 .
the behavior of stock prices around institutional trades .
the journal of finance , 50 , 11471174 .
chordia , t. and subrahmanyam , a. , 2004 .
order imbalance and individual stock returns : theory and evidence .
journal of financial markets , 72 , 485518 .
ding , z. , granger , c. w. j. and engle , r. f. , 1993 .
a long memory property of stock returns and a new model . journal of empirical finance , 1 , 83106 .
eisler , z. and kertecz , j. , 2006 .
size matters , some stylized facts of the market revisited .
european journal of physics b , 51 , 145154 .
engle , r. , ferstenberg , r. and russel , j. , 2008 .
measuring and modeling execution cost and risk .
technical report 08 - 09 .
university of chicago .
evans , m. d. d. and lyons , r. k. , 2002 .
order flow and exchange rate dynamics .
journal of political economy , 110 , 170180 .
farmer , j. d. , gillemot , l. , lillo , f. , mike , s. and sen , a. , 2004 .
what really causes large price changes ?
quantitative finance , 4 , 383397 .
farmer , j. d. , patelli , p. and zovko , ilija , 2005 .
the predictive power of zero intelligence in financial markets .
proceedings of the national academy of sciences of the united states of america , 102 , 22542259 .
farmer , j.d . , gerig , a. , lillo , f. and mike , s. , 2006 .
market efficiency and the long - memory of supply and demand : is price impact variable and permanent or fixed and temporary ?
quantitative finance , 6 , 107112 .
gabaix , x. , gopikrishnan , p. , plerou , v. and stanley , h. e. , 2003 . a theory of power - law distributions in financial market fluctuations .
nature , 423 , 267270 .
gabaix , x. , gopikrishnan , p. , plerou , v. and stanley , h.e .
institutional investors and stock market volatility . quarterly journal of economics , 121 , 461504 .
gatheral , j. , 2010 .
no - dynamic - arbitrage and market impact . quantitative finance , 10 , 749759 .
gerig , a.n . , 2007 .
a theory for market impact : how order flow affects stock price .
university of illinois .
glosten , lawrence r. , 2003 .
discriminatory limit order books , uniform price clearing and optimality .
technical report .
columbia .
glosten , l.r .
, 1994 . is the electronic limit order book inevitable ?
journal of finance , 49 , 11271161 .
gopikrishnan , p. , plerou , v. , gabaix , x. and stanley , h. e. , 2000 .
statistical properties of share volume traded in financial markets .
physical review e , 62 , r4493r4496 ; part a. hasbrouck , j. , 1991 .
measuring the information content of stock trades . the journal of finance , 46 , 179207 .
hausman , j. a. , lo , a. w. and mackinlay , a. c. , 1992 . an ordered probit analysis of transaction stock prices . journal of financial economics , 31 , 319379 .
hopman , c. , 2006 .
do supply and demand drive stock prices ?
quantitative finance ; to appear .
huberman , g. and stanzl , w. , 2004 . price manipulation and quasi - arbitrage .
econometrica , 72 , 12471275 .
keim , d. b. and madhavan , a. , 1996 .
the upstairs market for large - block transactions : analysis and measurement of price effects .
the review of financial studies , 9 , 136 .
kempf , a. and korn , o. , 1999 .
market depth and order size .
journal of financial markets , 2 , 2948 .
kyle , a. s. , 1985 .
continuous auctions and insider trading .
econometrica , 53 , 13151335 .
lillo , f. and farmer , j. d. , 2004 .
the long memory of the efficient market .
studies in nonlinear dynamics & econometrics , 8 , .
lillo , f. , farmer , j. d. and mantegna , r. n. , 2003 .
master curve for price impact function .
nature , 421 , 129130 .
lillo , f. , mike , s. and farmer , j. d. , 2005 .
theory for long memory in supply and demand .
physical review e , 7106 , 066122 .
moro , e. , moyano , l. g. , vicente , j. , gerig , a. , farmer , j. d. , vaglica , g. , lillo , f. and mantegna , r.n . , 2009 .
market impact and trading protocols of hidden orders in stock markets .
physical review e. , 80 , 066102 .
obizhaeva , a.a . and wang , j. , 2005 .
optimal trading strategy and supply / demand dynamcis . technical report .
afa 2006 boston meetings paper .
plerou , v. , gopikrishnan , p. , amaral , l. a. n. , meyer , m. and stanley , h. e. , 1999 .
scaling of the distribution of price fluctuations of individual companies .
physical review e , 60 , 65196529 ; part a. plerou , v. , gopikrishnan , p. , gabaix , x. and stanley , h. e. , 2002 . quantifying stock price response to demand fluctuations .
physical review e , 66 , article no . 027104 .
potters , m. and bouchaud , j - p . , 2003 .
more statistical properties of order books and price impact .
physica a , 324 , 133140 .
racz , e. , eisler , z. and kertesz , j. , 2009 .
comment on ` tests of scaling and universality . ' by plerou and stanley .
technical report .
schwartzkopf , y. and farmer , j.d . , 2010 .
technical report . for a preliminary report ,
see y. schwartzkopf s caltech ph.d thesis , complex phenomena in social and financial systems : from bird population growth to the dynamics of the mutual fund industry .
torre , n. , 1997 .
barra market impact model handbook .
berkeley : barra inc .
toth , b. , lemperiere , y. , deremble , c. , de lataillade , j. , kockelkoren , j. and bouchaud , j - p . , 2011 .
anomalous price impact and the critical nature of liquidity in financial markets .
phyiscal review x , 1 , 021006 ; http://arxiv.org/abs/1105.1694 .
toth , b. , lillo , f. and farmer , j.d . , 2010 .
segmentation algorithm for non - stationary compound poisson processes .
technical report .
santa fe institute .
toth , b. , palit , i. , lillo , f. and farmer , j. d. , 2011 .
why is order flow so persistent ?
technical report .
arxiv:1108.1632 .
vaglica , g. , 2008 . scaling laws of strategic behavior and specialization of strategies in agent dynamics of a financial market . thesis .
university of palermo .
vaglica , g. , lillo , f. , moro , e. and mantegna , r. , 2008 . scaling laws of strategic behavior and size heterogeneity in agent dynamics .
physical review e. , 77 , 0036110 .
viswanathan , j. and wang , james j. d. , 2002 .
market architecture : limit order books versus dealership markets .
journal of financial markets , 5 , 127167 .
weber , p. and rosenow , b. , 2006 .
large stock price changes : volume or liquidity ?
quantitative finance , 6 , 714 .
wyart , m. , bouchaud , j .-
, kockelkoren , j. , potters , m. and vettorazzo , m. , 2006 .
relation between bid - ask spread , impact and volatility in double auction markets .
technical report .
zhang , y. c. , 1999 . toward a theory of marginally efficient markets .
physica a , 269 , 3044 .
* proposition 1 . * _ the martingale condition implies zero overall profits , i.e. _ @xmath111\equiv\sum_{n=1}^\infty p_n \pi_n=0 .
e_1[n\pi_n]\equiv\sum_{n=1}^m p_n n \pi_n=0 .
\label{appeq1}\ ] ] * proof . * given the definition of @xmath220 and @xmath106 we can write the prices as @xmath221 with these expressions for @xmath222 , @xmath122 can be rewritten as @xmath223 where in the last equality we have used the martingale condition of eq .
( [ returnratio ] ) . for @xmath224
the profit per share is @xmath225 by substituting these two last expressions in eq .
( [ appeq1 ] ) we obtain @xmath226=\sum_{n=1}^{m-1}n\tilde r_n \sum_{i
= n+1}^{m } p_i-\sum_{n=1}^{m-1}p_n\sum_{i=1}^{n-1}i\tilde r_i - p_m\sum_{i=1}^{m-1 } i\tilde r_i = \nonumber \\
= \sum_{n=1}^{m-1}n\tilde r_n \sum_{j = n+1}^{m } p_j-\sum_{n=1}^{m}p_n\sum_{i=1}^{n-1}i\tilde r_i\end{aligned}\ ] ] by explicitly computing the coefficients of each @xmath227 , it is easy to show they vanish for each @xmath228 , i.e. @xmath229=0 $ ] .
* infinite support*. this proposition does not hold when @xmath6 has infinite support . in order to show this let us consider the expected profit for orders of length between @xmath189 and @xmath230 .
the following proposition holds : * proposition @xmath231 . *
_ the martingale condition for all intervals implies that for any integer @xmath232 , @xmath233 this equation holds both for finite and infinite support ( i.e. @xmath30 can be finite or infinite ) .
_ * proof .
* from the equation ( [ profpershare ] ) in the previous proposition , we know that martingale condition allows us to write the profit per lot traded as @xmath234 therefore the expected profit for orders shorter or equal to @xmath235 is @xmath236 this is equal to the quantity in eq .
( [ partsum2 ] ) .
moreover it is clear that both terms in brackets are non - negative .
this means that the market maker typically makes profits on short metaorders .
if the support of @xmath35 is infinite then the martingale condition at all intervals implies that @xmath237\equiv\sum_{n=1}^\infty p_n n\pi_n=\lim_{\bar n\to\infty } \left(\sum_{i=\bar n+1}^\infty p_i\right ) \left(\sum_{i=1}^{\bar n } i\tilde r_i\right).\ ] ] in the infinite support case the behavior of the limit in the last term of the above expression depends on the asymptotic behavior of @xmath35 and @xmath238 for large @xmath1 .
this is due to the fact that for large @xmath239 the first term in brackets goes to zero while the second term diverges .
it is possible to construct examples where @xmath229 $ ] goes to zero , to a finite value , or diverges .
this result shows that in the infinite support case the martingale condition does not imply zero overall immediate profits .
* proposition 2 . * _ if the second derivative of the immediate impact @xmath114 is bounded strictly below zero , in the limit where the number of informed traders @xmath115 , any nash equilibrium must satisfy the fair pricing condition @xmath116 for @xmath29 . on average market makers profit from orders of length one and take ( equal and opposite ) losses from orders of length @xmath30 .
_ the strategy of the proof is to show that counterexamples for which @xmath119 result in contradictions .
first consider a candidate equilibrium with @xmath240 for some value of @xmath1 , where @xmath29 .
assume a long - term trader @xmath117 buys @xmath241 shares at an average price @xmath242 .
after averaging over @xmath37 the shares are subsequently valued at price @xmath243 .
the profit is @xmath244 if long - term trader @xmath117 increases her order size by one share while all others hold their order size constant , her profit becomes @xmath245 the change in profit can be written @xmath246 the first term on the right ( @xmath247 ) represents the additional profit if it were possible to trade one extra share at the same average price , and the second term represents the reduction in profit because the average price increases . since there is nothing to distinguish the long - term traders , the equilibrium must be symmetric
, i.e. they all make the same decision .
thus if there are @xmath13 long - term traders buying @xmath241 shares and a day trader buying a random number of shares @xmath37 , at equilibrium @xmath248 . thus if @xmath13 is large , @xmath1 is also large . in the limit as @xmath13 is large the second term vanishes if @xmath249 this is true providing the second derivative of the function @xmath125 is bounded strictly below zero .
thus in this limit @xmath250 and the candidate equilibrium fails because the informed trader has an incentive to deviate . similarly if @xmath251 the informed traders take a loss which can be reduced by trading less .
when @xmath252 ( and as before @xmath29 ) no informed trader has an incentive to change her order size .
this is clear since in eq .
( [ deltaprofit ] ) with @xmath116 the change in profit is given by the second term alone , which is always negative .
a similar calculation shows that this is also true for decreasing order size , i.e. when @xmath116 , @xmath253 causes @xmath254 .
the cases @xmath21 and @xmath120 have to be examined separately because in these cases the fair pricing condition is incompatible with the martingale condition and informational efficiency ( i.e. with the conditions on the final price ) . for @xmath21
the market makers profit is @xmath255 and from ( eq .
[ shortterm ] ) the martingale condition is @xmath256 thus if @xmath257 , satisfaction of both the martingale condition and the fair pricing condition requires that @xmath258 , or equivalently that @xmath259 = 0 . in other words , if both conditions are satisfied then both the permanent and the temporary impact on the first step are identically zero , which would violate informational efficiency since @xmath260 . in section [ impactsolution ]
we show by construction that this holds for all @xmath1 , i.e. it is clear in eq .
( [ final ] ) and ( [ generalpermanentimpact ] ) that the impacts @xmath114 and @xmath68 are identically zero if @xmath261 . to have sensible impact functions we must have @xmath262 , which means that market making is profitable on the first timestep , so that they always make a profit .
this is false : the profit from a metaorder of length one is @xmath263 , where @xmath6 is a small number . in contrast , if a market maker participates only in the first trade of a large metaorder , her profit is @xmath264 , where @xmath265 is a large number . thus while @xmath266 , @xmath267 . ] .
similarly if @xmath120 the martingale condition implies @xmath268 , i.e. no reversion , and since @xmath125 is an increasing function the market maker takes a loss @xmath269 assuming @xmath116 for @xmath29 , the market makers profit @xmath270 and loss @xmath271 are related by eq .
( [ breakevenonaverage ] ) as @xmath272 for realistic size distributions we expect metaorders of size one to be much more common than those of size @xmath30 , i.e. @xmath273 . the ratio of the total profits is @xmath274 thus the market maker receives frequent but small profits on metaorders of length one and rare but large losses for metaorders of length @xmath30 .
long - term traders will rationally abstain from taking a loss on metaorders of length @xmath21 by simply not participating when they receive @xmath6 signals that are too weak ; thus , the trading volume at @xmath21 is due entirely to the day trader .
similarly , although @xmath275 , the long - term traders are unable to improve their profits by trading more , since we have bounded the total amount an individual can trade at @xmath32 so they are blocked from further increase .
thus the violations of the fair pricing condition when @xmath21 and @xmath120 occur naturally due to the institutional constraints that we have assumed and do not invalidate the equilibrium .
* proposition 3 . * _ the system of martingale conditions ( eq . [ shortterm ] ) and fair pricing conditions ( eq . [ fairpricing ] ) has solution @xmath276 _ * proof . * the solution of eq .
( [ solution2a ] ) is a direct consequence of the martingale conditions ( eq . [ shortterm ] ) .
the total profit of metaorders of length @xmath222 can be rewritten as ( see proof of proposition 1 ) @xmath277 the fair pricing conditions ( eq . [ fairpricing ] ) state that for @xmath278 it is @xmath279 , i.e. @xmath280 this is a recursive equation which determines @xmath238 once @xmath134 is given ( note that this equation does not hold for @xmath189 because we do not have fair pricing for metaorders of length one ) .
the solution of this equation is @xmath281 and we prove it by induction .
we assume that the solution holds for @xmath282 and we prove that it is true for @xmath283 .
( [ solreceqa ] ) holds for @xmath284 we can rewrite eq . ( [ receqa ] ) for @xmath283 as @xmath285 now by expanding the sum in brackets it is direct to show that @xmath286 since , by definition , @xmath287 the first two terms in the right hand side cancel and thus one obtains eq .
( [ solreceqa ] ) .
this equation is equivalent to eq .
( [ solutiona ] ) .
in fact @xmath288 i.e. our thesis , eq .
( [ solutiona ] ) .
as already discussed briefly in section [ finitesize ] , if the condition @xmath195 is violated this has an effect on the impact . in this section we consider the exact case of a finite support pareto distribution .
we show that when @xmath195 we obtain the same results of the previous section and we discuss what happens when @xmath197 .
we assume that the metaorder size distribution is a truncated pareto distribution for all @xmath289 , i.e. @xmath290 where the normalization constant @xmath291 is the harmonic number of order @xmath292 .
for the truncated pareto distribution the probability @xmath74 that a metaorder of size @xmath38 will continue is @xmath293 where @xmath177 is the generalized riemann zeta function ( also called the hurwitz zeta function ) . for small @xmath38
the function @xmath74 increases meaning that it is more and more likely that the order continues . in the regime of @xmath294 , @xmath74 is well approximated by the expression of eq .
( [ calpapprox ] ) for an infinite support pareto distribution .
however , around @xmath295 , @xmath74 starts to decrease meaning that it becomes more and more likely that the order is going to stop soon , with a corresponding effect on the impact .
the immediate impact can be easily calculated once the distribution of metaorder size is known by using eq .
( [ final ] ) . for truncated pareto distributed metaorder sizes ,
@xmath107 is ( for @xmath296 ) is @xmath297 for large @xmath38 but @xmath298 it is @xmath299 which is the same scaling as the infinite support pareto distribution ( see eq .
( [ temporaryimpact ] ) ) .
the same holds true for the permanent impact .
we have therefore shown that when @xmath298 the finite support of the metaorder size distribution is irrelevant and we obtain approximately the same results as in section [ paretosec ] .
the finite size effects and the role of the finiteness of the support becomes relevant when @xmath300 .
figure [ figimpact ] shows the total impact for @xmath301 and different values of @xmath181 .
it is clear that the impact is initially described by a power law , but then it becomes strongly convex when the order length becomes comparable with the maximal length . | we develop a theory for the market impact of large trading orders , which we call _ metaorders _ because they are typically split into small pieces and executed incrementally .
market impact is empirically observed to be a concave function of metaorder size , i.e. the impact per share of large metaorders is smaller than that of small metaorders .
we formulate a stylized model of an algorithmic execution service and derive a fair pricing condition , which says that the average transaction price of the metaorder is equal to the price after trading is completed .
we show that at equilibrium the distribution of trading volume adjusts to reflect information , and dictates the shape of the impact function .
the resulting theory makes empirically testable predictions for the functional form of both the temporary and permanent components of market impact .
based on the commonly observed asymptotic distribution for the volume of large trades , it says that market impact should increase asymptotically roughly as the square root of size , with average permanent impact relaxing to about two thirds of peak impact . |
# 1#2 # 1#2#3#4#5 w # 1 ( cc#5 & # 4 + # 2 & # 3 ) # 1#2#3#4#5 # 1 ( cc#5 & # 4 + # 2 & # 3 ) # 1#2#3 # 1#2#3 in the last decade many infinite hierarchies of exactly solvable models have been found . of foremost importance among these models
are the restricted solid - on - solid ( rsos ) models of andrews , baxter and forrester ( abf ) @xcite . in these models
each site of the lattice carries a height variable , restricted to the values @xmath3 with @xmath4 , subject to the rule that heights on neighbouring lattice sites differ by @xmath5 .
if the allowed heights are represented by the nodes in the following diagram , @xmath6 the adjacency rule requires that neighbouring lattice sites take values that are adjacent on the diagram .
andrews , baxter and forrester considered four different regimes , labelled i iv .
it was pointed out by huse @xcite that the critical line separating regimes iii and iv realises the complete unitary minimal series of conformal field theory .
this series is described by a central charge @xmath7 and a set of conformal weights , given by the kac formula @xmath8 ^ 2 - 1}{4h(h-1 ) } \qquad 1\leq r\leq h-2,\quad 1\leq s \leq h-1 .
\label{eq : a.kactable}\ ] ] the corresponding modular invariant partition function is @xcite @xmath9 where @xmath10 is the modular parameter and the @xmath11 are the virasoro characters given by @xmath12j } -q^{h(h-1)j^2+[hr+(h-1)s]j+rs}\right\ } \label{eq : a.virasoro}\ ] ] with @xmath13 . by giving a loop or polygon interpretation to the critical abf models , pasquier @xcite extended these models to arbitrary adjacency graphs . demanding that these new models be critical
restricts the graphs to the dynkin diagrams of the classical and affine simply - laced lie algebras shown in fig .
[ fig : a.adediagrams ] .
recently a new construction of solvable rsos models was found @xcite .
basically , the method is an extension of the work of pasquier , and related work of owczarek and baxter @xcite , to more general loop models .
application to the o@xmath14 model @xcite , which is closely related to the izergin - korepin model @xcite , has led to a new family of critical rsos models labelled by dynkin diagrams .
the same models were found independently by roche @xcite . in the approach of pasquier ,
the polygons , which are interpreted as domain walls separating regions of different height , densely cover the edges of the dual lattice . as a consequence ,
heights on adjacent sites are always different . in the new rsos models , two neighbouring sites of the lattice
either have the same or different height , so that the domain walls occupy some but not all edges of the dual lattice .
therefore it is natural , following @xcite , to term these new models _ dilute _ a - d - e models .
each member of the dilute a@xmath15 hierarchy possesses four distinct critical branches .
the central charge is given by @xmath16 where @xmath17 the first two branches give new realisations of the unitary minimal series with the modular invariant partition functions ( [ diagmipf ] ) .
the other two branches appear to be a direct product of this same series and an ising model , with modular invariant partition functions latexmath:[\[z={1\over4}\sum_{r'=1}^{2}\sum_{s'=1}^{3 } \sum_{r=1}^{h-2}\sum_{s=1}^{h-1 } as reported in @xcite , the models related to the a@xmath15 dynkin diagrams admit an off - critical extension .
a remarkable feature of these off - critical models is that , for odd values of @xmath19 , they break the @xmath0 symmetry of the underlying dynkin diagram .
the simplest of these symmetry breaking models belongs to the universality class of the ising model .
this allows the calculation of the magnetic exponent @xmath2 without the use of scaling relations .
this paper is devoted to the investigation of the models of the dilute a@xmath15 hierarchy .
first we briefly describe the whole family of dilute a - d - e models .
then , in section [ sec : a.offc ] , we define the off - critical a@xmath15 model and in section [ sec : a.free ] we calculate its free energy . from this
we extract the critical exponent @xmath20 when @xmath19 is even and @xmath21 when @xmath19 is odd .
the main body of the paper is concerned with the calculation of the order parameters of the dilute a models for odd values of @xmath19 . in section [ sec : a.lhp ] we compute the local height probabilities and in the subsequent section we use these results to evaluate generalised order parameters .
we also extract the set of associated critical exponents @xmath22 and derive the corresponding conformal weights . in section [ sec : a.phdiag ] we discuss the phase diagram , concentrating on @xmath23 , and in section [ sec : isfi ] we collect results concerning the ising model in a field .
finally , we summarise and discuss our main results . the results for the order parameters when @xmath19 is even will be presented in a future publication .
likewise , results for the critical models related to the other adjacency diagrams , among which is a solvable tricritical potts model @xcite , will be reported elsewhere . in this section
we define the family of dilute a - d - e models .
although we restrict the description to the square lattice , they can be defined on any planar lattice .
consider an arbitrary connected graph @xmath24 consisting of @xmath19 nodes and a number of bonds connecting distinct nodes .
label the nodes by an integer _
nodes @xmath26 and @xmath27 are called adjacent on @xmath24 if they are connected via a single bond . such a graph
is conveniently represented by an _ adjacency matrix _
@xmath28 with elements @xmath29 let @xmath30 denote the largest eigenvalue of @xmath28 and @xmath31 the perron - frobenius vector , _
i.e. , _ @xmath32 . with these ingredients we define an rsos model on the square lattice
@xmath33 as follows
. each site of @xmath33 can take one of @xmath19 different heights .
the boltzmann weight of a configuration is non - zero only if all pairs of neighbouring sites carry heights which are either equal or adjacent on @xmath24 .
the weight of an elementary face of the rsos model is given by @xmath34 where @xmath35 is the _
a_-th entry of @xmath31 and @xmath36 and @xmath37 can take any of the @xmath19 heights of the graph @xmath24 .
the generalised kronecker @xmath21 is defined as @xmath38 .
if we parametrise @xmath30 by @xmath39 then @xmath40 are given by of reference @xcite to @xmath41 . ]
@xmath42 we note that the weights @xmath43 and the weights @xmath44 are determined only up to a sign . for any graph @xmath24 ,
the rsos model defined by ( [ eq : da.rsos])-([eq : a.rhos ] ) satisfies the yang - baxter equation @xcite .
in fact , in @xcite it was shown that all models defined above have the same partition function as the o@xmath14 model @xcite . from equations ( [ eq : a.parn ] ) and ( [ eq : a.rhos ] ) it follows that there are four different branches that yield the same values of @xmath30 . using the periodicity of the weights
, we can restrict @xmath45 to the interval @xmath46 : @xmath47 these four branches correspond to ( part of ) the four branches defined in @xcite for the o@xmath14 model .
for @xmath48 the value of @xmath49 must be chosen imaginary , and the weights become complex , unlike the ordinary a - d - e models . as was pointed out in @xcite , the only adjacency graphs that have largest eigenvalue @xmath50 are the dynkin diagrams of the classical @xmath51 and affine @xmath52 simply - laced lie algebras shown in fig .
[ fig : a.adediagrams ] . for the classical case
the respective values of @xmath30 are listed in table [ tab : a.thetable ] .
the corresponding perron - frobenius vector can be found in @xcite . from the equivalence with the o@xmath14 model , the central charge of the
dilute a - d - e models is known @xcite .
the values on the four branches are listed in table [ tab : a.thetable ] for the classical algebras .
for the affine algebras we have @xmath53 and @xmath54 or @xmath55 .
@xmath56 [ sec : a.offc ] the dilute a@xmath15 model ( [ eq : da.rsos ] ) admits an extension away from criticality while remaining solvable . in terms of the theta functions of appendix [ app : a.elliptic ] , suppressing the dependence on the nome @xmath57 , the boltzmann weights of the off - critical a@xmath15 model are given by @xcite @xmath58 we note that in the critical limit , @xmath59 , the _ crossing factors _
@xmath60 reduce to the entries of the perron - frobenius vector @xmath35 : @xmath61 we also note that , for later convenience , we have relabelled the states of the model @xmath62 in comparison with those of our earlier definition of the model in @xcite .
the boltzmann weights ( [ eq : a.weights ] ) satisfy the following initial condition and crossing symmetry : @xmath63 and an inversion relation of the form _ g w ( .
cc d & g + a & b cc d & c +
g & b in equation ( [ eq : a.branch ] ) four different critical branches were defined .
this yields eight regimes for the off - critical a@xmath15 model @xmath64 for regimes @xmath65 and @xmath66 we exclude the @xmath67 case because the model becomes singular . at criticality all a@xmath15 models
satisfy the @xmath0 symmetry of the dynkin diagram , but the off - critical models , for odd values of @xmath19 , break this symmetry : @xmath68 for @xmath23 , if we make the identification @xmath69 , the model can be viewed as a spin-1 ising model . for @xmath70
the up - down symmetry of this model is broken .
we can therefore regard the nome @xmath57 of the @xmath71-functions as a magnetic field .
this is in contrast with the usual role of @xmath57 as a temperature - like variable ( see , _
@xcite ) . also for larger , odd values of @xmath19 we refer to @xmath57 as the magnetic field , even though it is not , in general , the leading magnetic operator . for odd @xmath19
the @xmath72 and @xmath73 regimes are equivalent because negating the magnetic field @xmath57 merely has the effect of relabelling the heights @xmath74 .
the dilute a@xmath15 model defined above is closely related to the izergin - korepin ( or a@xmath75 ) sos model @xcite . in the latter model
the weights are given by equation ( [ eq : a.weights ] ) with @xmath76 replaced by @xmath77 , where @xmath78 is an arbitrary constant , and @xmath79 .
if we set @xmath80 we obtain a@xmath75 rsos models based on the a@xmath15 algebra . only when @xmath81 or @xmath82 are these models physical . other choices of @xmath83 correspond to eigenvalues of the adjacency matrix of a@xmath15 which are not the largest . for all odd values of @xmath83
the models break the @xmath0 symmetry of the underlying dynkin diagram . another way to restrict the a@xmath75 sos model
is given by @xmath84 this possibility has been studied by kuniba in a more general way in his study of a@xmath85 rsos models @xcite .
however , this way of restricting the a@xmath75 sos model does not give rise to symmetry breaking models .
[ sec : a.free ] we calculate the free energy or , equivalently , the partition function per site @xmath86 of the dilute a model by the inversion relation method @xcite .
because the inversion relation ( [ eq : a.invrel ] ) is quadratic in @xmath57 we can restrict ourselves to the regimes with positive @xmath57 , and parametrise @xmath87 . in regimes @xmath89 and @xmath90 we assume that @xmath91 is analytic in the strip @xmath92 , and may be analytically extended just beyond these boundaries . the inversion relation ( [ eq : a.invrel ] ) implies @xmath93 while the crossing symmetry ( [ eq : a.crossing ] ) translates to @xmath94 we make the conjugate modulus transformation ( [ eq : a.congnome ] ) and a laurent expansion of @xmath95 in powers of @xmath96 . matching coefficients in equations ( [ eq : a.kinv ] ) and ( [ eq : a.cross ] ) we obtain for the free energy @xmath97 we now take the @xmath98 limit in the above expression to obtain the leading critical singularity . using the poisson summation formula , we find @xmath99 if we compare this with @xmath100 we find for the critical exponents @xmath20 and @xmath21 @xmath101 when @xmath67 in regime @xmath89 , expression ( [ eq : a.kappasing12 ] ) for the critical singularity has to be multiplied by @xmath102 . in regimes @xmath103 and @xmath104 the appropriate analyticity strip is @xmath105 , and the crossing symmetry becomes @xmath106 performing the same steps as before we obtain \(u ) = - 2 _ k=-^ . for the dominant singularity we find , apart from some exceptions we list below , @xmath107 comparing this with ( [ eq : a.defad ] ) , we find for the critical exponents @xmath20 and @xmath21 @xmath108 for the cases @xmath109 and 8 in regime @xmath103 and @xmath67 in regime @xmath104 equation ( [ eq : a.kappasing34 ] ) has to be multiplied by @xmath102 . when @xmath110 and 14 in regime @xmath103 and @xmath111 in regime @xmath104 the partition function per site is regular .
[ sec : a.lhp ] in this section , which forms the main part of our paper , we calculate the _ local height probabilities _ of the dilute a model for odd values of @xmath19 .
since negating the nome @xmath57 is nothing but a reversal of the magnetic field , we can restrict ourselves to the four ` + ' regimes .
( recall that for odd @xmath19 the boltzmann weights are symmetric under the transformation @xmath112 , @xmath62 . )
first we describe the groundstate configurations in each of the four regimes .
below we depict the set of groundstates by decoration of the adjacency diagram . comparing the various weights in ( [ eq : a.weights ] ) in the ordered limit , see appendix [ app : a.h - fun ] , we find that the following types of groundstates occur : 1
. completely flat or ferromagnetic configurations .
if the height of this configuration is @xmath26 we will denote this state by a solid circle on the adjacency graph @xmath24 at node @xmath26 .
conversely , flat configurations that do not yield a groundstate will be denoted by an open circle on @xmath24 .
antiferromagnetic configurations .
one sublattice has height @xmath26 and the other sublattice height @xmath113 .
this state , together with the state where the heights on the two sublattices are interchanged , is denoted by a double bond on @xmath24 between the nodes @xmath26 and @xmath27 . if we define the variable @xmath114 to be @xmath115 where @xmath116 @xmath117 denotes the integer part , the groundstates of the four regimes are as shown in fig .
[ fig : a.groundstate ] .
thus the total number of groundstates is @xmath118 and @xmath119 for the regimes @xmath120 , respectively .
the local height probability @xmath121 is the probability that a given site of the lattice has height @xmath26 given that the model is in the phase indexed by @xmath27 and @xmath122 . in the ferromagnetic phases @xmath123 and in the antiferromagnetic phases @xmath124 .
the technique of corner transfer matrices ( ctms ) which is used to calculate @xmath121 is well - known and the details of the method are given elsewhere ( see , _
@xcite ) . because the weights ( [ eq : a.weights ] ) satisfy the yang - baxter equation and possess the crossing symmetry ( [ eq : a.crossing ] )
, the local height probability in regimes @xmath89 and @xmath90 can be written as @xmath125 here the variable @xmath10 is related to @xmath87 by @xmath126 the one - dimensional configuration sums @xmath127 are given by @xmath128 the weight function @xmath129 is determined from the boltzmann weights in the ordered limit ( @xmath130 fixed ) .
their behaviour is @xmath131 where @xmath132 the values of the function @xmath133 for regimes @xmath89 and @xmath90 are listed in appendix [ app : a.h - fun ] . from ( [ eq : odc ] ) it follows directly that the one - dimensional configuration sums satisfy the following recurrence relation : @xmath134
the task is to solve this relation , with initial condition @xmath135 or , equivalently , @xmath136 and with the conditions @xmath137 which confine the heights in the recursion relation ( [ eq : a.recurrence ] ) to the set @xmath138 . in regimes @xmath103 and @xmath104 , where the spectral parameter @xmath139 is negative , the local height probability is given by @xmath140 where @xmath10 is now @xmath141 as for the regimes 1@xmath88 and 2@xmath88 , the one - dimensional configuration sums are expressed in @xmath129 by ( [ eq : odc ] ) . to determine @xmath129
, we now need to know the powers of @xmath142 in the ordered limit of the weights .
relative to regimes 1@xmath88 and 2@xmath88 , after the appropriate gauge factors @xmath143 are separated , and contributions which cancel in the ratio in equation ( [ eq : a.lhp34 ] ) are discarded , @xmath129 is effectively replaced @xmath144 .
we thus need only take @xmath145 in the regime @xmath89 ( @xmath90 ) solution of ( [ eq : a.recurrence ] ) to obtain the form of the solution in regime @xmath104 ( @xmath103 ) .
of course , @xmath10 must still be given its proper meaning from ( [ eq : a.qdef34 ] ) .
in this section we derive the solution of the recursion relation ( [ eq : a.recurrence ] ) for each of the four regimes .
in fact , as remarked above , the solution for regimes @xmath103 and @xmath104 is deduced from the solution for regimes @xmath89 and @xmath90 .
before we present the solution we need some preliminaries on the _ gaussian multinomials _
@xmath146 , defined by @xcite @xmath147 where @xmath148 in the limit @xmath149 the gaussian multinomials reduce to ordinary multinomials @xmath150 the following identities , which prove useful in solving the recurrence relation , may be derived by straightforward manipulation of the definition @xmath151 as a first step towards the solution of equation ( [ eq : a.recurrence ] ) we consider the special limit @xmath152 .
+ in this limit , setting @xmath153 , the recurrence relation reduces to the following combinatorial problem @xmath154 the fundamental solution of this equation is @xmath155 in order to satisfy the initial and confining conditions ( [ eq : a.initial ] ) and ( [ eq : a.confine ] ) we must take linear combinations of this solution in the following way : @xmath156 another piece of information can be gained by considering the @xmath157 limit .
it was noted in @xcite that , in this limit , the one - dimensional configuration sums for the abf models in regimes iii and iv become precisely the characters of the related virasoro algebra ( [ eq : a.virasoro ] ) .
expecting similar @xmath158 behaviour for our configuration sums , we generated large polynomials on the computer and multiplied these by @xmath159 .
the resulting polynomials , being very sparse , were then easily identified as virasoro characters , with @xmath160 and @xmath161 for @xmath162 and @xmath163 for @xmath164 , or as linear combinations of two characters . guided by these two limiting cases , replacing @xmath165 in ( [ eq : a.virasoro ] ) by @xmath82 according to ( [ eq : a.ltoh ] ) , we make the following ansatz for the configuration sums , with @xmath162 , which yield a single character : @xmath166j+\alpha(j , k ) } \mult{m}{k}{k \p 2(l \p 1)j \p a \m b } \right . }
\nonumber \\ & & \nonumber \\ \lefteqn { \hphantom{x_m^{a\ , b\ , c } = q^{\gamma \ , m } \ ! \sum_{j , k=-\infty}^{\infty } } \hskip -\wd0 \hskip -\wd1 \left .
-q^{(l+2)(l+1)j^2+[(l+2)a+(l+1)b]j+a b+\beta(j , k ) } \mult{m}{k}{k \p 2(l \p 1)j \p a \p b } \right\}. } \setbox0=\hbox { } \setbox1=\hbox{}\end{aligned}\ ] ] for @xmath164 we make a similar ansatz . the unknown functions @xmath20 and @xmath167 are assumed to be quadratic in their arguments @xmath168 and @xmath83 .
they depend implicitly on @xmath26 , @xmath27 and @xmath122 , as does the parameter @xmath169 .
for the configuration sums that , in the @xmath170 limit , yield linear combinations of characters , we make the appropriate linear combinations of the above ansatz .
fitting the ansatz for small values of @xmath171 with the correct polynomial , the coefficients in the functions @xmath172 @xmath173 and @xmath169 can be determined .
before we present the solution , we define the following auxiliary function : @xmath174j+k[k+2(l+1)j+a - b ] } \mult{m}{k}{k \p 2(l \p 1)j \p a \m b } \right . }
\label{eq : a.f } \\ & & \nonumber \\
\lefteqn{\hskip -\wd0 \left .
\qquad \quad -q^{(l+2)(l+1)j^2+[(l+2)a+(l+1)s]j+as+k[k+2(l+1)j+a+b ] } \mult{m}{k}{k \p 2(l \p 1)j \p a \p b } \right\ } } \nonumber \setbox0=\hbox{}\end{aligned}\ ] ] where we have suppressed the @xmath26 dependence .
this function has the following elementary properties @xmath175 it is also convenient to define the four sets @xmath176 where @xmath114 is given by ( [ eq : a.defl ] ) . with these definitions the solution of the recurrence relation reads rll x_m^abb-1=q^mh(b ,
b , b-1 ) & \ { l f_m^b(b ) + q^b f_m^b+1(b ) + q^b f_m^b+1(b ) + f_m^b(b ) + ( 1 q^m ) q^b f_m 1^b+2(b 1 ) . &
l b s_1\{1 } + b s_2 \{l 1 } + b s_3 + b s_4 + & & + x_m^abb= q^mh(b , b , b ) & \ { l f_m^b(b ) + q^b f_m^b+1(b ) + q^b f_m^b+1(b ) + ( 1 q^m ) q^-b f_m 1^b-1(b 1 ) + f_m^b(b ) + ( 1 q^m ) q^b
f_m 1^b+2(b 1 ) . &
l b s_1 + b s_2 + b s_3 + b s_4 + & & + x_m^abb+1= q^mh(b , b , b+1 ) & \ { l f_m^b(b ) + q^b f_m^b+1(b ) + q^b f_m^b+1(b ) + ( 1 q^m ) q^-b
f_m 1^b-1(b 1 ) + f_m^b(b ) . &
l b s_1 + b s_2 + b s_3\{0 } + b s_4\{l}. [ eq : a.solnr1 ] we note that in this solution we have included the terms @xmath177 and @xmath178 which , according to the confining condition ( [ eq : a.confine ] ) , should be identically zero . using the simple relations ( [ eq : a.elemprop ] ) it follows directly that this is indeed the case .
the advantage of including these terms in the above way is that it enables us to prove the solution without treating the boundary separately . from the definition ( [ eq : a.f ] ) it also follows immediately that the initial condition ( [ eq : a.initial ] ) is satisfied .
proving that ( [ eq : a.solnr1 ] ) is indeed the solution of the recurrence relation is now straightforward but rather tedious . inserting ( [ eq : a.solnr1 ] ) into the recursion relation , using the explicit form of the function @xmath129 as listed in equation ( [ eq : a.hfuncr1 ] ) , we find that only the following four relations need hold : ll f_m^b(b)-f_m-1^b(b ) & + = q^m-1 & b s_1 + f_m^b+1(b)-f_m-1^b+1(b ) & + = q^m-1 & b s_2 + f_m^b+1(b)-f_m-1^b+1(b 1 ) & + = q^m-1 & b s_3 + f_m^b(b)-f_m-1^b(b 1 ) & + = q^m-1 & b s_4 .
the restrictions on @xmath27 make these special cases of more general expressions , which we will show to be true for all values of @xmath27 .
if we widen their applicability in this way , the relations can in fact be combined to give simpler ones . in turn , these relations can be further simplified by requiring that they hold term - by - term in @xmath168 .
the resulting pair of relations are much stronger requirements than the original set , but still they hold . defining the function @xmath179 where we have set @xmath180 , these two relations are f_m^b(b)-f_m^b(b 1)=q^m [ eq : a.two1 ] f_m^b(b)-f_m-1^b(b)=q^m-1 .
[ eq : a.two2 ] the function @xmath181 consists of two polynomials .
one , the first term within the curly braces , has only positive coefficients and the other , the second term , has only negative coefficients . if we demand that the equations ( [ eq : a.two1 ] ) and ( [ eq : a.two2 ] ) be satisfied for the ` positive ' and ` negative ' polynomials independently , and we set @xmath182 , respectively , this yields @xmath183 for equation ( [ eq : a.two1 ] ) , and @xmath184 for equation ( [ eq : a.two2 ] ) .
the proof of these final two equations is now elementary .
equation ( [ eq : a.first ] ) follows immediately from ( [ eq : a.gidena ] ) and ( [ eq : a.gidenb ] ) if we set @xmath185 and @xmath186 , respectively .
equation ( [ eq : a.second ] ) follows from ( [ eq : a.gidena ] ) with @xmath185 . finding and proving the solution of the recurrence relation for regime @xmath90 proceeds along similar lines as in regime @xmath89 . to remove any confusion
we will denote the one - dimensional configuration sums in regime @xmath90 by @xmath187 .
first of all , the @xmath152 limit ( [ eq : a.qis1 ] ) is still valid .
generating large polynomials reveals that in this case the configuration sums yield virasoro characters with @xmath188 and @xmath189 for @xmath190 , and @xmath191 for @xmath192 , or linear combinations of two characters .
we replace @xmath165 in ( [ eq : a.virasoro ] ) by @xmath193 according to ( [ eq : a.ltoh ] ) and define the auxiliary function @xmath194j+k[k+2(l+1 ) j+a - b ] } \mult{m}{k}{k \p 2 ( l \p 1 ) j \p a \m b } \right . }
\label{eq : a.g } \\ & & \nonumber \\ \lefteqn{\quad \qquad \left .
\hskip -\wd0 -q^{(l+1)lj^2+[(l+1)r+la]j+ar+k[k+2(l+1)j+a+b ] } \mult{m}{k}{k \p 2 ( l \p 1 ) j \p a \p b } \right\ } } \setbox0=\hbox { } \nonumber\end{aligned}\ ] ] which satisfies the following simple relations @xmath195 again we define four sets @xmath196 the solution of the recurrence relation then reads rll y_m^abb-1=q^mh(b , b , b-1 ) & \ { l g_m^b(b ) + q^-b g_m^b-1(b ) + q^-b g_m^b-1(b ) + ( 1 q^m )
q^b g_m-1^b+1(b 1 ) + -(1 q^m ) q^-1 g_m-1 ^ 1(1 ) _ b,2 + g_m^b(b ) . &
l b t_1\{1 } + b t_2 \{l 1 } + + b t_3 + b t_4 + & & + y_m^abb = q^mh(b , b , b ) & \ { l g_m^b(b ) + q^-b g_m^b-1(b ) + q^-b g_m^b-1(b ) + ( 1 q^m ) q^b g_m-1^b+1(b 1 ) + g_m^b(b ) + ( 1 q^m ) q^-b g_m-1^b-2(b 1 ) . &
l b t_1 + b t_2 + b t_3 + b t_4 + & & + y_m^abb+1=q^mh(b , b , b+1 ) & \ { l g_m^b(b ) + q^-b g_m^b-1(b ) + q^-b g_m^b-1(b ) + g_m^b(b ) + ( 1 q^m ) q^-b g_m-1^b-2(b 1 ) . &
l b t_1 + b t_2 + b t_3 + b t_4\{l}. [ eq : a.solnr2 ] as in regime @xmath89 we have extended the solution to include the term @xmath197 , which , using ( [ eq : a.elemprop2 ] ) , indeed yields zero . however , in contrast to regime @xmath89 , we can no longer extend the above solution to the term @xmath198 which according to ( [ eq : a.solnr2 ] ) would not be zero .
therefore , in proving the recurrence relation we have to treat cases that involve this boundary term separately .
this ` irregularity ' is however compensated by the exceptional terms @xmath199 of ( [ eq : a.solnr2 ] ) and @xmath200 of ( [ eq : a.hfuncr2 ] ) . as a result , inserting the above solution into the recurrence relation , using ( [ eq : a.elemprop2 ] ) , we again obtain only four relations that should hold : ll g_m^b(b)-g_m-1^b(b ) & + = q^m-1 & b t_1 + g_m^b-1(b)-g_m-1^b-1(b ) & + = q^m-1 & b t_2 + g_m^b-1(b)-g_m-1^b-1(b 1 ) & + = q^m-1 & b t_3 + g_m^b(b)-g_m-1^b(b 1 ) & + = q^m-1 & b t_4 .
again these equations hold for all values of @xmath27 and we drop the restrictions on @xmath27 . doing so , the four equations can be combined to give simpler equations .
as before , these are true term - by - term in @xmath168 . setting @xmath201 , we find f_m^b(b)-f_m^b(b 1)=q^m f_m^b(b)-f_m-1^b(b)=q^m-1 with @xmath181 defined in ( [ eq : a.smallf ] ) .
finally , making the same splitting into ` positive ' and ` negative ' polynomials as before , now setting @xmath202 , yields precisely the equations ( [ eq : a.first ] ) and ( [ eq : a.second ] ) .
as explained previously , the form of the solution of the recurrence relation for regimes @xmath103 and @xmath104 can simply be obtained by replacing @xmath10 with @xmath203 in ( [ eq : a.solnr2 ] ) and ( [ eq : a.solnr1 ] ) , respectively .
it should again be stressed that the precise meaning of the variable @xmath10 in the various regimes is not related , and is given by equation ( [ eq : a.qdef ] ) in regimes @xmath89 and @xmath90 , and by ( [ eq : a.qdef34])in regimes 3@xmath88 and 4@xmath88 . in this section , using the solutions for the one - dimensional configuration sums @xmath204 and @xmath205 , we obtain expressions for the local height probabilities @xmath121 .
as described in section [ sec : a.groundstate ] , in regime 1@xmath88 we have only ferromagnetic groundstate configurations . hence we need to calculate p^bb(a ) = _ m b=1,3, ,l , l+1,l+3, ,l-1 .
[ eq : a.pbbreg1 ] from the solution ( [ eq : a.solnr1 ] ) for the one - dimensional configuration sums , we find that we have to consider the @xmath206 limit of the auxiliary function @xmath207 defined in ( [ eq : a.f ] ) .
to do so , we need the result @xmath208 which holds for arbitrary fixed @xmath26 . to establish this
we take the limit inside the sum and use the elementary result @xmath209 to obtain @xmath210 here the last equality follows from the @xmath10-analogue of kummer s theorem @xcite @xmath211 by setting @xmath212 . from the above considerations we conclude that , in the @xmath206 limit , the auxiliary function @xmath207 yields the virasoro characters defined in equation ( [ eq : a.virasoro ] ) @xmath213j } \m q^{(l+2)(l+1)j^2+[(l+2)a+(l+1)s]j+as } \right\ } } \nonumber \\ \lefteqn { \hphantom{\lim_{m\to\infty } f_m^s(b ) } \vphantom{\sum^a_a } = q^{\half(a - s)a-\delta_{a , s}^{(l+2)}+c/24}\:\chi_{a , s}^{(l+2)}(q ) } \\
\lefteqn { \hphantom{\lim_{m\to\infty } f_m^s(b ) } \sim q^{a^2\la/\pi } \:\chi_{a , s}^{(l+2)}(q ) . }
\nonumber\end{aligned}\ ] ] in the last step we have used @xmath214 , and have omitted terms independent of @xmath26 . substituting the appropriate elements of the solution ( [ eq : a.solnr1 ] ) , the @xmath215 , into the expression for local height probabilities , and using the limiting behaviour of @xmath207 , we find @xmath216 where @xmath217 we note that @xmath218 takes the values @xmath219 . by performing the conjugate modulus transformation ( [ eq : a.congnome ] ) we rewrite the virasoro characters as @xmath220\ ] ] where @xmath221 as a result the local height probabilities can be written in the form @xmath222 \label{eq : a.lhpreg1}\ ] ] where the normalisation factor in the denominator is given by @xmath223 \nonumber\\ & = & \sum_{a=0}^{2l+1 } s(a ) \ , \td\left(\case{\pi}{2 } \left(\case{a}{l+1 }
\m \case{s}{l+2}\right),t \right ) .
\label{eq : a.n1}\end{aligned}\ ] ] we recall that the nome @xmath10 is defined in terms of the nome @xmath57 as @xmath224 .
this yields the following relation between the nomes @xmath57 and @xmath225 : @xmath226 from the definition of the crossing factor @xmath227 and the simple identity @xmath228 we find an alternative form for @xmath31 : @xmath229 where we have again neglected @xmath26-independent factors . substituting this back into the definition of the normalisation factor
, we find that the summation in @xmath230 can actually be performed to give = .
[ eq : a.n1den ] the proof of this _ denominator identity _ , being rather technical and lengthy , is given in appendix [ app : a.den1 ] .
the working for regime 2@xmath88 is very similar to that of regime 1@xmath88 .
again we have only ferromagnetic groundstates , and we are interested in calculating p^bb(a ) = _ m b=1,3, ,l-1,l+2,l+4, ,l-1 .
[ eq : a.pbbreg2 ] from the solution ( [ eq : a.solnr2 ] ) for the one - dimensional configuration sums we see that we have to take the limit @xmath206 of the auxiliary function @xmath231 defined in ( [ eq : a.g ] ) . using the result ( [ eq : a.limgpoly ] ) yields @xmath232j } \m q^{(l+1)lj^2+[(l+1)r+la]j+ar } \right\ } } \nonumber \\ \lefteqn { \hphantom{\lim_{m\to\infty } g_m^s(b ) } \vphantom{\sum^a_a } = q^{\half(a - r)a-\delta_{r , a}^{(l+1)}+c/24 } \ : \chi_{r , a}^{(l+1)}(q ) } \\
\lefteqn { \hphantom{\lim_{m\to\infty } g_m^s(b ) } \sim q^{a^2\la/\pi } \:\chi_{r , a}^{(l+1)}(q ) } \nonumber\end{aligned}\ ] ] where we have used that @xmath233 . substituting the elements @xmath234 of the solution ( [ eq : a.solnr2 ] ) , and using the limiting behaviour of @xmath231
, we find that the local height probabilities are given by @xmath235 where @xmath236 we note that @xmath237 takes the values @xmath238 .
if we use the conjugate modulus form of the virasoro characters , we can rewrite this as @xmath239 \label{eq : a.lhpreg2}\ ] ] where the normalisation factor in the denominator is @xmath240 from the relation between the nomes @xmath57 and @xmath10 we find that @xmath57 and @xmath225 are again related as in equation ( [ eq : a.pnomet ] ) . inserting the form ( [ eq : a.alts ] ) for the crossing factor @xmath31 into the expression for the normalisation factor @xmath241
, we can carry out the summation over @xmath26 .
the result , proved in appendix [ app : a.den2 ] , is = .
[ eq : a.n2den ] the solution of the recurrence relation for regime 3@xmath88 is obtained by replacing @xmath10 by @xmath203 in the solution ( [ eq : a.solnr2 ] ) for regime 2@xmath88 .
we therefore need to consider the auxiliary function @xmath231 , with @xmath10 replaced by @xmath203 .
the effect of this replacement on the gaussian multinomials is given by @xmath242 applying this to @xmath231 , we find @xmath243j + k^2+[k+2(l+1)j+a - b]^2 -m[2k+2(l+1)j+a - b ] } \ !
\mults{m}{k}{k+2(l+1)j+a - b } \right . }
\\ & & \nonumber \\ \lefteqn { \hphantom{g_m^s(b)\times } \hskip -\wd0 \ , \left .
- q^{-(l+1)lj^2-[(l+1)r+la]j - ar + k^2+[k+2(l+1)j+a+b]^2 -m[2k+2(l+1)j+a+b ] } \ !
\mults{m}{k}{k+2(l+1)j+a+b } \right\ } . } \nonumber \setbox0=\hbox{}\end{aligned}\ ] ] in the first term we replace @xmath83 by @xmath244 and in the second term we replace @xmath83 by @xmath245 , where @xmath246 is determined by the requirement that @xmath247 the subscript of @xmath248 , indicating its dependence on @xmath26 , is included for later convenience .
after simplification , we thus obtain @xmath249j } \mults{m } { ( m- \mu_a - a + b)/2 - k - ( l+1)j } { ( m- \mu_a + a - b)/2 - k + ( l+1)j } \right . } \label{eq : a.invg}\\ & & \nonumber \\ \lefteqn{\hphantom{g_m^{2b - s}(b ) \times } \hskip -\wd0 \ , \left
. - q^{(l+1)(l+2)j^2+[(l+2)a+(l+1)s]j+as } \mults{m } { ( m - \mu_a - a - b)/2 - k -(l+1)j } { ( m - \mu_a + a + b)/2 - k + ( l+1)j } \right\}. } \nonumber \setbox0=\hbox{}\end{aligned}\ ] ] [ [ antiferromagnetic - phases ] ] antiferromagnetic phases + + + + + + + + + + + + + + + + + + + + + + + + in contrast to the previous two regimes , we now have antiferromagnetic as well as ferromagnetic phases . at first , we restrict our attention to the antiferromagnetic groundstates , and calculate @xmath250 it follows from the solution ( [ eq : a.solnr2 ] ) that we have to find expressions for the function @xmath231 of equation ( [ eq : a.invg ] ) in the @xmath206 limit .
to do so we use the result @xmath251 we therefore conclude that , in the limit @xmath206 , the auxiliary function @xmath252 is given by a product of virasoro characters @xmath253j } \m q^{(l+2)(l+1)j^2+[(l+2)a+(l+1)s]j+as}\right\ } } \nonumber \\ \lefteqn { \hphantom{\lim_{m\to\infty}q^{m^2/2}g_m^{2b - s}(b ) } = q^{\,(b^2-as)/2-\delta_{a , s}^{(l+2)}+c/24 + 1/48 } \chi_{\mu_a/2}(q ) \ : \chi_{a , s}^{(l+2)}(q ) \vphantom{\sum_a^a } } \label{eq : a.limg3 } \\ \lefteqn { \hphantom{\lim_{m\to\infty}q^{m^2/2}g_m^{2b - s}(b ) } \sim q^{-a^2\la/\pi } \chi_{\mu_a/2}(q ) \ : \chi_{a , s}^{(l+2)}(q ) } \nonumber\end{aligned}\ ] ] with @xmath254 . in the last step
we have used the following rogers - ramanujan identity @xcite for the @xmath255 characters @xmath256 and @xmath257 : @xmath258 in contrast to the ferromagnetic phases treated so far , the local height probabilities for antiferromagnetic phases depend on whether @xmath171 is taken to infinity through odd or even values .
we therefore write @xmath259 , where @xmath260 is defined to be the parity of @xmath261 @xmath262 for @xmath263 this gives @xmath264 replacing @xmath10 by @xmath203 in the solution ( [ eq : a.solnr2 ] ) for the one - dimensional configuration sums and using the result ( [ eq : a.limg3 ] ) , we find @xmath265 where @xmath266 we note that @xmath218 takes the values @xmath219 . performing a conjugate modulus transformation , we find that the local height probabilities are given by @xmath267 \label{eq : a.lhpreg3anti}\ ] ] where the normalisation factor is defined as @xmath268 from the relation between the nome @xmath87 and the nome @xmath269 we obtain @xmath270 we take the conjugate modulus transformation ( [ eq : a.congnome ] ) of the @xmath255 characters after first rewriting them using the formula @xcite @xmath271 furthermore , we rewrite the crossing factor @xmath31 by using the identity ( [ eq : a.alterns ] ) as well as the relation between the nomes @xmath57 and @xmath225 . substituting this into the defining relation of the normalisation factor @xmath272 , we can perform the summation , yielding @xmath273 the proof of the denominator identity is presented in appendix [ app : a.den3 ] .
[ [ ferromagnetic - phases ] ] ferromagnetic phases + + + + + + + + + + + + + + + + + + + + for the ferromagnetic phases of regime 3@xmath88 the local height probabilities are given by @xmath274 from the expressions for @xmath275 in equation ( [ eq : a.solnr2 ] ) it follows that we have to consider the following combinations of the function @xmath231 in the @xmath206 limit : @xmath276 here @xmath231 is given by equation ( [ eq : a.invg ] ) and we have neglected terms that have an extra factor @xmath277 , which do not contribute in the large @xmath171 limit
. we again take @xmath206 through even or odd values of @xmath171 with @xmath26 and @xmath27 fixed and @xmath278 chosen accordingly . for this
we need the result @xmath279 where we have used equation ( [ eq : a.limitgp ] ) .
if we apply this limit to the combination of @xmath231 functions , it follows that @xmath280 } \nonumber \\ \lefteqn { \qquad = q^{(b - a)b/2}\frac{1}{q(q ) } \sum_{k=0}^{\infty } \frac{q^{2k^2+(2\mu_a-1 ) k } } { ( q)_{2k+\mu_a } } } \nonumber\\ \lefteqn { \qquad \quad \times \ ! \ ! \sum_{j=-\infty}^{\infty } \left\{q^{(l+2)(l+1)j^2+[(l+2)a-(l+1)b]j } -q^{(l+2)(l+1)j^2+[(l+2)a+(l+1)b]j+ab}\right\ } } \label{eq : a.nomudep}\\ \lefteqn { \qquad = q^{(b - a)b/2-\delta_{a , b}^{(l+2)}+c/24 - 1/24 } \chi_{1/16}(q ) \chi_{a , b}^{(l+2)}(q ) } \vphantom{\sum_a^a } \nonumber \\ \lefteqn { \qquad \sim q^{-a\la/\pi } \chi_{a , b}^{(l+2)}(q ) . }
\nonumber\end{aligned}\ ] ] to obtain this result , we have used the rogers - ramanujan identity for the @xmath255 character @xmath281 : @xmath282 the left - hand side of this identity does not depend on the value of @xmath263 , and hence the result ( [ eq : a.nomudep ] ) is independent of the parity of @xmath171 . for a ferromagnetic phase
this is indeed what one would expect .
similarly we find that @xmath283 \sim q^{-a\la/\pi } \chi_{a , b+1}^{(l+2)}(q).\ ] ] replacing @xmath10 by @xmath203 in the solution ( [ eq : a.solnr2 ] ) for @xmath187 , and substituting the above results , we get @xmath284 where @xmath285 we note that @xmath218 takes the values @xmath286 . if we continue as for the antiferromagnetic phases , we arrive at the final result @xmath287 \label{eq : a.lhpreg3ferro}\ ] ] where the term occurring in the denominator reads @xmath288 again the summation over @xmath26 can be carried out , yielding @xmath289 a proof of the denominator identity leading to this result is given in appendix [ app : a.den3 ] .
the solution of the recurrence relation for regime 4@xmath88 is obtained by replacing @xmath10 by @xmath203 in the solution ( [ eq : a.solnr1 ] ) for regime 1@xmath88 .
consequently we need to consider the auxiliary function @xmath290 , with @xmath10 replaced with @xmath203 . using the inversion formula ( [ eq : a.qto1onq ] ) and carrying out the same sequence of transformations as for regime 3@xmath88
, we obtain @xmath291j } \mults{m } { ( m- \mu_a - a + b)/2 - k - ( l+1)j } { ( m- \mu_a + a - b)/2 - k + ( l+1)j } \right . } \label{eq : a.invf}\\ & & \nonumber \\ \lefteqn{\hphantom{f_m^{2b - r}(b ) \times } \hskip -\wd0 \ , \left . - q^{(l+1)lj^2+[(l+1)r+la]j+ar } \mults{m } { ( m - \mu_a - a - b)/2 - k -(l+1)j } { ( m - \mu_a + a + b)/2 - k + ( l+1)j } \right\}. } \nonumber \setbox0=\hbox{}\end{aligned}\ ] ] [ [ antiferromagnetic - phases-1 ] ] antiferromagnetic phases + + + + + + + + + + + + + + + + + + + + + + + + we again have antiferromagnetic as well as ferromagnetic phases .
both types admit treatment akin to that applied in regime 3@xmath88 . in the following we therefore leave out some of the details .
we begin by treating the antiferromagnetic groundstates and calculate @xmath292 from the solution ( [ eq : a.solnr1 ] ) we see that we have to obtain the @xmath206 limit of the function @xmath207 in equation ( [ eq : a.invf ] ) . we find that in this limit we get a product of virasoro characters @xmath293 where @xmath294 . for the local height probabilities
this gives @xmath295 where @xmath296 after a conjugate modulus transformation this gives @xmath297 \label{eq : a.lhpreg4anti}\ ] ] with the following normalisation factor : @xmath298 from the relation between the nome @xmath57 and @xmath10 we get @xmath299 using the result of appendix [ app : a.den4 ] for the normalisation factor @xmath300 , we finally obtain = .
[ eq : a.n5den ] [ [ ferromagnetic - phases-1 ] ] ferromagnetic phases + + + + + + + + + + + + + + + + + + + + for the ferromagnetic phases of regime 4@xmath88 , the local height probabilities are @xmath301 from the solution for the configuration sums @xmath302 , as listed in equation ( [ eq : a.solnr1 ] ) , it follows that we have to take the the @xmath206 limit of the following combinations of the function @xmath207 : @xmath303 where @xmath290 is given by equation ( [ eq : a.invf ] ) .
the infinite limit can easily be taken using ( [ eq : a.pairlim ] ) and ( [ eq : a.rr16 ] ) , and we arrive at @xmath304 \sim q^{-a\la/\pi } \chi_{b , a}^{(l+1)}(q ) } \nonumber \\ & & \nonumber \\ \lefteqn { \lim_{m\to\infty } q^{(m^2-m+b)/2 } \left [ f_{m}^{b}(b ) - q^{-m - b } f_{m-1}^{b+2}(b+1 ) \right ] \sim q^{-a\la/\pi } \chi_{b-1,a}^{(l+1)}(q ) . }
\end{aligned}\ ] ] again we observe that the dependence on the parity of @xmath171 has dropped out .
replacing @xmath10 by @xmath203 in the solution ( [ eq : a.solnr1 ] ) for @xmath302 and substituting the above results , we get @xmath305 where @xmath306 we note that @xmath237 takes the values @xmath307 . taking a conjugate modulus transformation , we obtain the final result @xmath308 \label{eq : a.lhpreg4ferro}\ ] ] with the following normalisation factor : @xmath309 performing the sum over @xmath26 , using the denominator identity for @xmath310 in appendix [ app : a.den4 ] , yields @xmath311 following huse @xcite , we define generalised order parameters in terms of the local height probabilities obtained in the previous section : @xmath312 where @xmath313 and @xmath314 .
\label{eq : clhp}\ ] ] for @xmath315 these order parameters are trivial because @xmath316 and @xmath317 . for ferromagnetic phases
there is no distinction between the two sublattice probabilities , and so @xmath318 .
we are not able in general to perform the sum in ( [ eq : gop ] ) . to determine the associated critical exponents we therefore expand the local height probabilities for the four regimes in powers of @xmath225 , and hence of @xmath57
. in this process the critical values of the local height probabilities may also obtained , and are given by @xmath319 in all regimes , and are independent of the boundary spins @xmath27 and @xmath122 . using the exponents obtained via the free energy
, we extract the conformal weights corresponding to the critical exponents of the generalised order parameters .
we first identify the scaling field corresponding to the elliptic nome @xmath57 . from the scaling relations @xmath320 and the exponents @xmath20 and @xmath21 obtained in section [ sec : a.free ] , we find that the conformal weight of the perturbing field is @xmath321 we therefore conclude that the nome @xmath57 corresponds to a perturbation in the direction of the spinless operator @xmath322 , with @xmath323 here @xmath324 denotes the operator of the unitary minimal model with central charge @xmath325 because this is a ferromagnetic regime , the only non - zero order parameters are @xmath326 where @xmath313 . to evaluate the associated critical exponents , we expand equation ( [ eq : a.n1den ] ) using the representation ( [ eq : a.thetasums ] ) of the @xmath71-functions as a series of trigonometric functions . in the @xmath327 limit this yields @xmath328.\ ] ] expanding also the @xmath329-functions in the expression ( [ eq : a.lhpreg1 ] ) for the local height probabilities , and using a simple trigonometric identity for the difference of cosines , gives @xmath330 setting @xmath331 we obtain the aforementioned result for the critical values of the local height probabilities ( [ eq : clhp ] ) .
if we substitute the above expansion into the expression for the order parameters , then use the orthogonality relation @xmath332 \label{eq : a.ortho}\ ] ] and interchange the order of summation , we find @xmath333 ^ 2 - 1}\sin\!\left(\case{[k+1 + 2m(l+1)]s\pi}{l+2}\right ) + o\!\left(t^{3l(l+2)}\right).\ ] ] to leading order we thus find @xmath334 hence , for @xmath335 , the order parameters vanish at criticality with a power law behaviour @xmath336 where the critical exponents @xmath22 are given by @xmath337 here we have used the fact that @xmath338 . to find the corresponding conformal weights @xmath339 we use the scaling relation @xmath340 from this relation we get @xmath341
thus we have obtained all ` diagonal ' weights of the kac table ( [ eq : a.kactable ] ) . to obtain the generalised order parameters for regime 2@xmath88 we closely follow the previous working .
the non - zero order parameters are again given by ( [ eq : a.orderhuse ] ) , with @xmath342 .
expanding the results ( [ eq : a.lhpreg2 ] ) and ( [ eq : a.n2den ] ) for the local height probabilities , we find @xmath343 ^ 2 - 1}\sin\!\left(\case{[k+1 + 2m(l+1)]r\pi}{l}\right ) + o\!\left(t^{3l(l+2)}\right ) .
\label{fred}\ ] ] for @xmath344 , the leading term is obtained for @xmath345 .
when @xmath346 , however , the amplitude of this term vanishes and we need the next - to - leading term @xmath347 . consequently we get @xmath348 where we have used the fact that @xmath237 is odd . from this
we can read off the exponents @xmath349 to find the corresponding conformal weights @xmath339 we again use the scaling relation ( [ eq : a.scaling ] ) .
in addition to the diagonal weights of the kac table this yields the conformal weight @xmath350 : @xmath351 using the identities @xmath352 and the result ( [ eq : a.n3denanti ] ) , it follows from equation ( [ eq : a.lhpreg3anti ] ) that @xmath353 for the order parameters ( [ eq : gop ] ) this yields the relation @xmath354 consequently we can restrict our attention to @xmath355 . as for the regimes 1@xmath88 and
2@xmath88 we first expand the normalisation factor ( [ eq : a.n3denanti ] ) .
we note , however , that we now need more terms in the expansion : @xmath356.\ ] ] expanding the @xmath329-functions appearing in the local height probabilities ( [ eq : a.lhpreg3anti ] ) we arrive at @xmath357 } \nonumber \\ & & \nonumber \\ & & \hspace{7 cm } + o\!\left(t^{2(l-2)(l+1)}\right)\end{aligned}\ ] ] if we apply the orthogonality relation ( [ eq : a.ortho ] ) , after substituting this expression in the definition of the order parameters , we obtain @xmath358 ^ 2 - 1}\sin\!\left(\case{[k+1 + 2m(l+1)]s\pi}{l+2}\right ) \right . }
\nonumber \\ & & \qquad \qquad \quad \ ; + t^{[k+1-l+2m(l+1)]^2+l^2 - 5 } \sin\!\left(\case{[k+1-l+2m(l+1)]s\pi}{l+2}\right ) \vphantom{\sum^a } \label{hannah } \\ & & \nonumber \\ & & \qquad \qquad \quad \ ; \left .
+ t^{[k+1+l+2m(l+1)]^2+l^2 - 5 } \sin\!\left(\case{[k+1+l+2m(l+1)]s\pi}{l+2}\right ) \right\ } + o\!\left(t^{2(l-2)(l+1)}\right ) .
\nonumber\end{aligned}\ ] ] except for @xmath346 , the leading term comes from the first term within the summation , with @xmath345 . defining the exponents @xmath359 as in equation ( [ eq : a.powerlaw ] )
, we can read off @xmath360 when @xmath346 the first and the third term , with @xmath345 and @xmath347 , respectively , are of the same order . using the fact that @xmath218 is odd we find that @xmath361 vanishes . because of the symmetry ( [ ted ] ) this is as it should be . using the relation ( [ ted2 ] )
we readily obtain the exponents @xmath362 from the scaling relation ( [ eq : a.scaling ] ) and the conformal weight @xmath363 in equation ( [ eq : a.deltap ] ) , we thus find @xmath364 we conclude that we have obtained all but one of the diagonal weights of the relevant kac table , shifted by the diagonal @xmath255 weights .
it is this structure , which we observe also in regime 4@xmath88 below , together with the products of virasoro characters which we found in the local height probabilities for these two regimes , which led to the proposal ( [ mipf ] ) for the modular invariant partition function for the corresponding critical branches .
the derivation of the order parameters for the ferromagnetic phases of regime 3@xmath88 proceeds along similar lines to the calculation for the antiferromagnetic phases .
the non - zero order parameters are again given by ( [ eq : a.orderhuse ] ) and , using the results ( [ eq : a.lhpreg3ferro ] ) and ( [ eq : a.n4den ] ) , are found to be @xmath365 ^ 2 - 1}\sin\!\left(\case{[k+1 + 2m(l+1)]s\pi}{l+2}\right ) \right . }
\nonumber \\ & & \qquad \qquad \ ; \;+ t^{[k+1-l+2m(l+1)]^2+l^2 - 5 } \sin\!\left(\case{[k+1-l+2m(l+1)]s\pi}{l+2}\right ) \vphantom{\sum^a } \\ & & \nonumber \\ & & \qquad \qquad \ ; \;\left .
+ t^{[k+1+l+2m(l+1)]^2+l^2 - 5 } \sin\!\left(\case{[k+1+l+2m(l+1)]s\pi}{l+2}\right ) \right\ } + o\!\left(t^{2(l-2)(l+1 ) } \right ) .
\nonumber\end{aligned}\ ] ] for @xmath344 the leading term comes from the first term within the curly braces , with @xmath345 : @xmath366 for @xmath346 the first term with @xmath345 and the third term with @xmath347 are of equal order .
using the fact that @xmath218 is even gives @xmath367 from these two equations we extract the exponents @xmath368 which are the exponents @xmath359 found in ( [ eq : onemissing ] ) , plus the ` missing ' exponent @xmath369 . from the results ( [ eq : a.lhpreg3anti ] ) and ( [ eq : a.n5den ] ) for the local height probabilities , and the symmetry relation @xmath370 we obtain the following results for the order parameters in the small field limit : @xmath371 from this we read off the critical exponents @xmath372 where @xmath373 .
we apply the scaling relation ( [ eq : a.scaling ] ) to the above results to find @xmath374 we have thus obtained the diagonal weights of the kac table , with @xmath375 , shifted by the diagonal @xmath255 weights . from the expression for the order parameters which survive in the ferromagnetic phases ( [ eq : a.orderhuse ] ) , and the results for the local height probabilities ( [ eq : a.lhpreg4ferro ] ) and ( [ eq : a.n6den ] ) , we have @xmath376 from this we obtain a subset of the exponents found for the antiferromagnetic phases .
we are also able to determine that @xmath377 which corresponds to the conformal weight @xmath378 .
[ sec : a.phdiag ] the study of the solvable manifolds spanned by the elliptic nome and the spectral parameter reveals some aspects of the role these manifolds play in a larger parameter space . in particular the off - critical regimes support coexistence between a specified number of phases , each associated with a groundstate , detailed in section [ sec : a.groundstate ] .
the critical branches are the critical points terminating these coexistence lines . in the following
we discuss some aspects of the phase diagram of the dilute a models .
the discussion is not limited to the solvable manifolds , but is restricted to the region where the weights are @xmath0 symmetric .
when the @xmath0 symmetry @xmath379 is obeyed , the weights permit a duality transformation .
this transformation is a direct generalisation of the orbifold duality , relating the a@xmath380 and d@xmath381 temperley - lieb models @xcite , to the dilute a and d models .
since the a@xmath1 and d@xmath1 dynkin diagrams are identical , we have a dual symmetry of the phase diagram in the case @xmath23 . because the phase diagram for @xmath382 is complicated by the large number of states of the model , and by the absence of a dual symmetry , we limit detailed discussion to the case @xmath23 .
only afterwards do we mention which conclusions are valid for general @xmath19 .
the duality transformation for the case @xmath23 can be performed by separating each degree of freedom into an ising - like variable @xmath383 , that discriminates between the states @xmath384 and @xmath385 , and the variable @xmath386 , that decides whether a state is @xmath387 , so that @xmath388 . for a given configuration of @xmath218 ,
the ising variables live on the sites where @xmath389 . since these ising variables take the same value on neighbouring sites , there is effectively only one ising variable in each cluster of @xmath389 sites that are mutually connected ( directly or indirectly ) by nearest neighbour links .
such clusters form an irregular lattice , with interactions between them wherever they are linked via a second neighbour bond . on this lattice
an ordinary kramers - wannier duality transformation can be performed which results in new ising variables on the dual lattice , formed in the same way , but by the clusters where @xmath390 .
the @xmath218-variables , fixed in the procedure so far , are now each replaced by @xmath391 .
the resulting duality transformation on the weights is @xmath392 } \nonumber \\ \lefteqn { \wt{}{2}{1}{2}{1 } = \w{}{1}{2}{1}{2 } + \w{}{3}{2}{1}{2 } } \nonumber \\ \lefteqn { \wt{}{1}{2}{1}{2 } = { \displaystyle{1}\over\displaystyle{2 } } \left [ \w{}{2}{1}{2}{1 } + \w{}{2}{1}{2}{3 } \right ] } \nonumber\end{aligned}\ ] ] where the weights are symmetric under exchange of the states 1 and 3 , and the transformation rules are invariant under reflection in the diagonals of the elementary squares .
the self - dual subspace contains all four solvable critical branches .
it constitutes a transition between a phase symmetric under exchange of the states 1 and 3 and one in which this symmetry is spontaneously broken .
the nature of this phase transition is not the same everywhere in the self - dual subspace . the phase structure within this subspace , which is equivalent to the phase diagram of the o@xmath14 loop model @xcite ,
is indicated in fig .
[ fig : pd ] .
it may be parametrised in terms of the weights @xmath40 .
if we keep the weights @xmath393 fixed , we may describe the self - dual parameter space in terms of the weight @xmath394 of the ferromagnetic states and the weights @xmath395 and @xmath396 of the antiferromagnetic states . when @xmath394 as well as @xmath395 and @xmath396 are small , the transition ( across the self - dual space ) is of second order , in the ising universality class .
it is described by the solvable critical point of branch 2 .
the character of the phases is ferromagnetic .
when the weight @xmath394 is sufficiently large the transition is of first order .
these regions are separated by an ising tricritical transition governed by branch 1 .
when , on the other hand , @xmath395 and @xmath396 are sufficiently large and @xmath394 is relatively small , antiferromagnetic configurations , in which one of the sublattices takes the value @xmath387 while the other sublattice fluctuates between @xmath384 and @xmath385 , are favoured .
there will be a phase separation between regions where one or the other sublattice has @xmath387 .
the onset of this phase separation is an ordinary ising transition , which appears to be independent of the transition across the self - dual subspace .
it is described by the critical point of branch 4 .
the independence of these transitions is reflected in the product structure of the order parameters on this branch , see _
e.g. _ , equation ( [ eq : prodstruc ] ) , and in the value of the central charge @xmath54 , see table [ tab : a.thetable ] .
it implies that the transition across the self - dual subspace is of the same ( ising ) universality class on either side of the ising transition within the subspace .
when the ferromagnetic weight @xmath394 and antiferromagnetic weights @xmath395 , @xmath396 are all large relative to the other weights , the system must make a choice between the first - order regime , in which the ferromagnetic configurations dominate , and the critical antiferromagnetic regime . from the fact that the surface tension between these phases can be made arbitrarily large ,
we conclude that the transition between the two regimes is first order .
the point where this first order transition is met by the other two transitions within the subspace is the critical point of branch 3 .
we further observe that the transition between the ferromagnetic and antiferromagnetic region of the self - dual subspace , persists away from the self - dual subspace , see fig .
[ fig : pd]b .
in fact , when the weights are taken far from self - duality , the model may be reduced to the interacting hard square model @xcite , in which the particles are represented by @xmath384 or @xmath385 , and the empty sites by @xmath387 .
of course , by dual symmetry , the roles are reversed in the opposite extreme .
the hard - square model is known to have an ising critical and tricritical transition into a sublattice - ordered phase .
we expect the critical transition to join continuously with branch 4 and the tricritical transition with branch 3 .
we now see that the critical point of branch 3 plays a central role .
it sits at the intersection of two phase - transition manifolds , _
i.e. , _ the ferro - antiferromagnetic transition and the self - dual subspace .
furthermore , it sits at the line where both transition sheets turn first order . for general @xmath19
we expect the same topology of the phase diagram . though there is no longer a dual symmetry that maps the weight space into itself
, the subspace given by ( [ eq : da.rsos ] ) still plays the role of a phase transition .
the universality class of the transition across this subspace varies with @xmath19 , and is that of regime iii - iv of the @xmath19-state abf model for branch 2 , and of the @xmath193-state abf model for branch 1 .
the transition between ferro- and antiferromagnetic regions remains ising - like for all @xmath19 .
where the transition manifolds intersect , the universality class is that of a direct product of an ising and an abf model ( except where they are first order ) .
again , for general @xmath19 there are directions in the parameter space in which the model can be reduced to the interacting hard square model . to conclude we remark that with the phase diagram proposed above we do not do full justice to the fact that the critical behaviour of branch 3 is that of the product of an ising model and branch 1 .
because the model in regime @xmath90 is in the universality class of the ising model in a magnetic field , we take a special interest in this case , and collect the results in this section .
as we have already observed , we can make the identification @xmath69 to relate the heights of the dilute a@xmath1 model to the states of a spin-1 ising model .
this particular model has the restriction that a ` @xmath72 ' and a ` @xmath73 ' spin may not be adjacent on the lattice . because the nome @xmath57 breaks the up - down symmetry , it plays the role of the magnetic field . in regime @xmath90
we see from table [ tab : a.thetable ] that the central charge of the model is @xmath255 .
using result ( [ eq : fe12 ] ) we can rewrite the free energy as @xmath397 the magnetisation of this ising model can be expressed as a linear combination of the local state probabilities : @xmath398 analogously , we can define a density for the spin-1 model , which is the probability that a site is occupied by a spin of either sign , @xmath399 when we group the following combination of elliptic functions : @xmath400 and make some simplifications , the magnetisation and density are @xmath401 the leading order behaviour is @xmath402 the first two expressions both lead to the exponent @xmath2 .
the last exponent is related by scaling to the conformal weights of the energy and spin operators : @xmath403 . [
sec : a.sumres ] in this paper we have considered the infinite hierarchy of dilute a models
. these models , which belong to the family of dilute a - d - e models , can be viewed as spin-1 generalisations of the solid - on - solid models of andrews , baxter and forrester @xcite . for any integer @xmath404 ,
the dilute a model defines a @xmath19-state solid - on - solid model , with heights labelled by the dynkin diagram of the classical lie algebra a@xmath15 . at the critical point ,
the boltzmann weights ( [ eq : a.weights ] ) of this solid - on - solid model obey the @xmath0 symmetry of the underlying dynkin diagram , but away from the critical point this symmetry is broken for odd values of @xmath19 . for each value of @xmath19
the model has four different critical branches .
two of these branches provide new realisations of the unitary minimal series and the two other branches can be viewed as the direct product of this same series and a @xmath255 model . away from criticality
the four branches yield eight distinct regimes . for all regimes
we have calculated the free energy . from this
we have obtained critical exponents @xmath20 , for even @xmath19 , and exponents @xmath21 , for odd @xmath19 . among these results
is the magnetic exponent @xmath2 of the ising model .
the main part of this paper is concerned , however , with the calculation of the generalised order parameters for the symmetry - breaking models of the dilute a hierarchy .
this calculation involved the evaluation of new sums - of - products identities for theta functions of the rogers - ramanujan type . from the resulting expressions for the order parameters we extracted a set of critical exponents @xmath22 .
for the models corresponding to the unitary minimal series , these exponents correspond to the diagonal conformal weights of the relevant kac table . for the models corresponding to a direct product of a unitary minimal model and a @xmath255 model ,
the critical exponents yield the diagonal weights of the kac table of the minimal model incremented by either one of the diagonal weights of the @xmath255 model .
we have not yet succeeded in computing the order parameters for the non - symmetry - breaking dilute a models , obtained for even values of @xmath19 . in this case
the boltzmann weights no longer enjoy the diagonal property ( [ eq : a.hdef ] ) and an additional diagonalisation is required .
we hope to report on these ` even ' members of the dilute a hierarchy in a future publication .
another open problem is the relation between the symmetry - breaking spin-1 ising model , corresponding to the dilute a@xmath1 models in the regimes 2@xmath405 and 2@xmath406 , and the integrable field theory of the critical ising model in a magnetic field , found by zamolodchikov @xcite . as shown by smirnov @xcite , zamolodchikov s @xmath31-matrix corresponds to an rsos projection of the izergin - korepin r - matrix .
hence the dilute a@xmath1 model provides a likely candidate for describing the corresponding solvable lattice model .
the precise relation between zamolodchikov s @xmath31-matrix , which has a structure related to the lie algebra e@xmath407 , and the dilute a@xmath1 model remains , however , unclear . by studying the excitation spectrum of the dilute a models , and in particular the a@xmath1 case
, we hope to establish the connection , if present , in the near future .
this work has been supported by the stichting voor fundamenteel onderzoek der materie ( fom ) and the australian research council .
[ app : a.h - fun ] in this appendix we give the weights ( [ eq : a.weights ] ) in a form which is more suitable for considering the ordered or strong - field limit , and hence for obtaining the weight function @xmath129 defined in equation ( [ eq : a.hdef ] ) .
we first define the new variables @xmath408 in terms of these variables , the ordered limit is given by @xmath409 with @xmath410 fixed . using the conjugate nome expressions ( [ eq : a.congnome ] ) of the theta functions and setting @xmath411
the boltzmann weights in conjugate modulus parametrisation read @xmath412 where we have omitted an overall normalisation factor @xmath413 and @xmath414 is defined in equation ( [ eq : a.gdef ] ) . in this alternative representation of the boltzmann weights
, we can readily extract their leading behaviour in the ordered limit . to do so , we make use of the simple properties of the function @xmath415 as listed in equation ( [ eq : a.limefun ] ) of appendix [ app : a.elliptic ] .
it is clear from the definition of @xmath416 that the result depends on the value of @xmath45 . for regime @xmath89 , using definition ( [ eq : a.setreg1 ] ) , we get @xmath417 to facilitate the proof of solution ( [ eq : a.solnr1 ] ) we include the following two terms to the above list : @xmath418 and @xmath419 .
[ app : a.denom ] in this appendix we prove the six denominator identities used in section [ sec : a.thermo ] to normalise the local height probabilities .
all identities involve sums over products of elliptic functions similar to the sums - of - products identities of rogers and ramanujan . to prove this identity we start by recasting the the left - hand side using the representation ( [ eq : a.thetasums ] ) of the @xmath71-functions as infinite sums : @xmath426 summing the roots of unity over @xmath26 , we find that the above expression vanishes unless @xmath427 , with @xmath428 .
hence we get @xmath429 ^ 2}\end{aligned}\ ] ] where we have used the fact that @xmath218 is odd .
we now carry out a sequence of transformations to diagonalise the quadratic exponent of @xmath225 .
first we make the substitution @xmath430 followed by @xmath431
. then we replace the sum over @xmath83 by a sum over @xmath432 and @xmath20 by setting @xmath433 , where @xmath20 has to be summed over @xmath434 and 2 .
similarly , we replace the sum over @xmath171 by a sum over @xmath30 and @xmath167 by setting @xmath435 .
finally we make the replacements @xmath436 and @xmath437 .
after all these transformations we obtain @xmath438 where , in the second step , we have used the definition ( [ eq : a.efunc ] ) of the function @xmath415 .
the terms with @xmath439 are identically zero and the two terms with @xmath440 , @xmath441 cancel . the remaining four terms factorise to give @xmath442 \vphantom{\sum_a^a } } \nonumber \\ \lefteqn { \qquad \qquad \times \ ! \left [ e\!\left(-\e^{6\pi \is s/(l+2 ) } \ , t^{12l(l+1)},t^{36l(l+1)}\right ) \right . \vphantom{\sum_a^a } } \label{eq : a.lhsntt}\\ \lefteqn { \qquad \qquad \qquad \qquad \qquad \qquad \left . -\e^{2\pi
\is s/(l+2 ) } e\!\left(-\e^{-6\pi \is s/(l+2 ) } \ , t^{12l(l+1 ) } , t^{36l(l+1)}\right ) \right ] .
\vphantom{\sum^a } } \nonumber\end{aligned}\ ] ] we now note the following two elliptic function identities : @xmath443 to prove the first identity we note that the function @xmath444 , defined as the ratio of the left - hand side over the right - hand side , has the following periodicity property : @xmath445 . since the function @xmath446 is analytic in @xmath447 , the only possible poles in the period annulus @xmath448 are the zeros @xmath449 and @xmath450 of the function @xmath451 . using formula ( [ eq : a.limefun ] ) to manipulate the function @xmath415 it is readily verified that these poles have zero residue .
hence , by liouville s theorem , @xmath452 is constant .
setting @xmath453 shows that this constant is one .
the second identity is actually a corollary of the first . replacing @xmath57 by @xmath454 , then setting @xmath455 and again using formula ( [ eq : a.limefun ] ) , yields @xmath456 applying both identities and the relation ( [ eq : a.thetasase ] ) of appendix [ app : a.elliptic ] , we obtain the desired right - hand side of ( [ eq : a.denom1 ] ) . as before , we carry out a sequence of transformations to diagonalise the quadratic exponent of @xmath225 .
first we make the replacements @xmath461 and @xmath431 .
then we set @xmath433 and @xmath435 and finally we substitute @xmath436 followed by @xmath437 . after making these transformations
we obtain @xmath462 the terms with @xmath439 vanish and the terms with @xmath440 cancel .
the remaining four terms can be written as @xmath463 \vphantom{\sum_a^a } } \\
\lefteqn{\qquad \qquad\times \ ! \left [ e\!\left(-\e^{6\pi \is
r / l } \ , t^{12(l+1)(l+2)},t^{36(l+1)(l+2)}\right ) \right . \vphantom{\sum_a^a } } \nonumber \\
\lefteqn{\qquad \qquad \qquad \qquad \qquad \qquad \left .
- \e^{2\pi \is
r / l } e\!\left(-\e^{-6\pi \is r / l } \ , t^{12(l+1)(l+2 ) } , t^{36(l+1)(l+2)}\right ) \right ] . \vphantom{\sum^a } } \nonumber\end{aligned}\ ] ] if we again apply the two identities of equation ( [ eq : a.eid ] ) and use the relation ( [ eq : a.thetasase ] ) , we obtain the required right - hand side of equation ( [ eq : a.denom2 ] ) .
we split the summation on the left - hand side depending on the parity of @xmath26 .
using the fact that @xmath467 , we get @xmath468 we now represent the theta - functions with @xmath26-dependent arguments by series to obtain @xmath469 . }
\end{aligned}\ ] ] performing the sum over @xmath26 , we find that @xmath470 and hence @xmath471 ^ 2 } \\ & & \qquad \qquad \quad \times \ : \left [ ( -1)^m \td\!\left(\case{\pi}{8 } , ( -1)^{\sigma}t^{(l+1)(l+2)/4}\right ) + \td\!\left(\case{3\pi}{8 } , ( -1)^{\sigma}t^{(l+1)(l+2)/4}\right)\right ] \nonumber\end{aligned}\ ] ] where we have used the fact that @xmath218 is odd .
again we carry out a sequence of transformations to diagonalise the quadratic exponent of @xmath225 .
first we make the replacements @xmath472 and @xmath431 , followed by the substitutions @xmath433 and @xmath473 .
finally we set @xmath436 and @xmath437 to obtain @xmath474 \vphantom{\sum_a^a } \nonumber \\ & & \quad \qquad \times \ : \e^{\pi \is s(\beta+1)/(l+2 ) } e\!\left(-\e^{4\pi \is s/(l+2 ) } \ , t^{2(l+1)(l-2)(3+\beta ) } , t^{8(l+1)(l-2)}\right ) \vphantom{\sum_a^a } \\ & & \quad \qquad \times \ : e\!\left(t^{4(l^2 - 4)(2+\alpha)},t^{12(l^2 - 4)}\right ) e\!\left(-t^{2(l+1)(l+2)(5 + 4\alpha+3\beta ) } , t^{24(l+1)(l+2)}\right ) .
\vphantom{\sum^a } \nonumber\end{aligned}\ ] ] in this expression the terms with @xmath439 vanish and the terms with @xmath440 and @xmath475 cancel . using the simple identity @xmath476 which follows directly from ( [ eq : a.thetasums ] ) , the remaining four terms factorise as @xmath477 } \nonumber \\ \lefteqn{\vphantom{\sum_a^a } \qquad \times \left [ e\!\left(-\e^{4\pi \is s/(l+2 ) } \ , t^{2(l+1)(l-2 ) } , t^{8(l+1)(l-2)}\right ) \right . } \label{eq : a.fred } \\
\lefteqn{\vphantom{\sum_a^a } \qquad \qquad \qquad \qquad \qquad \qquad \left .
-\e^{2\pi \is s/(l+2 ) } e\!\left(-\e^{-4\pi \is s/(l+2 ) } \ , t^{2(l+1)(l-2)},t^{8(l+1)(l-2)}\right ) \right].}\nonumber\end{aligned}\ ] ] we now note the identities @xmath478 the first one follows directly from liouville s theorem and the second one can be obtained from the first upon setting @xmath479 after replacing @xmath57 by @xmath454 .
if we apply these two identities , as well as equation ( [ eq : a.thetasase ] ) , we find precisely the right - hand side of equation ( [ eq : a.denom3 ] ) .
this completes the proof . in order to diagonalise the quadratic exponent of @xmath225
, we make the replacements @xmath430 and @xmath431 followed by the substitutions @xmath433 and @xmath483 .
further replacing @xmath436 and @xmath437 leads to the result @xmath484 the terms with @xmath439 vanish and the other terms may be combined to give @xmath485 } \nonumber \\ \lefteqn{\vphantom{\sum_a^a } \qquad \times \left [ e\!\left(\e^{4\pi \is s/(l+2)}\,t^{2(l+1)(l-2)},t^{8(l+1)(l-2)}\right ) \right . } \\
\lefteqn{\vphantom{\sum_a^a } \qquad \qquad \qquad \qquad \qquad \qquad \left .
-\e^{2\pi \is s/(l+2 ) } e\!\left(\e^{-4\pi \is s/(l+2)}\ , t^{2(l+1)(l-2)},t^{8(l+1)(l-2)}\right ) \right ] . }
\nonumber\end{aligned}\ ] ] if we multiply this expression with the prefactors outside the sum in equation ( [ eq : a.denom3b ] ) and use the following identities @xmath486 we find precisely the right - hand side of equation ( [ eq : a.denom3b ] ) .
the proof of these two relations is similar to those obtained earlier .
the first identities follows from application of liouville s theorem and the second identity is a consequence of the first . replacing the nome @xmath57 by @xmath454 and then setting @xmath455 yields @xmath487 following the proof for the antiferromagnetic phase of regime 3@xmath88 , we split the summation on the left - hand side depending on the parity of @xmath26 . representing the theta functions with @xmath26-dependent arguments by series yields @xmath489 . }
\end{aligned}\ ] ] summing the roots of unity over @xmath26 we see that @xmath490 and as a result we have @xmath491 ^ 2 } } \nonumber\\ & & \qquad \qquad \qquad \qquad \times \left [ ( -1)^m \td\!\left(\case{\pi}{8 } , ( -1)^{\sigma}t^{l(l+1)/4}\right)+ \td\!\left(\case{3\pi}{8 } , ( -1)^{\sigma}t^{l(l+1)/4}\right)\right]\end{aligned}\ ] ] where we have used the fact that @xmath237 is odd .
we diagonalise the quadratic exponent of @xmath225 by carrying out the following sequence of replacements .
we first set @xmath492 and @xmath431 .
we then replace @xmath433 and @xmath473 and finally we substitute @xmath436 followed by @xmath437 .
we then have @xmath493 \vphantom{\sum_a^a } \nonumber \\ & & \qquad \quad \times \ : \e^{\pi \is r(\beta+1)/l } e\!\left(- \e^{4\pi \is r / l}\ , t^{2(l+1)(l+4)(3+\beta ) } , t^{8(l+1)(l+4)}\right ) \vphantom{\sum_a^a } \\ & & \qquad \quad \times \ : e\!\left(t^{4l(l+4)(2+\alpha)},t^{12l(l+4)}\right ) e\!\left(-t^{2l(l+1)(5 + 4\alpha+3\beta ) } , t^{24l(l+1)}\right ) .
\vphantom{\sum^a } \nonumber\end{aligned}\ ] ] as before , the terms with @xmath439 vanish and the terms with @xmath440 and @xmath475 cancel . using the relation ( [ eq : a.simpleident ] ) ,
the remaining four terms can be written as @xmath494 \vphantom{\sum_a^a } } \nonumber \\
\lefteqn{\qquad \qquad \times \ ! \left
[ e\!\left(-\e^{4\pi \is r /l } \ ,
t^{2(l+1)(l+4 ) } , t^{8(l+1)(l+4)}\right ) \right . \vphantom{\sum_a^a } } \\ \lefteqn { \qquad \qquad \qquad \qquad \qquad \qquad \left . -\e^{2\pi \is r / l } e\!\left(-\e^{-4\pi \is r / l } \ , t^{2(l+1)(l+4)},t^{8(l+1)(l+4)}\right ) \right ] . \vphantom{\sum^a } } \nonumber\end{aligned}\ ] ] after application of the identities ( [ eq : a.useful ] ) , we obtain the required right - hand side of ( [ eq : a.denom4 ] )
. this expression is diagonalised by carrying out the sequence of transformations @xmath461 , @xmath431 , @xmath433 , @xmath483 , @xmath436 and @xmath437 .
this gives @xmath497 here the terms with @xmath439 vanish .
the other terms may be combined to @xmath498 \vphantom{\sum_a^a } } \nonumber \\
\lefteqn{\qquad \qquad \times \ ! \left [ e\!\left(\e^{4\pi \is r / l } \ , t^{2(l+1)(l+4)},t^{8(l+1)(l+4)}\right ) \right . \vphantom{\sum_a^a } } \\
\lefteqn { \qquad \qquad \qquad \qquad \qquad \qquad \left .
-\e^{2\pi \is r / l } e\!\left(\e^{-4\pi \is r / l } \ , t^{2(l+1)(l+4)},t^{8(l+1)(l+4)}\right ) \right ] .
\vphantom{\sum^a } } \nonumber\end{aligned}\ ] ] if we again use the identities ( [ eq : a.handy ] ) , and multiply the above expression by the correct prefactors from the left - hand side of ( [ eq : a.denom4b ] ) , we indeed obtain the right - hand side of ( [ eq : a.denom4b ] ) .
[ app : a.elliptic ] in this appendix we list several definitions and relations for the jacobian @xmath71-functions , used in the main text . for a more complete introduction ,
we refer the reader to _
@xcite .
the four standard @xmath71-functions , of _ nome _ @xmath57 , @xmath499 , are defined as the following infinite sums @xcite : @xmath500 by virtue of jacobi s triple product identity @xcite , the theta functions admit a representation as infinite products @xmath501 another function which proves to be useful is @xmath502 from its definition it follows immediately that @xmath503 where the function @xmath504 is defined as @xmath505 the @xmath71-functions can be expressed in terms of the @xmath415-function as @xmath506 likewise , the _ conjugate modulus _ transformation of the theta functions , which relates @xmath71-functions of nome @xmath87 to those of nome @xmath507 can be written as @xmath508 we note that these relations follow directly from the poisson summation formula .
s. o. warnaar , b. nienhuis and k. a. seaton , _ a critical ising model in a magnetic field _
preprint itfa 93 - 04 , amsterdam , to appear in _ yang - baxter equations in paris _ j .- m .
maillard ed .
( world scientific , singapore ) . | the free energy and local height probabilities of the dilute a models with broken @xmath0 symmetry are calculated analytically using inversion and corner transfer matrix methods .
these models possess four critical branches .
the first two branches provide new realisations of the unitary minimal series and the other two branches give a direct product of this series with an ising model .
we identify the integrable perturbations which move the dilute a models away from the critical limit .
generalised order parameters are defined and their critical exponents extracted .
the associated conformal weights are found to occur on the diagonal of the relevant kac table . in an appropriate regime
the dilute a@xmath1 model lies in the universality class of the ising model in a magnetic field . in this case
we obtain the magnetic exponent @xmath2 directly , without the use of scaling relations .
3.6 mm 3.6 mm # 1 |
about half of the nuclei with @xmath0 observed in nature are formed by the so - called rapid neutron - capture process ( or r - process ) of nucleosynthesis , occurring in explosive stellar events .
the r - process is believed to take place in environments characterized by high neutron densities ( @xmath1 ) , so that successive neutron captures proceed into neutron - rich regions well off the @xmath2-stability valley forming exotic nuclei that can not be produced and therefore studied in the laboratory .
if the temperatures or the neutron densities characterizing the r - process are low enough to break the @xmath3 equilibrium , the r - abundance distribution depends directly on the neutron capture rates by the so - produced exotic neutron - rich nuclei @xcite .
the neutron capture rates are commonly evaluated within the framework of the statistical model of hauser - feshbach ( although the direct capture contribution can play an important role for such exotic nuclei ) .
this model makes the fundamental assumption that the capture process takes place with the intermediary formation of a compound nucleus in thermodynamic equilibrium . in this approach ,
the maxwellian - averaged @xmath4 rate at temperatures of relevance in r - process environments strongly depends on the electromagnetic interaction , i.e the photon de - excitation probability .
the well known challenge of understanding the r - process abundances thus requires that one be able to make reliable extrapolations of the e1-strength function out towards the neutron - drip line . to put the description of the r - process on safer grounds , a great effort must therefore be made to improve the reliability of the nuclear model .
generally speaking ,
the more microscopic the underlying theory , the greater will be one s confidence in the extrapolations out towards the neutron - drip line , provided , of course , the available experimental data are also well fitted .
large scale prediction of e1-strength functions are usually performed using phenomenological lorentzian models of the giant dipole resonance ( gdr ) @xcite .
several refinements can be made , such as the energy dependence of the width and its temperature dependence @xcite to describe all available experimental data .
the lorentzian gdr approach suffers , however , from shortcomings of various sorts . on the one hand , it is unable to predict the enhancement of the e1 strength at energies around the neutron separation energy demonstrated by various experiments , such as the nuclear resonance fluorescence . on the other hand ,
even if a lorentzian - like function provides a suitable representation of the e1 strength for stable nuclei , the location of its maximum and its width remain to be predicted from some systematics or underlying model for each nucleus . for astrophysical applications , these properties have often been obtained from a droplet - type model @xcite .
this approach clearly lacks reliability when dealing with exotic nuclei , as already demonstrated by @xcite .
recently an attempt was made to derive microscopically the e1 strength for the whole nuclear chart @xcite .
the dipole response was calculated with the quasiparticle random phase approximation ( qrpa ) on top of hartree - fock+bcs ( hfbcs ) description @xcite .
the only input of this approach was the skyrme effective interaction injected in the hfbcs model .
these microscopic calculations predicted the presence of a systematic low - lying component in the e1 strength for very neutron - rich nuclei .
this low - lying component influences the neutron capture rate , especially if located in the vicinity of the neutron separation energy @xmath5 . in our previous hfbcs and qrpa microscopic
approach @xcite , the pairing correlation in the bcs model was determined assuming a simple constant - gap pairing interaction .
in addition , in the case of the highly neutron - rich nuclei that are of particular interest in the context of the r - process , the validity of the bcs approach to pairing is questionable , essentially because of the role played by the continuum of single - particle neutron states ( see @xcite , and references therein ) .
therefore the impact of the newly - derived e1-strength functions on the cross section prediction could only be evaluated qualitatively .
it was found that the radiative neutron capture cross sections by neutron - rich nuclei were systematically increased by the hfbcs+qrpa calculations @xcite with respect to the one obtained using lorentzian - like strength functions .
predictions with different forces have been compared , but no conclusions could be drawn regarding their intrinsic quality to predict the e1 strength .
the final large - scale hfbcs+qrpa calculations performed in @xcite were obtained on the basis of the skyrme force denoted sly4 @xcite . in the present paper
we calculate the dipole strength with one of the most accurate and reliable microscopic model available to date , namely the hartree - fock - bogoliubov ( hfb ) and qrpa models @xcite . as recalled in sect .
2.1 , the ground state is described within the hfb model .
effective interactions of the skyrme type characterized by different values of the nucleon effective mass and prescriptions for the pairing interaction are considered .
the collective gdr mode is obtained by qrpa calculations on top of the hfb calculations , as described in sect .
the residual interaction is derived self - consistently from the nucleon - nucleon effective interaction , which is the only input of the hfb calculation . to describe the damping of the collective motions on microscopic grounds , the second - rpa ( srpa ) described by @xcite
is adopted ( sect .
this approach strongly improves the reliability of the predictions by eliminating the phenomenological spreading of the qrpa strength determined in our previous hfbcs+qrpa calculations @xcite .
this new approach allows us to determine on a more quantitative and reliable ground the photoabsorption cross section and consequently to judge the ability of the forces to reproduce experimental data . in order to select the most adequate interaction for the e1-strength calculation , the hfb+qrpa prediction
are compared with photoabsorption data for 48 spherical nuclei @xcite ( sect . 3 ) .
the hfb+qrpa predictions are further compared , in sect . 3 , with low - energy experimental data and generalized to include deformation effects in a phenomenological way .
all these drastic improvements compared to the previous hfbcs and qrpa models allow to provide quantitative predictions of the e1-strength function , also for exotic neutron - rich nuclei ( sect . 4 ) .
the predicted gdr strengths are used to estimate all the radiative neutron capture rates of relevance for nucleosynthesis applications ( sect .
the long - term goal of microscopic models is to describe on the same ground a wide variety of nuclear structure properties ( in particular , magicity and pairing correlations in open - shell nuclei ) for both stable and exotic nuclei .
the hfb and qrpa models allows to treat , in a self - consistent way , pairing effects on the ground state as well as collective excitations for nuclei ranging from the valley of stability to the drip - line .
the qrpa considers nuclear excitation as a collective superposition of two quasiparticle ( qp ) states built on top of the hfb ground state @xcite .
this collective aspect of the excitation makes the qrpa an accurate tool to investigate the e1-strength function , in both closed and open shell nuclei .
the hfb calculations considered in the present work are fully detailed in @xcite .
they are based on the conventional skyrme force of the form @xmath6 the pairing force acting between like nucleons is treated in the full bogoliubov framework with a @xmath7-function pairing force of the form @xcite
@xmath8~ \delta(\mbox{\boldmath$r$}_{ij } ) \quad , \label{eq_surf}\ ] ] where @xmath9 is the density and @xmath10 its equilibrium value in symmetric nuclear matter .
two types of pairing forces are considered here , a volume density - independent force characterized by @xmath11 and a volume plus surface ( i.e density - dependent ) force with @xmath12 and @xmath13 .
this latter prescription originates from the calculations of the pairing gap in infinite nuclear matter at different densities performed by garrido et al .
@xcite using a bare " or realistic " nucleon - nucleon interaction .
this density - dependent pairing has also been found to be compatible with experimental nuclear masses by @xcite , provided the space of single - particle states over which such a pairing force is allowed to act is truncated to about @xmath14 mev around the fermi energy .
note finally that in the present approach the strength parameter @xmath15 is allowed to be different for neutrons and protons , and also to be slightly stronger for an odd number of nucleons ( @xmath16 ) than for an even number ( @xmath17 ) , i.e. , the pairing force between neutrons , for example , depends on whether @xmath18 is even or odd . for odd-@xmath19 and odd - odd nuclei ,
the blocking approximation is used , as detailed in @xcite .
based on this skyrme - hfb approach , a number of effective forces have been determined recently @xcite , the parameters of the underlying forces being fitted _ exclusively _ to all the 2135 available experimental masses @xcite , with some additional constraints regarding the stability of neutron matter and the incompressibility of nuclear matter .
the parameters corresponding to these six forces named bsk2-bsk7 are summarized in table [ tab_sky ] .
are also included in table [ tab_sky ] the effective isoscalar ( @xmath20 ) , isovector ( @xmath21 ) nucleon mass and the root - mean - square ( rms ) deviations @xmath22 between the measured and estimated masses for the 2135 nuclei with @xmath23 .
more details about these forces can be found in @xcite .
the major differences between these six forces are found in the density - dependence of the pairing force and the adopted isoscalar effective nucleon mass . while bsk2 and bsk3 have been built without constraining the effective mass ( leading to a value of @xmath24 ) , bsk4 and bsk5 are constrained by @xmath25 , as inferred from the extended brckner - hartree - fock calculations of asymmetric nuclear matter @xcite and bsk6 and bsk7 by @xmath26 , as inspired from the more traditional symmetric nuclear - matter calculations ( e.g @xcite ) .
the mass - data fits with the bsk47 interaction are almost as good as those obtained with bsk23 , in which @xmath27 is unconstrained , so that such a mass fit can not be used to discriminate between the different skyrme forces and the corresponding optimal choice for the nucleon effective mass or the pairing interaction . for comparison purposes ,
we also consider the sly4 skyrme force @xcite used in our previous hfbcs+qrpa calculation @xcite .
the sly4 parameters are given in table [ tab_sky ] , the pairing interaction corresponding to the one determined in @xcite ( the mass rms deviation is however not available for the sly4 force ) .
.some properties of the skyrme forces bsk2-bsk7 and sly4 ( see text for more details ) .
the last line corresponds to the rms deviation between predicted and experimental masses for the full set of 2135 spherical and deformed nuclei . [ cols="^,^,^,^,^,^,^,^",options="header " , ] as illustrated in figs .
[ fig_gdr1]-[fig_gdr3 ] and table [ tab_cst ] , the prediction of the gdr parameters is force - dependent .
as far as the position of the peak energy is concerned , most of the forces overpredict the peak energy of light nuclei and underpredict it for the heavier species .
the best agreement is found for the bsk6 - 7 forces which still overestimate @xmath28 for the lightest elements , but give an excellent agreement for @xmath29 elements .
it can be concluded that , within the present hfb+qrpa model , skyrme forces need to have a low effective nucleon mass @xmath30 to correctly predict the gdr characteristics .
an effective mass as low as the one used in sly4 requires a particularly low value of the interference factor ( see table [ tab_cst ] ) which simultaneously give rise to fine structure effects not observed experimentally ( see fig . [ fig_gdr3 ] ) .
interestingly , the density dependence of the pairing interaction has a minor impact on the prediction of the e1-strength function .
almost no differences are found among the 2 couples of forces bsk4 vs bsk5 or bsk6 vs bsk7 ( see in particular fig . [ fig_gdr3 ] ) .
finally , note that the srpa effect to shift the energy peak was not taken into account by the phenomenological lorentzian damping adopted in our previous hfbcs+qrpa work @xcite , and obviously modifies the conclusion drawn regarding the ability of the sly4 force to predict the location of the gdr . the agreement found here for the sly4 interaction ( fig .
[ fig_gdr3 ] ) is worse than it used to be in @xcite where an rms deviation of 0.457 mev was obtained on the centroid energy for the same sample of spherical nuclei .
finally , regarding the amplitude of the e1-strength function , the qrpa equations are solved so as to exhaust the thomas - reiche - kuhn sum rule .
this adopted ewsr corresponds to @xmath31\ ] ] and is found to reproduce well the peak cross section measured experimentally , as illustrated in fig .
[ fig_sigmax ] .
the resulting deviation can be characterized by an rms deviation factor @xmath32 defined as @xmath33^{1/2 } \label{eq_rms}\ ] ] where @xmath34(@xmath35 ) is the theoretical ( experimental ) peak cross section and @xmath36 the number of nuclei in the experimental sample .
all these results show that among the six skyrme forces studied here , both the bsk6 and bsk7 forces not only reproduce extremely well the experimental masses ( with a rms deviation as low as 0.676 mev on the 2135 known masses ) , but also is well adapted to describe the e1 collective excitations . for this reason ,
all further calculations are performed with bsk7 as our standard force .
it should be stressed that for more than thirty years , phenomenological effective interactions were developed using exclusively ground state properties of nuclei , such as binding energies , radii or spectroscopic quantities .
this was initiated through the skyrme hartree - fock model by @xcite .
nuclear forces are traditionally determined by fitting such ground state properties for less than ten or so nuclei .
recently , progress has been achieved in determining the skyrme force by fitting essentially all the mass data @xcite . the only excited feature taken into account
so far was the giant monopolar resonance energy @xcite in order to predict the infinite matter compressibility modulus .
the present hfb+qrpa model ( with the srpa corrections ) allows to consider nuclear excitations such as gdr in the development of phenomenological effective interactions . in the case of deformed spheroidal nuclei ,
the gdr splits into two major resonances as a result of the different resonance conditions characterizing the oscillations of protons against neutrons along the axis of rotational symmetry and an arbitrary axis perpendicular to it . in the phenomenological approach ,
the lorentzian - type formula is generalized to a sum of two lorentzian - type functions of energies @xmath37 and width @xmath38 @xcite , such that @xmath39 where @xmath40 is the ratio of the diameter along the axis of symmetry to the diameter along an axis perpendicular to it . in turn , the width @xmath38 of each peak is given by the same deformation dependence as the respective energy @xmath37 @xcite .
a similar splitting of the resonance strength for deformed nuclei is applied within the srpa procedure given by eq .
( [ loren ] ) , the lorentzian function at a given energy @xmath41 being split with an equal strength into two lorentzian functions centered according to eq .
( [ eq_def ] ) and characterized by a width @xmath42 ( see eq .
[ width ] ) obtained from the same relations ( eq . [ eq_def ] ) . as already found in @xcite
, distributing the strength equally between the two resonance peaks gives optimal location and relative strength of both gdr centroid energies as observed experimentally .
we illustrate in fig .
[ fig_u235 ] how the photoabsorption cross section in @xmath43u peaked around 12 mev in the spherical approximation is split into the two observed peaks .
the same deformation effects are applied to all nuclei predicted to be deformed by the hfb calculation based on the bsk7 force . for practical astrophysics applications ,
it is of first importance to describe the tail of the gdr at low energies , i.e around the neutron separation energy , as reliably as possible @xcite .
experimental e1 strengths at low energies are available through average resonance capture ( arc ) data @xcite or recent measurements of @xmath44-ray spectra in light - ion reactions @xcite . however , such data are related to the so - called `` downwards '' e1-strength function which determines the average width of the @xmath44-decay , while the photoexcitation data considered so far depend on the `` upwards '' e1-strength function associated with @xmath44-absorption .
when dealing with @xmath44-decay data , a temperature - dependent correction factor is traditionally introduced in the expression of the gdr width to take the collision of quasiparticles into account @xcite . in order to guarantee the compatibility with photoabsorption data , we introduce in the srpa procedure such a collision term by adding to the width @xmath42 ( see eq .
[ width ] ) a temperature - dependent correction term as @xmath45 \label{tdep}\ ] ] where @xmath46 refers to the temperature of the absorbing state , @xmath28 is the peak energy of the gdr and @xmath47 a normalization constant . in all calculations performed in the present work ,
the temperature is derived from the microscopic statistical model of nuclear level densities @xcite . as shown below
, adopting @xmath48 gives excellent agreement with most of the available data .
[ fig_nd144 ] illustrates in the specific case of the spherical @xmath49nd nucleus , that the e1-strength data derived from primary photon spectra in the ( n,@xmath44 ) reaction around 68 mev @xcite or ( n,@xmath50 ) reaction around @xmath51 mev @xcite are correctly reproduced at low energies with the @xmath46-dependent correction given by eq .
( [ tdep ] ) with @xmath48 .
the energy dependence of the collision term introduced in eq .
( [ tdep ] ) and already suggested in @xcite is of particular importance , since it is responsible for the @xmath52 behavior of the e1-strength function observed experimentally @xcite .
it is also found to affect the e1 strength around the neutron binding energy , as seen in figs .
[ fig_nd144][fig_fe1 ] .
in addition , we compare in fig .
[ fig_fe1 ] the qrpa predictions with the compilation of experimental e1-strength functions at low energies ranging from 4 to 8 mev @xcite for nuclei from @xmath53 up to @xmath54 .
the data set includes resolved - resonance measurements , thermal - captures measurements and photonuclear data . in a certain number of cases the original experimental values need to be corrected , typically for non - statistical effects , so that only values recommended by @xcite are considered in fig .
[ fig_fe1 ] .
qrpa predictions are globally in good agreement with experimental data at low energies in the whole nuclear chart . the average and rms deviations , as defined in eq .
( [ eq_rms ] ) , on the 62 experimental data have been estimated .
the @xmath46-independent predictions underpredict the e1 strength by an average factor of 1.6 , while on average the @xmath46-dependent formula ( assuming @xmath48 in eq
. [ tdep ] ) is in perfect agreement with the data .
the respective rms deviation factors are @xmath55 and 2.1 for the @xmath46-independent and @xmath46-dependent results .
these results show that including a @xmath46-dependence in the e1 strength to describe the @xmath44-decay data globally improves the agreement .
a qualitative agreement is also obtained with the e1-strength function derived at low energy from primary photon spectra in light - ion reactions @xcite , although the fine structure pattern are not reproduced .
large - scale qrpa calculations based on the bsk7 skyrme force have been performed for all @xmath56 nuclei lying between the proton and the neutron driplines , i.e some 8300 nuclei .
the srpa is applied to all distributions . in the neutron - deficient region , as well as along the valley of @xmath2-stability , the resulting e1-strength functions are very similar to the empirical lorentzian - like approximation .
when dealing with neutron - rich nuclei , the qrpa predictions start deviating from a simple lorentzian shape and results quantitatively similar to @xcite are obtained . in particular ,
some extra strength is found to be located at an energy lower than the gdr energy .
the more exotic the nucleus , the stronger this low - energy component .
this is illustrated in fig .
[ fig_sn ] for the e1-strength function in the sn isotopic chain .
all nuclei shown in fig .
[ fig_sn ] are predicted to be spherical in the hfb calculations based on the bsk7 force @xcite . for the @xmath57 neutron - rich isotopes , an important part of the strength
is concentrated at low energies ( @xmath58 mev ) .
phenomenological models are unable to predict such low energy components .
in particular for @xmath59sn , all phenomenological systematics ( as used for cross section calculation ) predict a @xmath44-ray strength peaked around 15 mev with a full width at half maximum of about 4.5 mev @xcite which is obviously very different from the microscopic estimate ( fig .
[ fig_sn ] ) .
more generally , the present hfb+qrpa calculation confirms that the neutron excess affects the spreading of the isovector dipole strength , as well as the centroid of the strength function .
the energy shift is larger than predicted by the usual @xmath60 or @xmath61 dependence given by the phenomenological liquid drop approximations @xcite .
the above - described feature of the qrpa e1-strength function for nuclei with a large neutron excess is qualitatively independent of the adopted effective interaction .
the radiative neutron capture cross section is estimated within the statistical model of hauser - feshbach making use of the most code @xcite .
it should be noted that this version makes use of the nuclear ground state properties derived coherently from the same microscopic hfb method with the bsk7 skyrme force @xcite .
it also benefits from the improved nuclear level density prescription based on the microscopic statistical model , also used to estimate the nuclear temperature in eq .
( [ tdep ] ) @xcite .
the direct capture contribution as well as the possible overestimate of the statistical predictions for resonance - deficient nuclei are effects that could have an important impact on the radiative neutron captures by exotic nuclei @xcite , but are not included in the present study .
the maxwellian - averaged radiative neutron capture rate at a temperature @xmath62 k , typical of the r - process nucleosynthesis , obtained with the qrpa e1-strength are compared in fig .
[ fig_rate ] with those based on the hybrid lorentzian - type formula @xcite .
these rates are sensitive to the neutron capture cross section at incident energies around 130 kev , and therefore depend on the e1 strength in a narrow range of a few hundred kev around @xmath5 .
the temperature - dependent hybrid formula corresponds to a generalization of the energy- and temperature - dependent lorentzian formula including an improved description of the e1-strength function at energies below @xmath5 as derived from @xcite .
the hybrid e1 strength differs from the qrpa estimate not only in the location of the centroid energy , but also in the low - energy tail .
no extra low - lying strength is included in the phenomenological hybrid formula , but its temperature dependence increases the e1 strength at low energies and is responsible for its non - zero @xmath63 limit .
the newly - derived strength gives an increase of the rate by a factor up to 6 close to the neutron drip line .
r - process nuclei characterized by @xmath64 mev are seen to have a neutron capture rate about at least twice faster than the one predicted with the phenomenological hybrid formula .
this is due to the shift of the gdr to lower energies compared with the usually adopted liquid - drop @xmath61 rule , as well as to the appearance of some extra strength at low energies as explained above .
both effects tend to increase the e1 strength at energies below the gdr , i.e in the energy window of relevance in the neutron capture process . for less exotic nuclei ,
the qrpa impact is relatively small , differences being mainly due to the exact position of the gdr energy and the resulting low - energy tail .
when compared to our previous hfbcs+qrpa predictions @xcite , the hfb+qrpa model gives larger neutron capture rates close to the neutron drip line , but lower rates for many of the @xmath65 \lsimeq 2 $ ] nuclei , as seen in fig .
[ fig_rate ] ( lower panel ) .
these differences justify the use of the hfb approach for exotic neutron - rich nuclei .
the e1-strength function is estimated with one of the most accurate and reliable microscopic model available to date , namely the hartree - fock - bogoliubov ( hfb ) and qrpa models .
the spreading width of the gdr is taken into account by analogy with the srpa method .
the analysis of hfb+qrpa model based on various skyrme forces with different pairing prescriptions and parameterizations shows that the effective nucleon - nucleon interaction can be constrained with the gdr data .
in particular , it is found that the skyrme force characterized with a low effective mass @xmath66 is a necessary condition to reproduce the location and width of the gdr , at least within the present hfb+qrpa model to which the srpa is applied .
in contrast , gdr data can not be used to discriminate between the surface or volume property of the pairing interaction .
in addition to its reliability , it is shown that the hfb+qrpa model also gives accurate predictions and that globally it agrees fairly well with experimental data .
the present hfb+qrpa model brings important improvement with respect to our previous hfbcs+qrpa model and can provide quantitative predictions of the dipole strength .
large - scale calculations of the e1-strength function are performed and used to estimate the radiative neutron capture rates of relevance for the r - process nucleosynthesis .
a systematic increase of the reaction rates for exotic neutron - rich nuclei is found .
further improvements may be useful . a proper treatment of the continuum states and its impact on the dipole strength is an important issue .
it is expected to be significant for drip - line nuclei .
continuum - qrpa models are available @xcite and study along these lines are in progress .
the particle - vibration coupling also affects the low - energy strength and could contribute to an extra increase of the radiative neutron capture rate by exotic nuclei . *
acknowledgments * m.s . and s.g .
are fnrs research fellow and associate , respectively .
this work has been performed within the scientific collaboration ( tournesol ) between the wallonie
bruxelles community and france .
+ 100 s. goriely , phys .
lett . b436
( 1998 ) 10 .
mccullagh , m.l .
stelts , r.e .
chrien , phys .
c23 ( 1981 ) 1394 .
kadmenskii , v.p .
markushev , v.i .
furman , sov .
37 ( 1983 ) 165 .
j. kopecky , r. e. chrien , nucl . phys .
a468 ( 1987 ) 285 .
myers , w.j .
swiatecki , et al .
c15 ( 1977 ) 2032 .
f. catara , e.g. lanza , m.a nagarajan , a. vitturi , nucl .
a624 ( 1997 ) , 449 .
s. goriely , e. khan , nucl .
a706 ( 2002 ) 10 .
e. khan , nguyen van giai , phys .
b472 ( 2000 ) 253 .
j. dobaczewski , w. nazarewicz , t.r .
werner , j.f .
berger , c.r .
chinn , j. decharge , phys .
c53 ( 1996 ) 2809 .
e. chabanat , p. bonche , p. haensel , j. meyer , r. schaeffer , nucl .
a635 ( 1998 ) 231 .
m. grasso , n. sandulescu , nguyen van giai , r. j. liotta , phys .
c64 ( 2001 ) 064321 .
e. khan , n. sandulescu , nguyen van giai , m. grasso , phys .
c66 ( 2002 ) 024309 .
s. drod , s. nishizaki , j. speth , j. wambach , phys .
197 ( 1990 ) 1 .
dietrich , b. l. berman , at .
data nucl .
data tables 38 ( 1989 ) 199 .
photonuclear data for applications ; cross sections and spectra , iaea - tecdoc-1178 ( 2000 ) .
p. ring , p. schuck , _ the nuclear many - body problem , springer - verlag _ ( 1980 )
. m. samyn , s. goriely , p .- h .
heenen , j.m .
pearson , f. tondeur , nucl .
a700 ( 2002 ) 142 . s. goriely , m. samyn , p .- h .
heenen , j.m .
pearson , f. tondeur , phys .
c 66 ( 2002 ) 024326 .
m. samyn , s. goriely , j.m .
pearson , nucl .
phys . a ( 2003 ) submitted .
s. goriely , m. samyn , m. bender , j.m .
pearson , nucl .
a ( 2003 ) in preparation .
bckman , a.d .
jackson , j. speth , phys .
b56 ( 1975 ) 209 .
g. col , p.f .
bortignon , nucl .
a696 ( 2001 ) 427 .
p. schuck , s. ayik , nucl .
a687 ( 2001 ) 220c .
smith , j. wambach , phys .
c38 ( 1988 ) 100 .
baker , l. bimbot , c. djalali , et al .
289 ( 1997 ) 235 .
d. vautherin , j.d.m .
brink , phys .
c5 ( 1972 ) 626 .
nguyen van giai , h. sagawa , nucl .
a371 ( 1981 ) 1 .
j. kopecky , m. uhl , phys . rev .
c41 ( 1990 ) 1941 .
a. voinov , m. guttormsen , e. melby , j. rekstad , a. schiller , s. siem , phys .
c * 63 * , 044313 ( 2001 ) .
s. siem , m. guttormsen , k. ingeberg , e. melby , j. rekstad , a. schiller , a. voinov , phys .
c * 65 * , 044318 ( 2002 ) . p.f .
bortignon , nucl .
a687 ( 2001 ) 329c .
p. demetriou , s. goriely , nucl .
a695 ( 2001 ) 95 .
popov , in proc .
conf . , physics and application , vol .
p. oblozinsky ) , p. 121
reference input parameter library , iaea - tecdoc , in press ( 2003 ) . | large - scale qrpa calculations of the e1 strength are performed on top of hfb calculations in order to derive the radiative neutron capture cross sections for the whole nuclear chart .
the spreading width of the gdr is taken into account by analogy with the second - rpa ( srpa ) method . the accuracy of hfb+qrpa model based on various skyrme forces with different pairing prescription and parameterization is analyzed .
it is shown that the present model allows to constrain the effective nucleon - nucleon interaction with the gdr data and to provide quantitative predictions of dipole strengths .
16.2 cm -.54 cm -.54 cm nuclear reactions : qrpa , e1-strength , nuclear forces |
the aim of topological data analysis is to understand the shape of unstructured data often given as finitely many points in a metric space .
usually , the shape of such a point cloud is studied through a filtration of complexes built on given points .
for instance , the vietoris - rips complex contains edges , triangles , tetrahedra spanned by points whose pairwise distances are less than a certain scale .
the persistent homology of the resulting filtration over all scales captures topological features that persist over a long time interval .
if given points are densely sampled from a compact set in @xmath7 , the vietoris - rips complex at a certain scale correctly represents the topology of the set @xcite . for
a cloud of @xmath0 points , the vietoris - rips complex may contain up to @xmath8 simplices in dimension @xmath9 , so this large size is the main drawback .
don sheehy @xcite recently approximated the full filtration of vietoris - rips complexes on @xmath0 points in a metric space by a filtration that has a size @xmath10 and approximates the persistent homology with a multiplicative error close to @xmath11 .
the sheehy - vietoris - rips complex uses a net - tree @xcite as a black box .
if we run the best algorithm @xcite for persistent homology on the sheehy approximation to the vietoris - rips filtration , the overall running time for approximating persistent homology will be @xmath12 .
this overquadratic time is a bottleneck , but allows us to replace a sophisticated construction of a net - tree by a simpler algorithm for @xmath13-farthest neighbors in a metric space .
[ prm : approximate ] for a cloud of @xmath0 points in a metric space , approximate the persistent homology of the vietoris - rips filtration in a subquadratic time @xmath6 .
we solve problem [ prm : approximate ] in theorem [ thm : approximate ] by building a sublinear - size approximation to the vietoris - rips filtration on @xmath0 given points in a metric space and then running the best algorithm for computing the zigzag persistent homology .
due to stability of persistent homology @xcite , the error of approximation at the homology level can be controlled at the level of filtration .
[ thm : approximate ] the vietoris - rips filtration has a sublinear - size approximation that leads to a simple @xmath6 time algorithm for approximating persistent homology of the vietoris - rips filtration on @xmath0 points in a metric space .
the running time also depends on the error of approximation and on the doubling dimension of the metric space , see proposition [ pro : linear - size - zigzag ] .
our algorithm can improve the filtration on the fly without starting from scratch to get a smaller error of approximation and at a higher computational cost .
[ dfn : doubling ] in a metric space @xmath14 with a distance @xmath15 , the ( closed ) ball with a center @xmath16 and a radius @xmath17 is @xmath18 . the _ doubling constant _
@xmath19 of @xmath14 is the minimum number of balls of a radius @xmath20 that can cover any ball of radius @xmath21 .
the _ doubling dimension _ of @xmath14 is @xmath22 , so @xmath23 . if a metric space @xmath14 is finite , then the _ spread _ @xmath24 ( or the _ aspect ratio _ ) is the ratio of the largest to smallest interpoint distances @xmath25 over all distinct @xmath26 .
definition [ dfn : doubling ] implies that any subspace of a finite metric space @xmath14 with a doubling dimension @xmath27 has a doubling dimension at most @xmath27 .
[ lem : max - num - points ] @xcite any metric space @xmath14 with a doubling constant @xmath19 , a doubling dimension @xmath27 and a spread @xmath24 has at most @xmath28 points .
[ dfn : complex ] a _ simplicial complex _ is a finite set @xmath29 of vertices and a collection of subsets @xmath30 called _ simplices _ such that all subsets of a simplex are also simplices .
the dimension of a simplex @xmath31 is @xmath32 .
any simplex has the euclidean topology from this geometric realization : @xmath33 then we can define the topology on any simplicial complex by gluing all its simplices along their common subsimplices as in combinatorial definition [ dfn : complex ] .
[ dfn : vr - filtration ] for a finite metric space @xmath14 and any @xmath34 , the _ vietoris - rips _
complex @xmath35 consists of simplices whose all vertices have pairwise distances @xmath36 .
so we get the _ vietoris - rips filtration _ @xmath37 .
we consider the simplicial homology @xmath38 of a complex @xmath39 over a field @xcite .
[ dfn : zigzag - module ] @xcite a _ zigzag filtration _ @xmath40 is a sequence of simplicial complexes with inclusions @xmath41 , where each double arrow @xmath42 is either a forward inclusion @xmath43 or a backward inclusion @xmath44 .
any zigzag filtration gives rise to the _ persistence module _ of homology groups with homomorphisms @xmath45 , where @xmath46 denotes the vector space of homology groups @xmath47 in all dimensions @xmath48 .
[ dfn : pers - diagram ] @xcite the _ persistence diagram _ @xmath49 of a module @xmath50 is a set of pairs @xmath51 with multiplicities plus all diagonal points @xmath52 with the infinite multiplicity . each single pair @xmath53 represents a homology class that persists over the time interval @xmath54 in the persistence module @xmath50 , so a homology class is born at @xmath55 and dies at @xmath56 .
pairs @xmath51 in a persistence diagram away from the diagonal @xmath57 have a high persistence @xmath58 and represent topological features , not noise .
let the distance in @xmath59 be @xmath60 .
recall that the _ @xmath20-offset _ of a subset @xmath61 is @xmath62 for @xmath17 .
[ thm : pers - stability ] ( * ? ? ?
* theorem 4.4 ) let two non - zigzag modules @xmath63 and @xmath64 of vector spaces be _ interleaved _ such that there are homomorphisms @xmath65 and @xmath66 commuting with the inclusions @xmath67 and @xmath68 for all @xmath69 .
then their persistence diagrams are @xmath20-close in the sense that @xmath70 is in the @xmath20-offset of @xmath71 and @xmath71 is in the @xmath20-offset of @xmath70 .
theorem [ thm : pers - stability ] also works for a multiplicative error @xmath72 instead of @xmath73 .
[ thm : pers - computation ] @xcite if a zigzag filtration @xmath40 of complexes consists of @xmath2 updates ( additions or deletions of simplices ) , then the persistence diagram @xmath74 can be computed in @xmath1 time , where @xmath3 is the time needed for multiplication of two @xmath4 matrices .
[ dfn : contiguous ] a map between ( vertices of simplicial ) complexes @xmath75 is _ simplicial _ if any simplex maps to a simplex .
two maps @xmath76 are _ contiguous _ if @xmath77 is a simplex in @xmath78 for any simplex @xmath79 .
[ lem : contiguous ] ( * ? ? ?
* theorems 12.4 , 12.5 ) let @xmath80 be the inclusion of simplicial complexes .
if there is a simplicial retraction @xmath81 , i.e. @xmath82 , such that @xmath83 and @xmath84 are contiguous , then the inclusion @xmath85 induces the isomorphism @xmath86 between all homology groups .
[ dfn : farthest - first ] in a finite metric space @xmath14 , the distance from a point @xmath87 to a subset @xmath88 is @xmath89 .
choose any @xmath90 and enumerate all other @xmath91 according to the _ farthest - first traversal _ ( or a _ greedy permutation _ ) such that @xmath92 .
then set @xmath93 for @xmath94 and @xmath95 .
[ fig : k - farthest - neighbors ] [ lem : find - k - farthest ] @xcite in a metric space of @xmath0 points , starting from any @xmath90 , a simple algorithm finds @xmath13-farthest neighbors @xmath96 in @xmath97 time . for any @xmath98
, we keep the distance @xmath99 to already chosen points . in the @xmath85-th iteration
we choose @xmath100 , set @xmath101 .
each of @xmath13 iterations has @xmath10 time .
there is a much more sophisticated algorithm @xcite to compute a greedy permutation on any @xmath0-point metric space @xmath14 in @xmath102 time .
[ lem : k - clustering ] @xcite in a finite metric space @xmath14 , the gonzalez algorithm from lemma [ lem : find - k - farthest ] gives a 2-approximation to the np - hard @xmath13-center clustering problem to find points @xmath96 minimizing @xmath103 .
lemma [ lem : k - clustering ] means that a farthest - first traversal in any @xmath0-point metric space @xmath104 gives a good @xmath13-point approximation @xmath105 to @xmath106 in the sense that the error @xmath107 can be only twice larger than the optimal error .
[ pro : additive - error ] let @xmath108 be @xmath13-farthest neighbors in a finite metric space @xmath14
. then the persistence diagram @xmath109 approximates @xmath110 with the additive error @xmath111 . for any @xmath112 ,
let @xmath113 be the point closest to @xmath87
. then @xmath114 by definition [ dfn : farthest - first ] .
the projection @xmath115 induces the inclusion @xmath116 .
indeed , for any edge in @xmath35 between @xmath117 with @xmath36 , there is an edge between their projections @xmath118 in @xmath119 with @xmath120 by the triangle inequality .
the conclusion follows from theorem [ thm : pers - stability ] for the interleaved homology of the complexes @xmath121 . in a finite metric space @xmath14
we define a hierarchy of levels @xmath122 similar to nets @xmath123 in @xcite , but we use radii @xmath124 from a farthest - first traversal instead of a radii of subtrees in a more sophisticated net - tree .
[ dfn : levels ] fix a farthest - first traversal in a finite metric space @xmath14 , so each point @xmath112 has its radius @xmath124 .
the _ deletion time _ of @xmath112 is @xmath125 . for @xmath34 , the _ level _
is @xmath126 .
the level @xmath127 changes only at discrete values @xmath128 .
we may set @xmath129 for @xmath130 .
then the descending continuous filtration @xmath131 can be rewritten a discrete sequence @xmath132 .
lemma [ lem : cover - packing ] is similar to ( * ? ? ? * lemma 6.1 ) , where the packing property @xmath133 has a small packing constant @xmath134 hidden in a net - tree .
[ lem : cover - packing ] fix a farthest - first traversal in a finite metric space @xmath14 .
then the levels @xmath127 from definition [ dfn : levels ] satisfy the following properties .
@xmath135 the _ covering _ property : any point @xmath112 has @xmath136 .
@xmath135 the _ packing _ property : any distinct @xmath137 have @xmath138 .
_ let @xmath139 .
the next point is @xmath140 by definition [ dfn : farthest - first ] .
then @xmath141 .
the last inequality follows from definition [ dfn : levels ] and the fact that @xmath142 . hence @xmath136 as required in the covering property . to prove the packing property ,
assume that the point @xmath143 has a higher index than the point @xmath144 .
definitions [ dfn : farthest - first ] and [ dfn : levels ] imply that @xmath145 similarly to @xcite , we introduce the weight @xmath146 of a point @xmath112 using our @xmath147 and @xmath124 from a farthest - first traversal in @xmath14 .
[ dfn : weights ] the deletion time was introduced in definition [ dfn : levels ] for points in a finite metric space @xmath14 with a hierarchy of levels @xmath127 .
the _ weight _ is @xmath148 [ fig : weight - function ] [ dfn : relaxed - distance ] fix a farthest - first traversal in a metric space @xmath14 .
then the _ relaxed distance _ for @xmath117 is @xmath149 .
[ dfn : projections ] for each level @xmath127 from definition [ dfn : levels ] in a finite metric space @xmath14 , we define the projection @xmath150 by the formula @xmath151 lemma
[ lem : projection - properties ] is similar to ( * ? ? ?
* lemma 7.1 ) .
however , part ( i ) relies only on the levels @xmath127 without using parent links and radii of subtrees in a net - tree .
[ lem : projection - properties ] for a finite metric space @xmath14 with a projection @xmath150 from definition [ dfn : projections ] , any point @xmath112 has the following three properties .
\(i ) there is a point @xmath144 such that @xmath152 .
\(ii ) the projection @xmath153 satisfies @xmath154 .
\(iii ) the projection @xmath153 satisfies @xmath155 for any @xmath98 .
\(i ) if @xmath156 , then the inequality is trivial for @xmath157 . now we assume that @xmath158 , so @xmath159 by definition [ dfn : levels ] and @xmath160 by definition [ dfn : weights ] .
take the maximum set @xmath161 such that @xmath162 and @xmath163 . the point @xmath164 exists , because @xmath165 , so the radii eventually become smaller than @xmath166
let @xmath167 be the closest point from the set @xmath161 to the given point @xmath87 .
we estimate the corresponding distance as follows : @xmath168 since @xmath169 , then @xmath170 .
the inequality @xmath171 means that @xmath144 by definition [ dfn : levels ] .
the inequality @xmath172 gives @xmath173 by definition [ dfn : weights ] .
so @xmath174 .
parts ( ii ) and ( iii ) are proved in appendix similarly to ( * ? ? ?
* lemma 7.1 ) .
here we remind the results from @xcite , see proofs in the appendix .
[ lem : rvr - filtration ] for a finite metric space @xmath14 with the relaxed distance @xmath175 from definition [ dfn : weights ] , if @xmath176 for @xmath117 , then @xmath177 .
[ dfn : rvr - filtration ] for a finite metric space @xmath14 and @xmath34 , the _ relaxed vietoris - rips complex _ @xmath178 consists of simplices whose vertices have pairwise relaxed distances @xmath179 .
lemma [ lem : rvr - filtration ] implies that @xmath180 for @xmath69 .
so we get the _ relaxed vietoris - rips filtration _ @xmath181 .
[ lem : rvr - approximation ] for a finite metric space @xmath14 with any @xmath34 and @xmath182 , we have @xmath183 .
hence the persistence diagram @xmath184 approximates @xmath185 with the multiplicative error @xmath186 .
we remind the sheehy approximation to the vietoris - rips complex @xcite and introduce a similar smaller complex for a @xmath13-point subspace .
[ dfn : svr - complex ] ( i ) for a finite metric space @xmath14 and @xmath182 , the nets @xmath123 @xcite are defined as @xmath127 in definition [ dfn : levels ] using radii @xmath124 of a net - tree .
sheehy - vietoris - rips complex _ is @xmath187 .
\(ii ) let @xmath188 be the first @xmath13 points in a farthest - first traversal of @xmath106 , see definition [ dfn : farthest - first ] .
we define the levels @xmath127 on the subspace @xmath189 and introduce another sparse vietoris - rips complex @xmath190 .
if we consider levels @xmath127 on the whole space @xmath106 , then @xmath191 and @xmath192 differ only by the definition of @xmath124 .
the sheehy - vietoris - rips complex @xmath191 uses an implicit radius of a subtree in a net - tree .
the new complex @xmath192 uses the radii from a farthest - first traversal in definition [ dfn : farthest - first ] .
since our aim is to trade the size for the error of approximation , we build the smaller complex @xmath192 on a @xmath13-point subspace of the original @xmath0-point space @xmath106 .
lemma [ lem : isomorph - homology ] is based on lemma [ lem : projection - properties ] very similarly to ( * ? ? ?
* lemma 7.3 ) .
[ lem : isomorph - homology ] let @xmath188 be first @xmath13 points in a farthest - first traversal of a finite metric space @xmath14 , so the levels @xmath127 can be defined on the subspace @xmath193 as in definition [ dfn : levels ] . then , for any @xmath34 , the inclusion @xmath194 induces an isomorphism in homology . the projection @xmath195 induces the simplicial retraction @xmath196 .
indeed , all points of @xmath197 remain fixed . moreover , any edge between @xmath198 in @xmath199 with @xmath179 maps to the edge between @xmath200 in @xmath201 with @xmath202 by lemma [ lem : projection - properties](iii ) .
to apply lemma [ lem : contiguous ] about an induced isomorphism in homology , it remains to show that @xmath203 is contiguous to @xmath204 on @xmath199 .
actually , for any edge between @xmath198 with @xmath179 , the tetrahedron on @xmath205 belongs to @xmath199 , because all 6 edges have @xmath206 .
namely , use lemma [ lem : projection - properties](ii ) for @xmath207 and @xmath208 , then [ lem : projection - properties](iii ) for other 4 edges .
[ pro : error - approx ] let @xmath108 be @xmath13-farthest neighbors in a finite metric space @xmath14 . then the persistence diagram @xmath209 approximates @xmath110 with the additive error @xmath210 . by lemma [ lem : isomorph - homology ] the persistence diagram @xmath209 coincides with @xmath211 , which approximates @xmath109 with the multiplicative error @xmath186 by lemma [ lem : rvr - approximation ] .
since @xmath109 approximates @xmath110 with the additive error @xmath111 by proposition [ pro : additive - error ] , the overall additive error of approximating @xmath110 is @xmath210 .
recall that a first point @xmath212 from a farthest - first traversal in a finite metric space is usually chosen arbitrarily and we set @xmath95 in definition [ dfn : farthest - first ] .
[ dfn : critical - alpha ] let @xmath108 be @xmath13-farthest neighbors in a finite metric space @xmath14 .
the _ critical values _ of @xmath213 are @xmath214 , @xmath215 , and all values satisfying @xmath216 , @xmath217 .
[ lem : sparse - zigzag ] if @xmath127 are levels on @xmath13-farthest neighbors @xmath193 , then the sparse vietoris - rips complex @xmath218 changes only at the critical values from definition [ dfn : critical - alpha ] and becomes smaller only at @xmath219 . over each interval @xmath220 , the complex
@xmath218 has the same vertices @xmath221 . at @xmath222
we denote the complex @xmath218 by @xmath223
. then edges between @xmath198 ( possibly simplices if cliques are formed ) can be only added at critical values @xmath224 .
if the open interval @xmath225 contains @xmath226 critical values @xmath224 , then we get an ascending filtration of complexes @xmath227 .
at @xmath219 when we update @xmath228 to @xmath229 , the complex loses its vertex @xmath230 ( with all incident simplices ) and then starts to grow again until @xmath231 is lost at @xmath232 and so on .
finally , @xmath233 may have only a single edge between @xmath234 at @xmath235 , then @xmath236 at @xmath237 .
so we get the full zigzag filtration @xmath238 from the descending filtration @xmath239 after replacing each @xmath240 by the ascending subfiltration @xmath241 .
lemma [ lem : higher - neighbors ] is similar to ( * ? ? ?
* lemma 9.2 ) and is based on lemma [ lem : max - num - points ] .
[ lem : higher - neighbors ] let @xmath108 be @xmath13-farthest neighbors in a finite metric space @xmath14 . for any @xmath242 , @xmath243 , let @xmath244 be the set of vertices that are connected by an edge with @xmath230 in @xmath218 for @xmath213 just below @xmath245 , @xmath246 then @xmath247 , where @xmath27 is the doubling dimension of @xmath14 .
the set @xmath248 contains points @xmath249 only with smaller indices , hence @xmath250 for @xmath219 .
the minimum distance between any points in @xmath127 is at least @xmath251 by the packing property in lemma [ lem : cover - packing ] . for any point @xmath252
, we estimate : @xmath253 then the distance between any two @xmath254 is at most @xmath255 .
then the spread ( ratio of the largest to smallest interpoint distances ) of @xmath248 is at most @xmath256 .
so @xmath257 by lemma [ lem : max - num - points ] .
proposition [ pro : linear - size - zigzag ] is similar to ( * ? ? ? * theorem 9.3 ) and is based on lemma [ lem : sparse - zigzag ] with an explicit description of the discrete zigzag filtration @xmath238 .
[ pro : linear - size - zigzag ] the zigzag filtration @xmath238 built ( as in lemma [ lem : sparse - zigzag ] ) on @xmath13-farthest neighbors in a finite metric space @xmath14 of a doubling dimension @xmath27 has the total number of updates @xmath258 up to dimension @xmath9 . at
@xmath219 the complex @xmath218 loses @xmath230 with all links ( simplices ) to its neighbors in @xmath248 .
it is essential that later we will not add any of these deleted simplices .
hence each of @xmath13 vertices @xmath230 generates the number of updates equal to the doubled number @xmath259 of all deleted simplices up to dimension @xmath9 .
it remains to use @xmath247 from lemma [ lem : higher - neighbors ] .
* proof of theorem [ thm : approximate ] . * starting from any @xmath90 , we build @xmath13-farthest neighbors @xmath188 by the simple @xmath97-algorithm in lemma [ lem : find - k - farthest ] .
so we have computed all @xmath260 , @xmath215 giving critical values @xmath219 by definition [ dfn : critical - alpha ] .
the final additive error of approximation for persistent homology will be @xmath210 by proposition [ pro : error - approx ] .
if we need a smaller error of approximation , then at this moment we may decide to continue a farthest - first traversal ( without starting from scratch ) to get a smaller @xmath107 . for each pair
@xmath261 , @xmath262 , we solve the linear equations @xmath263 to find @xmath264 more critical values of @xmath213 .
we sort all @xmath264 critical values in @xmath265 time .
each critical value is linked to the corresponding point @xmath230 or pair @xmath261 . for the smallest critical value @xmath213 , the initial complex @xmath218 consists of @xmath13 isolated points @xmath188 .
each complex @xmath218 will be represented by a list of maximal ( by inclusion ) simplices . in the sparse filtration @xmath266
, these maximal simplices correspond to cliques ( sets of pairwise connected vertices ) .
for each point @xmath230 , @xmath243 , we will maintain the list @xmath267 of subsets of @xmath268 that form with @xmath230 a maximal simplex in @xmath218 .
hence all vertices from @xmath267 are in the set @xmath248 from lemma [ lem : higher - neighbors ]
. then the list @xmath267 has at most @xmath269 simplices up to dimension @xmath9 by lemma [ lem : max - num - points ] .
we will go through all critical values in the increasing order updating @xmath218 and @xmath267 .
* case 1 : * a value @xmath213 is critical due to a pair @xmath261 , @xmath270 .
the edge between @xmath271 enters the complex @xmath218 by definition [ dfn : svr - complex](ii ) .
we need to add simplices on any new cliques containing the edge @xmath261 .
we go through each maximal simplex @xmath272 and check if all vertices of @xmath273 have edges to @xmath249 , which takes @xmath274 time for simplices up to dimension @xmath9 .
if yes , we replace the old maximal simplex @xmath273 by the larger simplex @xmath275 in @xmath267 , then we add the new maximal simplex @xmath276 to @xmath218 .
* case 2 : * a value @xmath213 is critical due to @xmath277 .
we remove from @xmath218 the vertex @xmath230 and all maximal simplices in the current list @xmath267 .
there are no other simplices containing @xmath230 since only the points @xmath278 survive .
the running time in cases 1 and 2 above for each of @xmath264 critical values does nt depend on @xmath13 and the overall time is @xmath279 so far .
finally , the algorithm of @xcite computes the persistent homology of the sparse vietoris - rips filtration @xmath266 in time @xmath280 . here
@xmath3 and @xmath258 is the number of updates in the sparse zigzag filtration @xmath192 up to dimension @xmath9 by proposition [ pro : linear - size - zigzag ] .
so the simple algorithm based on a farthest - first traversal approximates the persistent homology in time @xmath281 .
if we choose @xmath282 , the running time is subquadratic as required : @xmath283 .
the sophisticated construction of a full farthest - first traversal by har - peled and mendel @xcite takes only @xmath102 time , where @xmath19 is the doubling constant of the metric space . it might be possible to adapt their algorithm for only @xmath13-farthest neighbors to get the overall time @xmath284 for @xmath285
we are open to collaboration on any related problems .
we thank any referees in advance for critical comments and suggestions .
references d. attali , a. lieutier , d. salinas , vietoris - rips complexes also provide topologically correct reconstructions of sampled shapes . _ computational geometry _ 46 ( 2013 ) 448 - 465 .
g. carlsson , v. de silva .
zigzag persistence .
_ 10 : 367 - 405 , 2010 .
f. chazal , d. cohen - steiner , m. glisse , l. j. guibas , s. y. oudot .
proximity of persistence modules and their diagrams . in _ proceedings of socg 2009 _ , 237 - 246 .
c. chen , m. kerber .
an output - sensitive algorithm for persistent homology . _ computational geometry _ 46 ( 2013 ) 435 - 447 .
h. edelsbrunner , j. harer .
computational topology .
an introduction , ams , providence , rhode island , 2010 .
t. f. gonzalez .
clustering to minimize the maximum intercluster distance .
_ theoretical computer science _ 38 : 293 - 306 , 1985 .
n. milosavljevic , d. morozov , p. skraba .
zigzag persistent homology in matrix multiplication time . in _ proceedings of socg
2011 _ , p. 216 - 225 .
j. munkres , elements of algebraic topology , perseus , cambridge , massachusets , usa , 1984 .
s. har - peled , m. mendel .
fast construction of nets of low dimensional metrics , and their applications .
_ siam journal on computing _
35(5 ) : 1148 - 1184 , 2006 .
d. sheehy , linear - size approximations to the vietoris - rips filtration , _ discrete comput .
geometry _ , 49(4 ) : 778 - 796 , 2013 .
* proofs of parts ( ii ) and ( iii ) in lemma [ lem : projection - properties ] . *
+ ( ii ) we use @xmath286 from part ( i ) and definitions [ dfn : relaxed - distance ] , [ dfn : projections ] : @xmath287 @xmath288 @xmath289 * proof of lemma [ lem : rvr - filtration ] . *
+ the highest gradient of the piecewise linear function @xmath146 is @xmath292 . hence if @xmath69 then @xmath293 and similarly @xmath294 .
the relaxed distance by definition [ dfn : relaxed - distance ] is @xmath295 @xmath296 * proof of lemma [ lem : rvr - approximation ] . *
+ if @xmath297 for weights @xmath298 , then @xmath36 , so we get the inclusion @xmath299 . to prove the inclusion @xmath300 , we notice that @xmath301 . hence if @xmath302 , then @xmath303 . | the vietoris - rips filtration for an @xmath0-point metric space is a sequence of large simplicial complexes adding a topological structure to the otherwise disconnected space .
the persistent homology is a key tool in topological data analysis and studies topological features of data that persist over many scales .
the fastest algorithm for computing persistent homology of a filtration has time @xmath1 , where @xmath2 is the number of updates ( additions or deletions of simplices ) , @xmath3 is the time for multiplication of @xmath4 matrices . for a space of @xmath0 points given by their pairwise distances , we approximate the vietoris - rips filtration by a zigzag filtration consisting of @xmath5 updates , which is sublinear in @xmath0 .
the constant depends on a given error of approximation and on the doubling dimension of the metric space .
then the persistent homology of this sublinear - size filtration can be computed in time @xmath6 , which is subquadratic in @xmath0 . |
in order to construct @xmath1-dimensional homotopy quantum field theories ( hqft s ) , the second author introduced for a group @xmath0 the notions of a ribbon @xmath0-category and a modular ( ribbon ) @xmath0-category . the aim of this paper is to analyze the categories of representations of quantum groups ( at roots of unity ) from this prospective .
the role of @xmath0 will be played by the center of the corresponding lie algebra .
we begin with a general theory of ribbon @xmath0-categories with abelian @xmath0 .
according to @xcite , any modular ( ribbon ) @xmath0-category @xmath8 gives rise to a 3-dimensional hqft with target @xmath2 .
such an hqft comprises several ingredients including a homotopy modular functor " assigning modules to pairs ( a surface @xmath9 , an element of @xmath10 ) and a numerical invariant of pairs ( a closed oriented 3-manifold @xmath6 , a cohomology class @xmath11 ) .
we introduce here a larger class of weakly non - degenerate premodular @xmath0-categories .
similarly to @xcite , each such category @xmath8 gives rise to a numerical invariant @xmath12 of pairs @xmath13 as above .
if @xmath14 satisfies an additional assumption of regularity and @xmath0 is finite then the standard witten - reshetikhin - turaev invariant @xmath15 is also defined and splits as @xmath16 where @xmath17 is the first betti number of @xmath6 .
note that the invariant @xmath12 extends to an hqft provided @xmath8 is a modular @xmath0-category .
let @xmath3 be a simple complex lie algebra and @xmath5 be a complex root of unity . we show that under certain assumptions on the order of @xmath5 , the pair @xmath18 gives rise to a premodular ribbon @xmath0-category @xmath19 where @xmath20 is the center of @xmath3 .
( for the center groups of simple lie algebras , see table 1 below ) .
the definition of @xmath8 is based on a study of the representations of the quantum group @xmath4 , cf .
we specify conditions on @xmath5 which ensure that @xmath8 is regular so that we have numerical invariants of 1-cohomology classes and a splitting of the standard wrt - invariant as above .
another set of conditions ensures that @xmath21 is a modular @xmath0-category .
the resulting 3-dimensional hqft is however not very interesting since it splits as a product of a standard tqft and a homological hqft ( cf .
remark [ quq ] ) .
the paper consists of three sections . in sect . 1 we discuss the theory of premodular @xmath0-categories with abelian @xmath0 .
in sect . 2 we recall the definition of @xmath12 and briefly discuss the homotopy modular functor . in sect .
3 we consider the category @xmath22 .
the first author would like to thank a. beliakova , c. blanchet , a. bruguieres , g. masbaum for helpful discussions .
throughout this section , @xmath0 denotes an abelian group and @xmath23 a field of characteristic zero . we shall use the standard notions of the theory of monoidal categories , see @xcite .
recall that a left duality in a monoidal category @xmath8 associates to any object @xmath24 an object @xmath25 and two morphisms @xmath26 and @xmath27 satisfying the identities @xmath28 here @xmath29 denotes the unit object of @xmath30 and for simplicity we omit the associativity isomorphisms and the canonical isomorphisms @xmath31 .
an object of @xmath30 isomorphic to @xmath29 is said to be _
trivial_. a monoidal category @xmath14 is _ @xmath23-additive _ if all the @xmath32 in @xmath14 are @xmath23-modules and both the composition and the tensor product of morphisms are bilinear over @xmath23 .
we say that a @xmath23-additive monoidal category @xmath14 _ splits as a disjoint union of subcategories _
@xmath33 numerated by certain @xmath34 if : - each @xmath35 is a full subcategory of @xmath14 ; - each object of @xmath14 belongs to @xmath35 for a unique @xmath34 ; - if @xmath36 and @xmath37 with @xmath38 then @xmath39 .
a _ monoidal @xmath0-category over @xmath23 _ is a @xmath23-additive monoidal category with left duality @xmath14 which splits as a disjoint union of subcategories @xmath40 numerated by @xmath41 such that \(i ) if @xmath36 and @xmath37 then @xmath42 ; \(ii ) if @xmath36 then @xmath43 . we shall write @xmath44 and call the subcategories @xmath45 of @xmath14 the _ components _ of @xmath14 .
the category @xmath46 corresponding to the neutral element @xmath47 is called the _ neutral component _ of @xmath14 . conditions ( i ) and ( ii ) show that @xmath46 is closed under tensor multiplication and taking the dual object . condition ( i ) implies that @xmath48 .
thus , @xmath46 is a monoidal category with left duality .
the standard notions of braidings and twists in monoidal categories apply in this setting without any changes . a braiding ( resp .
twist ) in a monoidal @xmath0-category @xmath14 is a system of invertible morphisms @xmath49 ( resp .
@xmath50 ) satisfying the usual conditions , see @xcite .
we say that a monoidal @xmath0-category is _ ribbon _ if it is ribbon in the sense of @xcite , i.e. , if it has braiding and twist compatible with each other and with duality .
the standard theory of ribbon categories applies to any ribbon @xmath0-category @xmath8 .
suppose that @xmath51 is a framed oriented @xmath52-component link in @xmath53 whose components are ordered , and @xmath54 are @xmath23-linear combinations of objects of @xmath55 .
then there is defined the quantum reshetikhin - turaev invariant @xmath56 . in particular , for any object @xmath57 , we have a dimension @xmath58 where @xmath59 is an oriented unknot with framing 0 . for any endomorphism @xmath60
, we have a well - defined trace @xmath61 so that @xmath62 .
let @xmath14 be a ribbon @xmath0-category .
an object @xmath63 of @xmath14 is _ simple _ if @xmath64 .
it is clear that an object isomorphic or dual to a simple object is itself simple .
the assumption that @xmath23 is a field and ( * ? ? ?
* lemma ii.4.2.3 ) imply that any non - zero morphism between simple objects is an isomorphism .
we say that an object @xmath63 of @xmath14 is _ dominated by simple objects _ if there is a finite set of simple objects @xmath65 of @xmath14 ( possibly with repetitions ) and morphisms @xmath66 such that @xmath67 . clearly , if @xmath36 then without loss of generality we can assume that @xmath68 for all @xmath69 .
we say that a ribbon @xmath0-category @xmath14 is _ premodular _ if it satisfies the following three axioms : ( 1.3.1 ) the unit object @xmath70 is simple ; ( 1.3.2 ) for each @xmath71 , the set @xmath72 of the isomorphism classes of simple objects of @xmath35 is finite ; ( 1.3.3 ) for each @xmath71 , any object of @xmath35 is dominated by simple objects of @xmath35 .
according to ( * ? ? ?
* remark 7.8 ) the set @xmath73 is a subgroup of @xmath0 .
the theory certainly reduces to @xmath74 . for simplicity
, we will assume from now on that @xmath75 .
recall that @xmath59 is an unknot with framing @xmath76 .
let @xmath77 be an unknot with framing @xmath78 .
for each @xmath71 , consider the formal linear combination @xmath79 where @xmath80 is a simple object in @xmath35 representing @xmath81 .
define @xmath82 @xmath83 where the twist @xmath84 acts on @xmath80 as the scalar operator @xmath85 .
we recall here the following result ( * ? ? ?
* lemma 6.6.1 ) . if @xmath86 , then @xmath87 .
[ product ] the equality here is understood as an equality in the verlinde algebra of @xmath8 .
applying to both sides the @xmath23-linear homomorphism @xmath88 from the verlinde algebra to @xmath23 , sending the class of any object to its dimension , we obtain that @xmath89 . taking as @xmath63 a simple object in @xmath90 and using the fact that @xmath91 ( see ( * ? ? ?
* lemma 6.5 ) ) we obtain the following corollary . for every @xmath92 , @xmath93 [ equal ]
one has that @xmath94 .
[ product2 ] this follows from lemma [ product ] and corollary [ equal ] .
an important property of any premodular category is the sliding property .
here is the version for premodular g - category .
( graded sliding property ) let @xmath8 be a premodular @xmath0-category .
let @xmath95 be framed oriented ordered links in @xmath53 such that @xmath96 is obtained from @xmath51 by sliding the second component over the first one ( see figure [ sliding ] ) .
then for every @xmath97 , one has @xmath98
@xmath99 the proof is the same as in the case of a premodular category , see for example @xcite .
one has to take into account the colors of the components of tensor product decomposition .
this proposition extends to tangles in the obvious way .
one has to be a bit careful about colorings of a tangle .
any non - circle component must be colored with an object of @xmath8 , not a @xmath23-linear combination of objects as for circle components .
suppose that @xmath100 is a set of objects of a ribbon @xmath0-category @xmath8 .
an object @xmath101 is _ @xmath102-transparent _ if for every @xmath103 , @xmath104 this means that one can always move a string colored by @xmath63 past a string colored by @xmath105 , see figure [ transparent ] .
@xmath106 let @xmath107 denote the set of all @xmath108-transparent simple objects of @xmath8 .
let @xmath63 be a simple object of a premodular @xmath0-category @xmath8 with @xmath109 .
the operator of the tangle in figure [ s]a is non - zero for some @xmath71 if and only if @xmath110 , i.e. , @xmath63 is @xmath108-transparent .
[ trans3 ] @xmath111 since @xmath63 is simple , the operator in figure [ s ] is equal to a scalar operator @xmath112 with @xmath113 .
suppose @xmath114 .
let @xmath115 .
figure [ transparent2 ] shows that @xmath110 ; the second equality uses the graded sliding property .
( this argument was first used by blanchet and beliakova in @xcite . )
@xmath116 now suppose @xmath117 .
it is clear that @xmath118 . by corollary [ product2 ]
, one has @xmath119 .
the operator of the tangle in figure [ s]b is @xmath120 .
it is also equal to ( by combining the two parallel components ) @xmath121 .
thus @xmath122 , and @xmath123 . under the assumption of lemma [ trans3 ] , if @xmath63 is @xmath46-transparent , then @xmath63 is almost " transparent with respect to any object in @xmath8 .
more precisely , the same argument as in the proof of lemma [ trans3 ] shows that for any @xmath124 , one has @xmath125 a premodular @xmath0-category @xmath14 is _ weakly non - degenerate _ if @xmath126 .
it is known ( see @xcite ) that weak non - degeneracy implies @xmath109 .
a premodular @xmath0-category @xmath14 is _ regular _ if it is weakly non - degenerate and @xmath127 .
if @xmath8 is a regular premodular @xmath0-category and @xmath128 then @xmath129 [ vanish ] consider the equality in figure [ proof4 ] which is obtained by sliding the top component of the left hand side over the bottom one .
@xmath130 by the regularity of @xmath14 , there are no @xmath108-transparent simple objects in @xmath35 .
hence , by lemma [ trans3 ] , the right hand side on figure [ proof4 ] is @xmath76 .
it follows that @xmath131 , since @xmath132 .
similarly @xmath133 .
let @xmath8 be a premodular @xmath0-category . for @xmath134
, choose simple objects @xmath135 representing @xmath136 , respectively , and set @xmath137 here @xmath138 is the standard hopf link in @xmath53 with framing @xmath76 .
it follows from the properties of the trace that @xmath139 does not depend on the choice of @xmath140 in the isomorphism classes @xmath136 .
a premodular @xmath0-category @xmath8 is _ modular _ if the following axiom is satisfied .
( _ non - degeneracy axiom _ ) the ( finite ) square matrix @xmath141_{i , j\in i_0 } $ ] is invertible over @xmath23 .
it follows from this axiom that the neutral component @xmath46 of @xmath14 is a modular ribbon category in the usual , ungraded sense ( see @xcite ) .
it is known that the non - degeneracy implies the weak non - degeneracy .
a modular ribbon @xmath0-category @xmath8 may be non - modular in the ungraded sense because the set @xmath142 of the isomorphism classes of simple objects in @xmath8 may be infinite , or because the full @xmath143-matrix @xmath141_{i , j\in i } $ ] may be non - invertible .
suppose that @xmath0 is finite and a modular g - category @xmath55 is regular .
then @xmath8 is a modular category in the ungraded sense .
[ ttt ] according to @xcite , @xcite , a premodular category @xmath8 is modular if and only if it has no non - trivial @xmath8-transparent simple objects and @xmath144 . if @xmath63 is a @xmath8-transparent simple object of @xmath8 then @xmath145 .
but since @xmath46 is modular , @xmath63 is a trivial object . by corollary [ equal ] ,
@xmath146 , since @xmath46 is modular .
* 2.1 . invariants of 3-dimensional @xmath0-manifolds .
* fix an abelian group @xmath0 .
let @xmath21 be a weakly non - degenerate premodular ribbon @xmath0-category over a field of zero characteristic @xmath23 .
we explain here following @xcite that @xmath21 gives rise to a topological invariant of 1-dimensional cohomology classes of 3-manifolds with coefficients in @xmath0 .
fix an element @xmath147 such that @xmath148 .
let @xmath6 be a closed connected oriented 3-dimensional manifold and @xmath149 .
present @xmath6 as the result of surgery on @xmath53 along a framed oriented link @xmath150 . recall that @xmath6 is obtained by gluing @xmath52 solid tori to the exterior of @xmath51 in @xmath53
this allows us to consider for @xmath151 , the value , @xmath152 , of @xmath153 on the meridian of @xmath154 and provide @xmath154 with the color @xmath155 .
let @xmath156 ( resp .
@xmath157 ) be the number of positive ( resp .
negative ) squares in the diagonal decomposition of the intersection form @xmath158 where @xmath159 is the compact oriented 4-manifold bounded by @xmath6 and obtained from the 4-ball @xmath160 by attaching 2-handles along tubular neighborhoods of the components of @xmath51 in @xmath161 . set @xmath162 where @xmath163 is the first betti number of @xmath6 .
it follows from ( * ? ? ?
* theorem 7.3 ) that @xmath164 is a homeomorphism invariant of the pair @xmath165 ( in @xcite this is stated for modular @xmath0-categories , but the proof remains true for weakly non - degenerate premodular categories ) .
the factor @xmath166 appears here for normalization purposes . as in the standard theory ,
the invariant @xmath167 generalizes to an invariant of triples @xmath168 where @xmath6 is as above , @xmath169 is a colored ribbon graph in @xmath6 and @xmath170 . here
a coloring is understood as the usual coloring of @xmath169 over @xmath8 such that the color of every 1-stratum @xmath171 of @xmath169 is an object in @xmath172 where @xmath173 is the value of @xmath153 on the meridian of @xmath171 .
this notion of a coloring applies in particular to framed oriented links in @xmath6 so that we obtain an isotopy invariant of triples @xmath174 where @xmath6 is as above , @xmath51 is a colored framed oriented link in @xmath6 and @xmath175 .
suppose now that @xmath0 is finite and @xmath8 is regular .
set @xmath176 . by lemma
[ vanish ] , @xmath177 we can therefore consider the standard witten - reshetikhin - turaev invariant @xmath178 of @xmath6 where @xmath179 . if @xmath0 is finite and @xmath8 is a regular premodular @xmath0-category , then for any closed connected oriented 3-manifold @xmath6 , @xmath180 the proof in @xcite can be applied to get the result . a modular @xmath0-category @xmath8 ( not necessarily regular )
gives rise to a @xmath1-dimensional hqft , see @xcite .
in particular , the homotopy modular functor @xmath181 associated with @xmath55 assigns @xmath23-modules to so - called extended @xmath0-surfaces and assigns @xmath23-linear isomorphisms of these modules to weak homeomorphisms of such surfaces .
we recall here the definition of an extended @xmath0-surface and the corresponding @xmath23-module in our abelian case .
let @xmath182 be a closed oriented surface .
a point @xmath183 is _ marked _ if is equipped with a sign @xmath184 and a tangent direction , i.e. , a ray @xmath185 where @xmath186 is a non - zero tangent vector at @xmath187 .
a _ marking _ of @xmath182 is a finite ( possibly void ) set of distinct marked points @xmath188 .
@xmath0-marking _ of @xmath182 is a marking @xmath188 endowed with a cohomology class @xmath189
. a @xmath0-marking @xmath188 is _ colored _ if it is equipped with a function @xmath190 which assigns to every point @xmath191 an object @xmath192 where @xmath193 is a small loop in @xmath194 encircling @xmath187 in the direction induced by the orientation of @xmath182 if @xmath195 and in the opposite direction otherwise .
an _ extended @xmath0-surface _ comprises a closed oriented surface @xmath182 , a colored @xmath0-marking @xmath188 , and a lagrangian space @xmath196 .
the corresponding @xmath23-module is defined as follows .
assume that @xmath182 is a connected surface of genus @xmath197 and @xmath198 .
for @xmath199 , set @xmath200 and @xmath201 .
the group @xmath202 is generated by the homotopy classes of loops @xmath203 homotopic to @xmath204 , respectively , and by @xmath205 elements @xmath206 subject to the only relation @xmath207\ , ... \,[a_n , b_n]= 1\ ] ] where @xmath208= ab a^{-1}b^{-1}$ ] .
then @xmath209 @xmath210 where for an object @xmath211 we set @xmath212 and @xmath213 .
observe that the dimension of @xmath214 over @xmath23 does not depend on the values of @xmath153 on @xmath215 .
this implies that this dimension does not depend on the choice of @xmath153 . for @xmath216 , this dimension
is computed by the verlinde formula . in particular ,
if @xmath217 we obtain @xmath218 the same formula can be deduced from the generalised verlinde formula for hqft s given in ( * ? ? ?
4.11 ) . since we restrict ourselves here to abelian @xmath0 , the homotopy classes of maps from a manifold @xmath6 to the eilenberg - maclane space @xmath219 are classified by elements of @xmath220
this allows one to formulate hqft s with target @xmath219 in terms of 1-dimensional cohomology classes .
we call the hqft s arising from modular @xmath0-categories with abelian @xmath0 and trivial crossed structure ( as everywhere in this paper ) _ abelian_. the abelian hqft s are simpler than the general hqft s
. one of simplifications is that in the abelian case we can forget about base points ( in the general case one has to work in the pointed category ) .
also , in the abelian case the canonical action of @xmath0 on the spaces of conformal blocks @xmath221 defined in ( * ? ? ?
10.3 ) is trivial .
let @xmath222 be a finite - dimensional simple lie algebra over @xmath223 . fix a cartan subalgebra and a system of basis roots @xmath224 .
choose the inner product on @xmath225-span@xmath226 such that the square length of any _ short _ root is 2 .
let @xmath227 be the quotient of the weight lattice @xmath228 by the root lattice @xmath229 .
it is known that @xmath0 is isomorphic to the center of the simply - connected lie group associated with @xmath222 , and @xmath230 is equal to the determinant of the cartan matrix .
let @xmath231 be the maximal absolute value of the non - diagonal entries of the cartan matrix .
let @xmath232 denote the coxeter number of @xmath3 and @xmath233 the dual coxeter number of @xmath3 .
by @xmath234 we denote the smallest positive integer such that @xmath235 for all @xmath236 .
the data associated with simple lie algebras is given in table 1 .
the quantum group @xmath237 , as defined in @xcite , is a hopf algebra over @xmath238 , with @xmath186 a formal variable .
there is an integral version of @xmath239 , defined over the ring @xmath240 $ ] , introduced by lusztig . for @xmath241 , let @xmath242 be the hopf algebra over @xmath223 obtained by tensoring the integral version of @xmath237 with @xmath223 , where @xmath243 is considered as a @xmath240$]-module by @xmath244 ( see @xcite ) .
let @xmath5 be a root of unity .
the fusion category @xmath245 is the quotient of the category of all tilting @xmath242-modules by negligible modules and negligible morphisms ( see @xcite , there @xmath245 is denoted by @xmath246 ) .
the category @xmath245 is a semisimple monoidal @xmath223-abelian category with duality .
the isomorphism classes of simple objects in @xmath245 are parametrized by the dominant weights @xmath247 such that @xmath248 where @xmath249 is the half - sum of all positive roots and @xmath250 is defined as follows .
there are two cases depending on the order @xmath69 of the root of unity @xmath251 .
case 1 : the numbers @xmath69 and @xmath231 are co - prime and @xmath252 .
then @xmath253 where @xmath254 is the weyl chamber and @xmath255 is the short highest root , i.e. , the only root in @xmath254 with square length 2 . for a dominant weight @xmath247 such that @xmath248
there is a simple @xmath242-module @xmath263 , known as the weyl module , which is a deformation of the corresponding classical @xmath3-module .
it represents the isomorphism class numerated by @xmath247 .
the decomposition of a tensor product @xmath264 is described in @xcite ( the tensor product is denoted there by , but we will use the usual notation ) . for our purposes , we notice that if @xmath265 appears as a summand in @xmath266 , then @xmath267 for some @xmath268 where @xmath269 is the group of affine transformations of @xmath270 generated by the reflections in the walls of the simplex @xmath271 , the topological closure of @xmath250 , and the dot action is the one shifted by @xmath249 so that @xmath272 .
note that @xmath273 is a fundamental domain of @xmath269 .
for further use , we state here a simple lemma .
it is known that @xmath277 , where @xmath105 is the usual weyl group , and @xmath278 is the @xmath279-lattice spanned by the long roots .
( one considers @xmath278 as a group of translations ) .
obviously , @xmath280 . if @xmath281 , the statement is trivial . if @xmath282 , the statement is well - known in the theory of simple lie algebras .
there are a braiding and a twist in @xmath245 which make @xmath245 a ribbon category .
they depend on a choice of a complex root of unity @xmath283 such that @xmath284 , where @xmath234 is as in table 1 .
let us denote the resulting ribbon category by @xmath285 .
the tensor structure in @xmath285 and the corresponding verlinde algebra do not depend on the choice of @xmath283 . the ribbon category @xmath285 is premodular , since it has only a finite number of isomorphism classes of simple objects .
it is also hermitian , see @xcite , so that @xmath286 for every @xmath287 .
( this can also be deduced from the explicit formulas for the quantum dimensions of the weyl modules ) .
thus for every simple object @xmath190 , one has @xmath288 is a positive real number ( since @xmath289 ) . in case 2
if @xmath290 , then this hermitian structure is known to be unitary , see @xcite .
then @xmath291 for every @xmath287 .
again this can also be deduced from the explicit formulas for the quantum dimensions of the weyl modules .
it is clear that @xmath292 where @xmath69 is the order of @xmath251 .
when @xmath69 is divisible by @xmath231 , @xmath293 , and @xmath283 has order @xmath294 , the category @xmath285 is modular .
these are the well - known , and so far the main , examples of modular category .
[ [ tildecal - cfrak - gvarepsilonzeta - as - a - premodular - g - category ] ] @xmath285 as a premodular @xmath0-category ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ let @xmath295 be the projection .
for @xmath41 , let @xmath72 be the set of dominant weights @xmath247 such that @xmath296 and @xmath297 .
let @xmath298 be the set of all objects in @xmath285 which are direct sums of @xmath299 with @xmath300 .
set @xmath301 .
we consider @xmath8 as a full subcategory of @xmath285 .
if all the colors of a framed link @xmath51 are simple @xmath3-modules with highest weights in the root lattice , then the quantum invariant of @xmath51 is in @xmath311 $ ] , according to the integrality , see @xcite .
hence the fact that @xmath8 is a modular , or weakly non - degenerate , @xmath0-category does not depend on the choice of the @xmath234-th root @xmath283 of @xmath5 ; it totally depends on the order @xmath69 of @xmath251 .
note that the fact @xmath285 is a ( non - graded ) modular category does depend on the choice of @xmath283 .
the following proposition shows that in case 1 of sect .
[ cases ] , the category @xmath312 constructed above is a modular @xmath0-category , at least under the assumption that @xmath69 is co - prime with @xmath230 .
[ ref22 ] suppose @xmath3 is a lie algebra of series @xmath316 with odd @xmath317 .
assume that the order @xmath69 of @xmath251 is even but not divisible by 4 and @xmath318 .
then @xmath8 is a modular @xmath0-category .
( note that for @xmath316 , one has @xmath319 , and @xmath320 . ) in the cases of propositions [ ref11 ] and [ ref22 ] , it can be shown that the category @xmath21 is the product of its neutral component and a modular category associated with the center group @xmath0 ( see the definition in @xcite , see also @xcite ) .
the corresponding invariant of a 1-cohomology class on a 3-manifold @xmath6 is then the product of the invariant of the cohomology class @xmath76 with an invariant depending only on @xmath321 and the linking form .
the theory in this case is rather trivial .
[ quq ] one may ask when the @xmath0-category @xmath8 is weakly - nondegenerate . the answer is only when the order @xmath69 of @xmath251 is as one described in propositions [ ref11 ] , [ ref22 ] , [ pre ] , or one in the following additional cases : for the lie algebra @xmath338 with odd @xmath317 the number @xmath69 must satisfies @xmath339 ; for the lie algebra @xmath340 , one must have @xmath341 , where @xmath342 .
our aim here is to study the set @xmath107 of @xmath46-transparent objects of @xmath343 . to cover both cases 1 and 2 of sect .
[ cases ] we introduce a group @xmath344 equal to @xmath0 in case 1 and equal to @xmath345 in case 2 .
more precisely , set @xmath346 where @xmath347 , @xmath348 in case 1 and @xmath349 are the @xmath279-dual lattices in case 2 . in both cases
one has @xmath350 .
the group @xmath344 acts on @xmath271 as follows .
let @xmath351 with a lift @xmath352 and let @xmath353 .
the element @xmath354 may not lie in the simplex @xmath355 anymore , but there is @xmath275 which maps @xmath356 into @xmath271 .
set @xmath357 .
let the dot version of this action be the one shifted by @xmath249 so that @xmath358 .
\(ii ) if @xmath69 is divisible by @xmath231 , @xmath360 , and @xmath283 is a root of unity of order @xmath294 then every @xmath108-transparent simple object in @xmath361 is isomorphic to @xmath362 with @xmath363 .
[ sub ] to prove proposition [ sub ] we first recall the so - called second symmetry principle for quantum link invariants ( see @xcite ) .
it describes how these invariants change under the action of @xmath344 on the colors of link components .
we record two corollaries of this principle .
first , @xmath366 this means for the hopf link colored by @xmath63 and any element in @xmath46 , we can unlink the hopf link not altering the value of the quantum invariant .
thus the operator of figure [ s]a with @xmath378 is equal to @xmath379 . applying lemma [ trans3 ] ( for strings piercing the unknot with color @xmath384 )
, we obtain that @xmath385 where @xmath386 is the set of isomorphism classes of objects in @xmath387 . on the other hand
, @xmath8 is known to be modular in the ungraded sense .
hence a string colored by a simple object piercing through an unknot with color @xmath388 is non - zero only when the color of the string is a trivial object .
hence @xmath389 which , by corollary [ equal ] , is @xmath390 .
thus @xmath391 from part ( i ) we know that @xmath392 for every @xmath351 , and @xmath393 by ( [ dim ] ) .
since @xmath394 , we can have @xmath395 only when @xmath396 .
hence we have at least @xmath397 pairwise non - isomorphic simple objects in @xmath107 , each has the square of the quantum dimension equal 1 . since @xmath398 is a positive real number for any @xmath399
, we obtain that @xmath386 consists only of the isomorphism classes of @xmath400 with @xmath351 .
now we prove that @xmath411 . by the second symmetry principle in @xcite ,
the twist @xmath84 acts on each @xmath400 as multiplication by @xmath412 where @xmath352 is a lift of @xmath351 .
now ( b ) implies that the twist acts as the identity operator on all @xmath46-transparent objects . from ( [ dim ] )
we see that for @xmath413 one has @xmath414 . if @xmath415 ( in that case @xmath416 ) then @xmath417 is positive , hence @xmath418 .
when @xmath283 is an arbitrary root of unity of order @xmath294 , by considering a galois action , we also have @xmath418 .
we have @xmath420 .
let @xmath368 be the non - trivial element of @xmath344 .
explicit calculation shows that @xmath400 is not in @xmath108 .
it follows from proposition [ sub ] part ( ii ) that @xmath108 does not have any non - trivial transparent objects .
hence by a criterion of bruguieres ( simplified in ( * ? ? ?
* lemma 4.3 ) ) , @xmath46 is modular .
\1 . in the @xmath421-case
, there is a decomposition of quantum invariants considered by blanchet and beliakova , using idempotents in the birman - wenzl - murakami category .
their construction is probably different from ours because they do not consider spin representations .
see also @xcite .
\2 . for a quotient group @xmath422 of @xmath0
, one can consider any @xmath0-category as a @xmath422-category . by choosing @xmath422 appropriately
, we can make the proof of proposition [ pre ] valid even when the order @xmath69 of @xmath251 does not satisfy the conditions there . in this way one can get a cohomology decomposition for @xmath337 in the case @xmath423 , as in @xcite .
similar result holds true for the lie series @xmath234 .
suppose @xmath424 ( otherwise the theory is reduced to the ungraded case ) .
if the order @xmath69 of @xmath251 does not satisfy the conditions of proposition [ ref22 ] or [ ref11 ] , then it can be shown that @xmath46 contains a non - trivial @xmath108-transparent element of the form @xmath425 , hence @xmath8 can not be modular ( in the @xmath0-category sense ) .
kirby , r. , melvin , p. , on the 3-manifold invariants of witten and reshetikhin - turaev for @xmath426 .
( 1991 ) , 473545 .
kirillov , a. _ on an inner product in modular categories _ , jour .
ams , * 9 * ( 1996 ) , 11351169 . murakami , h. , _ quantum invariants for 3-manifolds _ in the 3rd korea - japan school of knots and links , proc . of applied math .
workshop , * 4 * ( 1994 ) , 129143 .
sawin , s. _ jones - witten invariants for non - simply connnected lie groups and the geometry of the weyl alcove _ , preprint math.qa/9905010 . | for a group @xmath0 , the notion of a ribbon @xmath0-category was introduced in @xcite with a view towards constructing @xmath1-dimensional homotopy quantum field theories ( hqft s ) with target @xmath2 .
we discuss here how to derive ribbon @xmath0-categories from a simple complex lie algebra @xmath3 where @xmath0 is the center of @xmath3 .
our construction is based on a study of representations of the quantum group @xmath4 at a root of unity @xmath5 . under certain assumptions on @xmath5 ,
the resulting @xmath0-categories give rise to numerical invariants of pairs ( a closed oriented 3-manifold @xmath6 , an element of @xmath7 ) and to 3-dimensional hqft s .
-0.2 cm -0.2 cm [ section ] [ thm]lemma [ thm]proposition [ thm]corollary [ thm]remark [ section ] [ section ] c i d zz rr = 12.5 cm = 19.0 cm = 0.5 cm = 0pt = 12pt = 12pt hom i d sgn |
superfluidity resulting from a bose - einstein condensate ( bec ) in atomic gases is characterized by the irrotational nature of the condensate flow .
the recent experimental observation of the scissors mode in a trapped bose - condensed gas @xcite clearly demonstrated the irrotational nature of a trapped superfluid bose gas .
the scissors mode is the oscillation of the atomic cloud with respect to a symmetry axis of an anisotropic trap potential @xcite , induced by a sudden rotation of the trap about this axis . above the bose - einstein transition temperature ( @xmath2 ) ,
the thermal cloud exhibits two normal mode frequencies corresponding to the rotational and irrotational motion .
in contrast , the pure condensate motion at @xmath1 only exhibits one frequency , since only irrotational motion is allowed . the above distinction between the condensate ( superfluid ) oscillation at @xmath3 and the thermal gas ( normal fluid ) oscillation above @xmath2
was clearly observed in the experiment reported in ref .
the observed frequencies of oscillations are in good agreement with the theoretical predictions in ref .
@xcite at @xmath1 and @xmath4 . at finite temperatures in the bose - condensed phase , where an appreciable fraction of the atoms are excited out of the condensate ,
one expects coupled motions of the superfluid and normal fluid components .
recent experiments at oxford @xcite observed such a coupled scissors mode oscillations at finite temperatures , and determined the temperature dependence of the frequency and damping rate of the oscillations of each component . in this paper , we discuss the scissors mode in a trapped bose - condensed gas at finite temperatures using the kinetic theory derived by zaremba , nikuni and griffin ( zng ) @xcite . in the zng kinetic theory
, one has a generalized gross - pitaevskii equation for the condensate atoms and a semi - classical kinetic equation for the noncondensate atoms .
the condensate and noncondensate are coupled through mean - field interactions as well as collisions between the atoms ( the so - called @xmath0 collisions ) . in this paper , we restrict ourselves to the collisionless regime , where the mean collision rate is much smaller than collective mode frequencies . generalizing the moment calculation approach used by gury - odelin and stringari @xcite , we derive coupled equations describing oscillations of the quadrupole moments of the condensate and noncondensate components at finite temperatures .
recently , jackson and zaremba @xcite have solved the same coupled zng equations numerically using fft / monte - carlo simulations to discuss the temperature - dependent oscillations associated with the scissors mode .
these authors found excellent agreement with the oxford data @xcite .
our work is more analytical and complements the numerical results in ref.@xcite .
we consider a bose - condensed gas confined in an anisotropic harmonic trap potential described by @xmath5 with @xmath6 and @xmath7 .
the parameter @xmath8 characterizes the deformation of the trap potential in the @xmath9-@xmath10 plane .
the coupled dynamics of the condensate and noncondensate @xcite is described by the generalized gross - pitaevskii ( gp ) equation for the condensate wavefunction @xmath11
@xmath12 \phi({\bf r},t ) , \label{eq : gp}\ ] ] and the semi - classical kinetic equation for the noncondensate distribution function @xmath13 @xmath14+c_{22}[f ] .
\label{eq : qk}\ ] ] here @xmath15 is the condensate density , and @xmath16 is the noncondensate density , @xmath17 and @xmath18 $ ] is the time - dependent effective potential acting on the noncondensate , including the hartree - fock ( hf ) mean field . as usual , we treat the interaction in the @xmath19-wave scattering approximation @xmath20 .
the dissipative term @xmath21 in the generalized gp equation ( [ eq : gp ] ) is due to the collisional exchange of atoms in the condensate and noncondensate , which is related to the @xmath0 collision integral in ( [ eq : qk ] ) , @xmath22 . \label{r_term}\ ] ] the explicit expressions for the two collision integrals in the kinetic equation ( [ eq : qk ] ) are given by @xcite @xmath23&=&\frac{2g^2}{(2\pi)^5\hbar^7}\int d{\bf p}_2 \int d{\bf p}_3 \int d{\bf p}_3 \int d{\bf p}_4 \cr & & \times\delta({\bf p}+{\bf p}_2-{\bf p}_3-{\bf p}_4 ) \delta(\tilde\varepsilon_{p_1}+\tilde\varepsilon_{p_2}-\tilde\varepsilon_{p_3 } -\tilde\varepsilon_{p_4})\cr & & \times [ ( 1+f)(1+f_2)f_3f_4-ff_2(1+f_3)(1+f_4 ) ] , \label{eq : c22}\end{aligned}\ ] ] @xmath24&=&\frac{2g^2n_c}{(2\pi)^2\hbar^4}\int d{\bf p}_1 \int d{\bf p}_2 \int d{\bf p}_3 \cr & & \times \delta ( m{\bf v}_c+{\bf p}_1-{\bf
p}_2-{\bf p}_3 ) \delta(\varepsilon_c+\tilde\varepsilon_{p_1}-\tilde\varepsilon_{p_2 } -\tilde\varepsilon_{p_3 } ) \cr & & \times[\delta({\bf p}_1-{\bf p})-\delta({\bf p}_2-{\bf p})-\delta({\bf p}_3-{\bf p } ) ] \cr & & \times [ ( 1+f_1)f_2f_3-f_1(1+f_2)(1+f_3 ) ] .
\label{eq : c12}\end{aligned}\ ] ] here @xmath25 and @xmath26 represent the local energy of the noncondensate and condensate atoms @xmath27 where the condensate chemical potential @xmath28 is defined in ( [ eq : muc ] ) .
it is convenient to rewrite the gp equation in ( [ eq : gp ] ) in terms of the amplitude and phase variables @xmath29 .
this leads to the quantum hydrodynamic equations for the condensate : @xmath30 [ hydro_c ] where the condensate velocity is @xmath31 and the condensate chemical potential @xmath28 is defined by @xmath32 throughout this paper , we use the thomas - fermi ( tf ) approximation , which neglects the quantum pressure term in the condensate chemical potential to give : @xmath33 within this tf approximation , the equilibrium condensate density profile is given by @xmath34 - 2\tilde n_0({\bf r } ) .
\label{eq : nc0}\ ] ] the equilibrium distribution @xmath35 describing the noncondensate atoms is given by the bose - einstein distribution function @xmath36 } -1 } , \label{eq : f0}\ ] ] where @xmath37 and @xmath38 $ ] .
the equilibrium density of the noncondensate described by ( [ eq : f0 ] ) is @xmath39 where @xmath40 is the thermal de broglie wavelength , @xmath41 is the bose - einstein function , and @xmath42}$ ] is the local fugacity .
we note that the equilibrium tf densities of the condensate and noncondensate determined from ( [ eq : nc0 ] ) - ( [ ntilde0 ] ) are isotropic in the renormalized coordinates @xmath43 , where @xmath44 .
thus @xmath45 and @xmath46 can be expressed as functions of @xmath47 .
of course , the expression for @xmath45 in ( [ eq : nc0 ] ) is only valid within the tf condensate radius @xmath48 , where @xmath49.\ ] ] outside the tf condensate radius @xmath50 , one has @xmath51 .
the scissors mode can be excited by a sudden rotation of the anisotropic trap potential ( [ eq : trap ] ) in the @xmath9-@xmath10 plane by a small angle @xmath52 .
this perturbation induces fluctuations in the quadrupole moment @xmath53 where @xmath54 . following ref .
@xcite , one can derive coupled moment equations involving the relevant quadrupole variables for the condensate and noncondensate , starting from ( [ hydro_c ] ) and ( [ eq : qk ] ) . in the linearized theory , these exact moment equations are given by ( after some work ) @xmath55 [ moments_eq1 ] the various condensate and noncondensate moments related to some function @xmath56 are defined by @xmath57 @xmath58 here the local velocity @xmath59 and the local density of the kinetic pressure tensor @xmath60 of the thermal cloud are defined in terms of the single - particle distribution function by @xmath61 @xmath62 in ( [ moments_eq1 ] ) , one also has moments involving the collision integrals @xmath0 and @xmath63 : @xmath64 @xmath65 + c_{22}[f({\bf
r},{\bf p},t)]\right\}. \label{pxy_coll}\ ] ] the condensate moments and noncondensate moments are coupled through the hf mean - field as well as the collisional exchange term @xmath66 .
if we neglect the coupling between the two components by setting @xmath67 and @xmath68 in ( [ xy_c ] ) - ( [ pxy_n ] ) , we recover the uncoupled moment equations derived in ref .
@xcite , namely the @xmath1 condensate equations in ( [ xy_c ] ) and ( [ xv_c ] ) , and the @xmath4 thermal cloud gas equations in ( [ xy_n ] ) - ( [ pxy_n ] ) .
the moment equations in ( [ moments_eq1 ] ) are exact consequences of the coupled zng equations ( [ eq : gp ] ) and ( [ eq : qk ] ) , but are obviously not closed because of the hf mean - field terms and the collision terms . to obtain a closed set of equations
, we must introduce an ansatz for the condensate and noncondensate oscillations .
for the condensate , we assume that the velocity field has the same irrotational form as the @xmath1 result , namely @xmath69 where the small parameter @xmath70 characterizes the amplitude of the condensate oscillation . for the noncondensate atoms , it is convenient to write the distribution function as @xmath71\psi({\bf r},{\bf p},t ) , \label{deltaf}\ ] ] where @xmath72 represents the small fluctuation of the noncondensate around static equilibrium .
we make the ansatz that @xmath72 can be written in the form @xmath73 with @xmath74 and @xmath37 . the first term in ( [ ansatz_n ] ) is associated with the density fluctuations , while the second term is associated with the velocity fluctuations .
we also assume that the noncondensate velocity field has the form @xmath75 the third term in ( [ ansatz_n ] ) represents the quadrupole deformation in the momentum distribution , characterized by a small parameter @xmath76 .
the form of ( [ ansatz_n ] ) , with ( [ ansatz_vn ] ) , is motivated by the exact solution @xcite of the collisionless kinetic equation above @xmath2 ( eq . ( [ eq : qk ] ) without @xmath0 , @xmath63 , and @xmath77 ) , these normal modes have frequencies @xmath78 . taking the time derivative of ( [ xv_c ] ) , ( [ xvp_n ] ) and ( [ xvm_n ] ) , and using ( [ ansatz_vc ] ) and ( [ ansatz_vn ] ) ,
we obtain the following set of equations : @xmath79 [ moments_eq2 ] the effect of the hf mean field is parameterized in ( [ moment_1])-([moment_3 ] ) by the following three quantities : @xmath80 [ \bbox{\nabla}n_{c0}\cdot\bbox{\nabla}(xy ) ] \right\}/\theta_{c,{\rm rigid } } , \label{delta1 } \\ \delta_2&\equiv & \left\{\frac{2g}{\omega_0 ^ 2}\int d{\bf r } [ \bbox{\nabla}\tilde n_0\cdot\bbox{\nabla}(xy ) ] [ \bbox{\nabla}n_{c0}\cdot\bbox{\nabla}(xy ) ] \right\}/\theta_{n,{\rm rigid } } , \label{delta2 } \\
\delta_3&\equiv&\left[\frac{1}{1-\epsilon^2}\frac{2g}{\omega_0 ^ 2 } \int d{\bf r}\tilde n_0({\bf r}\cdot\bbox{\nabla}n_0)\right ] /\theta_{n,{\rm rigid } } \cr & = & \left[\frac{2}{(1-\epsilon^2)\tilde\omega_0 ^ 2}\int d{\bf r}\tilde p_0({\bf r } ) \right]/\theta_{n,{\rm rigid}}-1 .
\label{delta3}\end{aligned}\ ] ] here @xmath81 is the equilibrium kinetic pressure , while @xmath82 and @xmath83 are the rigid values for the moment of inertia of the condensate and noncondensate components ( see , for example , refs.@xcite and @xcite ) : @xmath84 for later use , we also note that ( [ pxy_n ] ) can also be written as @xmath85 + \left\langle \frac{p_xp_y}{m}\right \rangle_{\rm coll}. \label{pxy_n2}\ ] ] if we neglect the collisions , ( [ moment_1])-([moment_3 ] ) provide a closed set of equations for the three moments @xmath86 , @xmath87 , and @xmath88 , which can be used to study the undamped scissors - mode oscillations involving the condensate and noncondensate components at finite temperatures . to truncate the collision terms @xmath66 and @xmath89 in ( [ moments_eq2 ] ) , we use ( [ deltaf ] ) in the collision integrals and linearize in @xmath72 as follows [ see also ref .
@xcite ] : @xmath23&\simeq&\frac{2g^2}{(2\pi)^5\hbar^7}\int d{\bf p}_2 \int d{\bf p}_3 \int d{\bf p}_4 \cr & & \times \delta({\bf p}+{\bf p}_2-{\bf p}_3-{\bf p}_4 ) \delta(\tilde\varepsilon_p+\tilde\varepsilon_{p_2}-\tilde\varepsilon_{p_3 } + \tilde\varepsilon_{p_4 } ) \cr & & \times f_0f_{20}(1+f_{30})(1+f_{40})(\psi_3+\psi_4-\psi_2-\psi)\equiv \hat l_{22}[\psi],\end{aligned}\ ] ] and @xmath24&\simeq & \frac{2 g^2 n_c}{(2\pi)^2\hbar^4 } \int d{\bf p}_1 \int d{\bf p}_2 \int d{\bf p}_3 \cr & & \times \delta(m{\bf v}_c+{\bf p}_1-{\bf p}_2-{\bf p}_3 ) \delta(\mu_{c0}+\tilde\varepsilon_{p_1 } -\tilde\varepsilon_{p_2}-\tilde\varepsilon_{p_3 } ) \cr & & \times [ \delta({\bf p}-{\bf p}_1)-\delta({\bf p}-{\bf p}_2 ) -\delta({\bf p}-{\bf p}_3 ) ] \cr & & \times ( 1+f_{10})f_{20}f_{30 } ( \beta g \delta n_c+\psi_2+\psi_3-\psi_1 ) \cr & & \equiv \beta g \delta
n_c \hat l_{12}[1]+\hat l_{12}[\psi ] .
\label{c12_1}\end{aligned}\ ] ] here the linearized @xmath90 operator in ( [ c12_1 ] ) is defined by @xmath91 & \equiv & \frac{2 g^2 n_c}{(2\pi)^2\hbar^4 } \int d{\bf p}_1 \int d{\bf p}_2 \int d{\bf p}_3 \cr & & \times \delta(m{\bf v}_c+{\bf p}_1-{\bf p}_2-{\bf p}_3 ) \delta(\mu_{c0}+\tilde\varepsilon_{p_1 } -\tilde\varepsilon_{p_2}-\tilde\varepsilon_{p_3 } ) \cr & & \times [ \delta({\bf p}-{\bf p}_1)-\delta({\bf p}-{\bf p}_2 ) -\delta({\bf p}-{\bf p}_3 ) ] \cr & & \times ( 1+f_{10})f_{20}f_{30 } ( \psi_2+\psi_3-\psi_1 ) . \label{l12}\end{aligned}\ ] ] using the ansatz for @xmath72 given by ( [ ansatz_n ] ) , one finds that the source function @xmath92 in ( [ r_term ] ) reduces to @xmath93 , \label{g12_approx}\ ] ] where collision time @xmath94 is defined by @xmath95.\ ] ] using ( [ g12_approx ] ) and working to first order in @xmath96 ( since we are working in the collisionless regime ) , one obtains the following equation of motion : @xmath97 .
\label{dxy12dt}\ ] ] here @xmath98 and @xmath99 are the averaged relaxation times associated with the @xmath0 collisional exchange of atoms : @xmath100 ^ 2\right\ } /\theta_{c,{\rm rigid } } , \label{tau1}\\ \frac{1}{\tau_{2}}&=&\left\ { \frac{g}{\omega_0 ^ 2}\int d{\bf r}\frac{\beta gn_{c0}({\bf r } ) } { \tau_{12}({\bf r})}[\bbox{\nabla}n_{c0}\cdot\bbox{\nabla}(xy)]^2\right\ } /\theta_{n,{\rm rigid}}. \label{tau2}\end{aligned}\ ] ] one can also show that , using the ansatz for @xmath13 in ( [ deltaf ] ) with ( [ ansatz_n ] ) , one can write @xmath101 defined by ( [ pxy_coll ] ) in the form of a relaxation time approximation @xmath102 with the associated relaxation time @xmath103 defined by @xmath104 where @xmath105 is the viscous relaxation time ( see also refs.@xcite ) defined by @xmath106+\hat l_{22}[p_xp_y]\right\ } } { \displaystyle \int\frac{d{\bf p}}{(2\pi\hbar)^3 } ( p_xp_y)^2f_0(1+f_0)}.\ ] ] using ( [ pxy_tau ] ) in ( [ pxy_n2 ] ) , we find the following equation of motion @xmath101 @xmath107 -\frac{1}{\tau_3}\left\langle\frac{p_xp_y}{m}\right\rangle_{\rm coll}. \label{pxydt}\ ] ] in summary , we have obtained a closed set of the three moment equations , as given by ( [ moments_eq2 ] ) , ( [ dxy12dt ] ) and ( [ pxydt ] ) .
they involve three parameters @xmath108 characterizing the hf mean - field coupling and three collisional relaxation times @xmath109 .
one problem in actual calculations of these quantities is that the integrands in the formulas for @xmath110 , @xmath111 , @xmath98 and @xmath99 have singularities at the tf condensate boundary ( when @xmath112 ) arising from the divergence in @xmath113 ( we recall @xmath114 ) .
these singular contributions are an unphysical artifact of the simple tf approximation . for our present purposes
, we can introduce a cutoff in the integrals where the tf approximation breaks down .
more explicitly , we follow refs . @xcite and introduce a cutoff at @xmath115 , where @xmath116,\ ] ] with @xmath117 . it is convenient to introduce a 3-component vector describing the dynamical moment variables :
@xmath118 the coupled moment equations we have derived then reduce to the following matrix equation @xmath119 the @xmath120 matrix @xmath121 is given by @xmath122 and the column vector @xmath123 describes the terms related to collisions @xmath124 to first order in @xmath125 , one has the following equation for @xmath123 : @xmath126 where the elements of the @xmath120 matrix @xmath127 are given by @xmath128 in ( [ k ] ) , we have introduced the renormalized relaxation times @xmath129 we can now look for normal mode solutions of ( [ eq_matrix ] ) with the time dependence @xmath130 . the matrix equation in ( [ eq_matrix ] ) then reduces to @xmath131 where @xmath132 is the renormalized mode frequency .
we discuss the solutions of ( [ eq_matrix2 ] ) in the next two sections .
in this section , we first discuss the collective oscillations of the condensate and noncondensate components neglecting the collisional damping .
this means we set @xmath133 in ( [ eq_matrix ] ) , or equivalently , @xmath134 in ( [ eq_matrix2 ] ) . in this approximation
, our eigenvalue equation reduces to @xmath135 neglecting the hf mean fields ( @xmath136 ) , the solution of ( [ eq_for_omega ] ) is given by the zeros of the determinant @xmath137 in this simple limit , the mode frequencies are @xcite @xmath138 and @xmath139 .
the thermal gas modes thus have a high frequency and a low frequency component . keeping @xmath108 finite ,
we next solve the eigenvalue problem ( [ eq_for_omega ] ) numerically , using the trap parameters of the recent oxford experiment @xcite , namely , @xmath140hz , @xmath141 .
the total number of atoms is taken to be @xmath142 . in fig . 1
, we plot the normal mode frequencies as a function of the temperature .
we find three normal mode frequencies , denoted by @xmath143 and @xmath144 . in the low temperature limit , @xmath145 approaches the well - known @xmath1 condensate mode frequency @xmath138 @xcite . at finite temperatures ,
the frequency is shifted due to the hf mean field of the noncondensate atoms .
expanding the solution of ( [ eq_for_omega ] ) to first order in the interaction parameters @xmath108 , one finds @xmath146 this formula for @xmath145 is a good approximation to the direct numerical solution of ( [ eq_for_omega ] ) .
this result can also be obtained directly by neglecting the noncondensate fluctuations in ( [ moment_1 ] ) , i.e. , @xmath147 .
thus the positive frequency shift in the condensate mode ( @xmath145 ) in fig .
1 is mainly due to the mean field of the static thermal cloud , @xmath148 . as for the noncondensate ( @xmath149 ) modes ,
the frequencies approach the pure thermal cloud frequencies found above @xmath2 @xcite , @xmath150 as @xmath151 approaches @xmath2 . at low temperatures , however , there are large frequency shifts in these noncondensate modes because of the coupling to a large condensate .
the solutions of ( [ eq_for_omega ] ) also give the relative amplitude of the quadrupole oscillations of the condensate and noncondensate components .
it is more convenient to use the rotation angles @xmath152 and @xmath153 of the condensate and noncondensate clouds , assuming that the two components rotate without distortion .
these rotation angles are related to the quadrupole moments through @xmath154 we note that using the tf density profiles for @xmath45 and @xmath46 in ( [ theta_nc ] ) , one obtains @xmath155 in fig.2(a)-(c ) , we plot the temperature dependence of rotation amplitudes in ( [ theta_xy ] ) of the condensate and noncondensate oscillations associated with the @xmath145 and @xmath149 modes .
we find that the @xmath145 and @xmath144 modes involve in - phase oscillations of the two components , while the @xmath156 mode involves out - of - phase oscillations . at @xmath157 ,
the @xmath145 mode mostly involves the condensate oscillations . for decreasing @xmath151 , the amplitude of the noncondensate component increases , and at low temperatures
, the thermal cloud is almost moving together with the condensate cloud .
in contrast , the @xmath149 modes involve mainly noncondensate oscillations at all temperatures .
all these results agree with those in ref .
@xcite .
we next consider the experimental situation @xcite where the scissors modes are excited by the sudden rotation of the trap potential by the angle @xmath52 at @xmath158 .
we assume that the atoms are initially in thermal equilibrium for @xmath159 .
the corresponding initial conditions for the angles @xmath160 and @xmath161 are found from the moment equations in ( [ moments_eq1 ] ) , @xmath162\theta_0,~~ \cr \theta_n'(0)&=&\theta_n'''(0)=\theta_n'''''(0)=0 . \label{initial}\end{aligned}\ ] ] the solution for @xmath160 and @xmath161 for @xmath163 with the initial conditions in ( [ initial ] ) is given by @xmath164 where the amplitudes of each mode are given by the following explicit formulas : @xmath165 } { ( \omega_i^2-\omega_j^2)(\omega_i^2-\omega_k^2)}. \label{coeff}\end{aligned}\ ] ] in ( [ coeff ] ) , @xmath166 represent the three mode indices ( @xmath167 ) and @xmath168 . in fig .
3 , we plot the temperature dependence of these amplitudes @xmath169 and @xmath170 .
we find that the oscillations of the condensate always have a large weight in the condensate mode ( @xmath145 ) .
in contrast , the thermal cloud oscillations make a large contribution to the noncondensate modes ( @xmath149 ) at @xmath157 , but the condensate mode ( @xmath145 ) is dominant at low temperatures .
this behavior is in agreement with the numerical results obtained by jackson and zaremba @xcite .
our results for the scissors mode can be used to derive an expression for the moment of inertia , a quantity which characterizes the rotational properties of a trapped bose gas .
zambelli and stringari @xcite derived the following exact relation between the moment of inertia @xmath171 and the quadrupole response function : @xmath172 where @xmath173 is the imaginary part of the quadrupole response function of the system .
one can also show that @xmath173 is related to the fourier transform of the quadrupole moment induced by the sudden rotation of the trap potential , @xmath174 using the solution for @xmath160 and @xmath161 in ( [ angles ] ) with the relations in ( [ theta_xy ] ) , we find that @xmath175 is given by @xmath176 \right .
\cr & & \left .
+ \theta_{n,{\rm rigid } } \sum_i b_i[\delta(\omega-\omega_i)+\delta(\omega+\omega_i ) ] \right\}. \label{q_omega}\end{aligned}\ ] ] one can show that @xmath173 given by ( [ chi_q ] ) with ( [ q_omega ] ) satisfies the exact sum rule relation @xcite @xmath177 here @xmath178 is the rigid value of the moment of inertia of the total system ( see ( [ theta_nc ] ) ) , @xmath179 where @xmath180 . using these results for @xmath181 in ( [ theta ] )
, one finds @xmath182 from the explicit formulas for @xmath169 and @xmath170 given in ( [ coeff ] ) , we find @xmath183 noting that @xmath184 which follows from the definitions of @xmath110 and @xmath111 in ( [ delta1 ] ) and ( [ delta2 ] ) , we finally obtain the following simple formula for moment of inertia : @xmath185 it is interesting to note that because of the interactions between the two components , the irrotational contribution @xmath186 and the rotational contribution @xmath83 in ( [ m_inertia ] ) can not be associated separately with the condensate and noncondensate quadrupole moments in ( [ m_inertia0 ] ) . thus obtaining the simple final expression given in ( [ m_inertia ] ) is nontrivial . in fig . 4
, we plot the moment of inertia in ( [ m_inertia ] ) as a function of temperature , as well as the separate irrotational contribution @xmath186 and the rotational contribution @xmath83 .
an expression analogous to ( [ m_inertia ] ) for the noninteracting trapped bose gas was obtained by stringari ( see especially fig . 1 of ref .
@xcite ) . the above result for the moment of inertia in ( [ m_inertia ] ) can be understood by considering the anisotropic trap potential in ( [ eq : trap ] ) rotating with the angular frequency @xmath187 in the @xmath188 plane . in this rotating frame ,
the stationary distribution of the noncondensate is given by @xcite @xmath189\}-1}.\ ] ] to first order in @xmath187 , the total angular momentum of the noncondensate component is given by [ see also ref .
@xcite ] @xmath190 thus the noncondensate part of the moment of inertia is given by the rigid value defined in ( [ theta_nc ] ) , i.e. , @xmath191 .
for the condensate contribution , one has to find the stationary solution of the gp equation in the rotating frame @xcite . in the linear regime , the stationary velocity field is determined from @xmath192=0.\ ] ] for the equilibrium tf condensate density @xmath193 determined from ( [ eq : nc0 ] ) , one can show that the stationary velocity field is given by @xmath194 the total angular momentum of the condensate component is then given by [ see ( [ q0tf ] ) ] @xmath195 thus the condensate part of the moment of inertia is given by @xmath196 .
this has the same form as the @xmath1 expression for the irrotational value of the moment of inertia of the tf condensate given in ref .
thus , the first term in ( [ m_inertia ] ) represents the irrotational value of the condensate component in the thomas - fermi limit , while the second term is the rigid value of the noncondensate component .
the expression in ( [ m_inertia ] ) is analogous to the non - interacting gas expression ( @xmath136 ) first obtained by stringari @xcite .
however , as we noted in the derivation of ( [ m_inertia ] ) , this equivalence is nontrivial when @xmath197 because of the cancellation of contributions in going from ( [ m_inertia0 ] ) to ( [ m_inertia ] ) .
we next consider the damping of the collective modes due to the collisions between atoms .
since we are working in the collisionless regime @xmath198 , in ( [ eq_matrix2 ] ) , we can treat @xmath127 as a perturbation to @xmath121 .
working to first order in @xmath125 , one has the following equation for normal mode frequencies : @xmath199 where @xmath200 is @xmath120 unit matrix and the function @xmath201 related to damping is found to be given by @xmath202 \cr & & + ( \tilde\omega^2 - 2+\delta_1)\left\{\frac{2}{\tilde\tau_3 } [ \tilde\omega^2 - 2\epsilon^2(1-\delta_2/2)]+\frac{1}{\tilde\tau_2 } [ \tilde\omega^2 - 2\epsilon^2(1+\delta_3)]\right\ } \cr & & -2\epsilon^2\frac{\delta_1\delta_2}{\tilde\tau_3 } -2\left[\tilde\omega^2 - 2\epsilon^2(1+\delta_3)\right]\frac{\delta_2}{\tilde\tau_1}.\end{aligned}\ ] ] we note that the lowest order or first term in ( [ det2 ] ) can be written in terms of the undamped frequencies @xmath203 determined from ( [ eq_for_omega ] ) , namely @xmath204 we now look for the damping of these frequencies , i.e. , @xmath205 to first order in @xmath125 , we find the following expression for the damping rates : @xmath206 expanding the solution ( [ gamma ] ) in @xmath108 , one can obtain approximate expressions for the damping rate @xmath207 . the lowest - order approximation
( neglecting @xmath208 and @xmath209 ) gives after some algebra @xmath210 we recall that within this approximation , @xmath138 and @xmath139 .
one can show that damping for the condensate mode given in ( [ gamma_approx ] ) is precisely equivalent to that obtained by williams and griffin @xcite , where the dynamics of the noncondensate was neglected . to see this
, we recall the ansatz ( [ ansatz_vc ] ) for @xmath211 and rewrite @xmath212 given in ( [ gamma_approx ] ) as [ see the definition of @xmath213 in ( [ tau1 ] ) ] @xmath214 ^ 2 } { \displaystyle m\int d{\bf r}~n_{c0 } v_c^2}. \label{gamma_c}\ ] ] using the quantum hydrodynamic equations for the condensate in ( [ hydro_c ] ) ( neglecting @xmath215 and @xmath216 ) , one can write ( [ gamma_c ] ) in terms of the density fluctuations @xmath217 in the undamped solution .
it is then straightforward to show that ( [ gamma_c ] ) is equivalent to the damping in eq .
( 21 ) of ref .
@xcite .
we have numerically evaluated ( [ gamma ] ) to find the damping of the collective modes . in fig . 4
, we plot the temperature dependence of the damping rates .
it turns out that the lowest - order formula for @xmath212 given in ( [ gamma_approx ] ) gives a good approximation for the full expression in ( [ gamma ] ) .
this means that the collisional damping of the condensate mode is almost entirely due to the @xmath0 collisional exchange with the static thermal cloud , as first discussed in ref .
@xcite , with little contribution from the @xmath63 collision term in ( [ pxy_coll ] ) .
since the viscous relaxation only involves noncondensate fluctuations , they only affect the condensate motion through the hf mean field @xmath218 .
thus the contribution of @xmath103 to the condensate damping @xmath212 is of second order in @xmath108 , which is a small correction .
in contrast , the damping of the @xmath149 modes is mainly due to the viscous relaxation associated with the @xmath103 relaxation time .
the increase of @xmath219 as the temperature decreases near @xmath2 is due to the increase in the @xmath0 collision rate .
this clearly indicates that the equilibration of the thermal cloud in a trapped bose - condensed gas is enhanced by the @xmath0 collisions with the condensate cloud @xcite . above @xmath2
, one does not have @xmath0 collisions and thus the damping @xmath219 is given by ( [ gamma_approx ] ) without the @xmath220 term , while the @xmath103 relaxation time now only involves the contribution from the @xmath63 collisions @xcite .
the damping of the condensate mode we have found is much smaller than the magnitude observed in the oxford experiments @xcite .
this is because the present moment analysis does not include landau damping of the condensate mode .
landau damping is due to the mean - field coupling of the condensate oscillations to a continuum of noncondensate excitations @xcite . in our moment calculation
, we truncate the hf mean field coupling using a simple ansatz for the noncondensate fluctuations , as given in ( [ ansatz_n ] ) .
this simple form describes the coherent motion of the thermal cloud , but completely neglects the effect from incoherent single - particle excitations which are responsible for landau damping .
recently , jackson and zaremba @xcite discussed the scissors mode at finite temperatures using a direct monte - carlo simulation of the zng semi - classical kinetic equation .
they found that for the same parameters used in the present study , landau damping is the main contribution to the damping of the condensate mode at all temperatures ( see also ref .
moreover , in ref .
@xcite , the @xmath63 collisions appear to mainly influence the damping by relaxing the system towards equilibrium and hence allowing the landau mechanism to be more effective .
the @xmath0 contribution to the damping of the condensate mode in ref .
@xcite is essentially in agreement with our results for the @xmath0 collisional damping rate in fig .
we note that the @xmath0 collisional damping of the condensate mode will increase in importance with the total number of atoms in the trapped gas .
we also note that the numerical results in ref .
@xcite find an increase of the damping of the noncondensate modes with decreasing temperature .
this is in contrast to our result for the collisional damping rate @xmath219 shown in fig .
the large damping in the noncondensate modes found in ref .
@xcite at low temperatures is presumably due to the anharmonicity in the effective potential @xmath221 for the noncondensate , arising from the large condensate mean field @xmath222 @xcite .
this effect is not included in our moment calculation because of our simple ansatz for the noncondensate fluctuations , given in ( [ ansatz_n ] ) .
in this paper , we have used the zng kinetic theory @xcite to study the scissors mode in a trapped bose - condensed gas at finite temperatures .
we derived three coupled moment equations starting from a generalized gp equation for the condensate and a semi - classical kinetic equation for the noncondensate .
our coupled moment equations are the natural extension to finite temperatures of a similar calculation carried out in ref .
@xcite for @xmath1 and @xmath4 .
we solved our coupled moment equations to find normal mode solutions .
these solutions were used to calculate the quadrupole response function and the moment of inertia , using the exact frequency moment sum rule formulas derived by zambelli and stringari @xcite .
it was shown that the moment of inertia is given by a simple expression in ( [ m_inertia ] ) , sum of the irrotational value of the condensate component and the rotational value of the noncondensate component .
this result clearly shows that the linear response of an trapped bose gas at finite temperatures involves the irrotational motion of the condensate component as well as the rotational motion of the noncondensate component .
we also evaluated the collisional damping for both the condensate and noncondensate oscillations .
our analysis naturally led to three collision times : ( 1)@xmath98 is the relaxation time of the condensate due to @xmath0 collisions .
( 2)@xmath99 is the relaxation time of the noncondensate due to @xmath0 collisions .
( 3)@xmath103 is the relaxation time of the noncondensate similar to that involved in the shear viscosity transport coefficient @xcite , which has a contribution from both @xmath0 and @xmath63 collisions .
we found that the noncondensate oscillations were damped mainly by a viscous - type relaxation ( @xmath223 ) , while the condensate damping was dominated by the collisional exchange due to @xmath0 collisions ( @xmath213 ) . if we neglect the dynamical mean - field coupling between the condensate and the noncondensate , one obtains an explicit expression for the condensate damping , which is identical to that given in ref .
@xcite assuming a static noncondensate .
we find that this simple static thermal cloud approximation provides a quantitatively good approximation for the @xmath0 collisional damping of the condensate mode .
however , our present discussion does not explain the damping observed in the oxford experiments @xcite . as discussed in section v , this is because a truncated moment calculation such as ours can not easily deal with the mechanism of landau damping , which is the dominant contribution to the collective mode damping in the collisionless or mean - field regime @xcite . in order to incorporate landau damping in the moment method ,
one will have to solve the exact set of moment equations ( [ xy_c ] ) - ( [ pxy_n ] ) including the incoherent motion of the thermal cloud .
our simple ansatz in ( [ ansatz_n ] ) only describes the coherent thermal cloud dynamics . in spite of the limitations mentioned above , the moment method developed in this paper
is very useful in understanding the detailed dynamics of a trapped bose - condensed gas at finite temperatures .
it provides an explicit physical picture of the coupled motions of the condensate and noncondensate components involved in the three scissors modes .
our work was inspired by , and complements , the recent numerical work of jackson and zaremba @xcite on the scissors mode .
our more analytical theory is essentially consistent with the results given in ref.@xcite , apart from missing the landau damping .
our method can be easily extended to study other collective modes , such as the monopole mode and the quadrupole modes @xcite , as well as to study the effect of a rotating thermal cloud at finite temperatures in vortex nucleation @xcite .
i thank a. griffin for useful suggestions and a critical reading of the manuscript .
i also have benefited from stimulating discussions with e. zaremba , b. jackson and j.e .
this research was supported by jsps of japan . calculated temperature dependence of three scissors - mode frequencies for @xmath142 @xmath224rb atoms in an anisotropic trap with @xmath140hz and @xmath141 @xcite .
the frequency is renormalized by @xmath225^{1/2}$ ] and the temperature is renormalized by the ideal gas transition temperature @xmath226^{1/3}$ ] .
the actual transition temperature is @xmath227 .
temperature dependence of coefficients @xmath169 and @xmath170 in ( [ angles ] ) and ( [ coeff ] ) for coupled scissors - mode oscillations induced by the sudden rotation of the trap potential .
panel ( a ) is a plot of the condensate rotation angle ( @xmath169 ) and panel ( b ) is a plot of the noncondensate angle ( @xmath170 ) , as defined in ( [ angles ] ) .
temperature dependence of the moment of inertia @xmath171 in ( [ m_inertia ] ) , normalized to @xmath229 defined by ( [ theta_rig ] ) .
we also plot the irrotational condensate component @xmath230 and the rotational noncondensate component @xmath231 , where @xmath82 and @xmath83 are defined in ( [ theta_nc ] ) . | we use a generalized gross - pitaevskii equation for the condensate and a semi - classical kinetic equation for the noncondensate atoms to discuss the scissors mode in a trapped bose - condensed gas at finite temperatures .
both equations include the effect of @xmath0 collisions between the condensate and noncondensate atoms .
we solve the coupled moment equations describing oscillations of the quadrupole moments of the condensate and noncondensate components to find the collective mode frequencies and collisional damping rates as a function of temperature .
our calculations extend those of gury - odelin and stringari at @xmath1 and in the normal phase .
they complement the numerical results of jackson and zaremba , although landau damping is left out of our approach .
our results are also used to calculate the quadrupole response function , which is related to the moment of inertia .
it is shown explicitly that the moment of inertia of a trapped bose gas at finite temperatures involves a sum of an irrotational component from the condensate and a rotational component from the thermal cloud atoms . |
it has been established that many - body processes play an important role in the _ transient _ optical response of semiconductors in the coherent regime@xcite .
the coulomb correlations between photoexcited carriers are especially strong in the presence of magnetic field . by suppressing kinetic energy of electrons and holes in two spatial dimensions ( magnetic confinement ) ,
a high magnetic field enhances the relative strength of interactions between them@xcite . in bulk semiconductors , a dominant role of coulomb correlations in magnetic field
was demonstrated in four - wave mixing ( fwm ) spectroscopy experiments@xcite .
for example , a huge ( several orders of magnitude ) enhancement of the fwm signal was observed as the field exceeded certain characteristic value . a crossover to
strongly - correlated regime occurs when the magnetic length , @xmath0 , becomes smaller than the excitonic bohr radius , @xmath1 . in quantum wells ( qw ) in perpendicular magnetic field ,
the energy spectrum is discrete so one would expect even stronger effect of interactions on the optical response .
the linear absorption spectrum is dominated by a bound magnetoexciton ( mx ) state that incorporates electron and hole transitions between landau levels ( ll ) in conduction and valence band , respectively . in strong field , such that the cyclotron energy , @xmath2 , is much larger than characteristic interaction energy , @xmath3 ( here @xmath4 is the dielectric constant ) , the processes involving transitions between different ll s are suppressed , and the lowest mx state is comprised of @xmath5 ll electron - hole ( _ e - h _ ) pair with magnetic - field - dependent energy dispersion@xcite . however , owing to the _ e - h _
symmetry for any given ll , such mx s do not interact with each other due to a cancellation of coulomb matrix elements between electrons and holes@xcite . for this reason , the _ nonlinear _ optical response of @xmath5 mx s is similar to that of noninteracting two - level systems@xcite unless there is a sufficient _ e - h _ asymmetry because of , e. g. , differing band offsets or disorder@xcite . in the latter case ,
coulomb correlations become important for both pump - probe@xcite or fwm@xcite spectroscopy ; however , in undoped qw s , such an asymmetry is weak . in weaker magnetic field
( @xmath6 ) , when ll mixing is strong , the coherent optical response in qw s was studied in hartree - fock approximation ( ha ) within semiconductor bloch equations technique@xcite . here
we study the role of many - body correlations in coherent optical spectroscopy of qw mx s excited to _ upper _ ( @xmath7 ) ll s .
we focus on the case of a sufficiently strong magnetic field , @xmath8 , so that individual optically - excited mx , with binding energy @xmath9 , is comprised of a single ( e. g. , @xmath10 ) ll _ e - h _ pair . for such fields , the processes involving inter - ll transitions do not contribute to linear response even if optical frequency is tuned to excite interband transitions at upper ll s .
however , as we demonstrate below , coulomb correlations between _ e - h _ pairs excited to @xmath11 ll s are significant .
such correlations originate from auger processes which involve inter - ll scattering of _ two _ photoexcited electrons .
for example , two electrons on @xmath10 ll can scatter to @xmath5 and @xmath12 ll s , as illustrated in fig .
1 . since this is a resonant process ( ll s are equidistant ) , it does not depend on the ll separation and , therefore , can take place even in a strong field , @xmath13 .
the inter - ll auger processes has been previously observed in luminescence experiments@xcite .
schematic representation of auger scattering of magnetoexcitons .
two _ e - h _ paires are excited by a pump pulse to @xmath10 ll ( a ) undergo resonant auger processes involving electrons ( b ) and holes ( c ) .
, width=384 ] we show that the auger processes play the dominant role in the nonlinear spectroscopy of qw s in strong magnetic field . since they involve inter - ll scattering of charged carriers , the _ e - h _ symmetry no longer holds which gives rise to interactions between mx s .
it should be emphasized that since the ll s are discrete , the amplitude of auger process can be large and , in fact , is restricted mainly by the inhomogeneous@xcite or homogeneous ( due to phonons ) ll broadening .
in fact , in strong field , the relevant energy scale , @xmath3 , is set by interactions , so that an adequate description of the nonlinear optical response should treat the auger processes nonperturbatively .
for example , the simple process described above ( fig .
1 ) is followed by auger scattering of the electrons _ back _ to @xmath10 ll a process completely irrelevant in the luminescence .
such _ multiple _
auger processes , by effectively causing interaction between mx s , give rise to _ coherent _ signature in the fwm spectroscopy .
we perform numerical calculations of fwm polarization in strong field by including auger - scattering _ exactly _ , in all orders of perturbation theory .
we find a strong enhancement as well as oscillations of the fwm signal for negative time delays .
these oscillations are identified as quantum beats originating from the interference between four - particle correlated states that are related to each other via auger processes . in section [ gen ]
to we outline the formalism for evaluating the nonlinear polarization and compute the relevant matrix elements . in section [ auger ]
we consider the contribution of the auger processes into polarization . in section [ num ]
we show the results of numerical calculations for the case of @xmath10 ll .
section [ conc ] concludes the paper .
we consider a 2d system in strong perpendicular magnetic field with two - band hamiltonian @xmath14 , where @xmath15 is free two - band hamiltonian and @xmath16 is the coulomb potential . in the fwm spectroscopy
, the sample is subjected to the probe and pump laser pulses , separated by time delay @xmath17 , with electric field intensities @xmath18 and @xmath19 and wave vectors @xmath20 and @xmath21 , respectively , and the signal along the direction @xmath22 is measured . in order to obtain the third - order optical response , we adopt the formalism of ref .
@xcite generalized to the case of arbitrary ll .
the third - order fwm optical polarization has the form @xmath23
with@xcite @xmath24.\end{aligned}\ ] ] where @xmath25 is the the interband dipole matrix element , @xmath26 is the laser central frequency ( we work in rotating frame ) , and @xmath27 is the ground state of @xmath28 . here
@xmath29 is the interband transition operator ( @xmath30 and @xmath31 are electron and hole annihilation operators , respectively ) , while the state @xmath32 stands for @xmath33 where the _ single _ _ e - h _ pair state @xmath34 and the _ two _ _ e - h _ pair state @xmath35 satisfy the equations @xmath36 and @xmath37 respectively ; these equations describe the time - evolution ( governed by the hamiltonian @xmath28 ) of a single - exciton and two - exciton states@xcite .
the polarization ( [ pol - fwm ] ) is determined by the time - dependence of relevant one and two - pair states . in order to solve eqs .
( [ eqp],[eqw ] ) , we use the standard basis for an electron in @xmath38th and a hole in @xmath39th lls@xcite , @xmath40 where @xmath41 is the center - of - mass momentum of e - h pair , @xmath42 , @xmath43 are the relative and the average coordinates , respectively , @xmath44 is the ll degeneracy , and @xmath45 is the unit vector in the direction of the magnetic field . here ,
@xmath46 and @xmath47 are the landau wave - functions for electron and hole , and @xmath48 is given by @xmath49 and @xmath50 for @xmath51 , where @xmath52 is the laguerre polynomial and @xmath53 is the complex coordinate .
note also relations @xmath54 . in this basis , the single - pair amplitude , @xmath55 can be easily found from eq .
( [ eqp ] ) .
the matrix elements of the coulomb potential , @xmath16 , have the form @xmath56 where @xmath57 is the complex momentum and we used the identity @xmath58 from eq .
( [ eqp ] ) , we then obtain @xmath59 , where @xmath60 is the linear polarization , due to the pump , of the @xmath39th ll , satisfying @xmath61 where @xmath62 is the detuning ( @xmath63 is the bandgap ) .
turning to the two - pair states , we introduce the amplitude @xmath64 where a complete orthogonal two - exciton basis in the hilbert subspace with zero total momentum is constructed from symmetrized product of single - pair states : @xmath65,\ ] ] the corresponding matrix elements of coulomb potential can be explicitly calculated as , @xmath66 \nonumber\\ & & + \frac{1}{l^2}\,v_{mn , m'n';m_1n_1,m'_1n'_1}({\bf p},{\bf q}).\end{aligned}\ ] ] where symmetrization with respect to @xmath67 is implicit hereafter .
the last term describes mx - mx interaction accompanied by the momentum exchange @xmath68,\end{aligned}\ ] ] where in the derivation we used the relations @xmath69 equation for two - mx amplitude @xmath70 is obtained by projecting eq .
( [ eqw ] ) onto two - mx basis states , @xmath71 where the last ( source ) term can be found using eq .
( [ eqp ] ) , @xmath72 ( n\delta_{{\bf p}0}\delta_{mn}\delta_{m'n'}-\delta_{mn'}\delta_{m'n}),\ ] ] where we used the expansion @xmath73 ( the second term comes from the exchange ) . in the following
we will be interested in the exciton - exciton interactions contribution into the fwm polarization .
correspondingly , we separate out the interactions - induced contribution by writing @xmath74 , where the first term , corresponding to non - interacting excitons , after being combined with the second term of ( [ tfwm ] ) , gives the pauli blocking contribution of non - interacting excitons , and will be included in the numerical calculations below . the second term of the above decomposition gives the exciton - exciton interaction contribution to the polarization , @xmath75,\ ] ] with @xmath76 where the function @xmath77 describes the propagation of an _ e - h _ pairs , created by pump and probe pulses , in the fwm direction , and can be expressed via the time - evolution operator @xmath78 as @xmath79 with the matrix elements @xmath80 satisfying @xmath81 and @xmath82 .
putting all together , we obtain @xmath83 where @xmath84 substituting the above decomposition @xmath85 into eq .
( [ w - eq ] ) , where the non - interacting term has the form @xmath86 we obtain the following equation for @xmath87 : @xmath88 where the matrix elements of the hamiltonian were calculated above and the source term is given by @xmath89 \nonumber\\ \times ( n\delta_{{\bf p}0}\delta_{mn}\delta_{m'n'}-\delta_{mn'}\delta_{m'n}),\end{aligned}\ ] ] with @xmath90 .
substitution of eqs .
( [ v - two ] ) and ( [ w0 ] ) into ( [ j ] ) yields a closed - form but rather complicated expression for @xmath91 . in the following we will need to solve eqs . ( [ wxx - eq ] ) only for certain ll s corresponding to the auger processes .
we consider the case of strong magnetic field so that @xmath92 . in this case
all _ single - pair _ processes involving inter - ll transitions are suppressed .
for example , neglecting non - diagonal coupling , eq .
( [ p - eq ] ) simplifies to @xmath93 and similarly the single - pair evolution operator has only diagonal matrix elements , @xmath94 with @xmath95 where @xmath96 is the mx homogeneous width . in the case
when two pairs are excited , inter - ll scattering becomes possible via auger processes ( see fig .
1 ) . let the pump pulse , tuned to @xmath39th ll , excite two _ e - h _ pairs .
then the electrons can auger - scatter with each other to @xmath97 and @xmath98 ll s , and holes can auger - scatter with each other to @xmath99 and @xmath100 ll s .
since these are the only _ resonant _ processes , all other inter - ll processes are suppressed .
it is now convenient to introduce new notations which refer to the ll number to which the _ e - h _ pairs were initially excited , @xmath101 and @xmath102 \bigr],\end{aligned}\ ] ] where @xmath103 and @xmath104 for @xmath105 .
since all non - auger inter - ll transitions are suppressed , we restrict ourselves only with intermediate states related to each other via auger processes .
then the relevant two - mx amplitudes @xmath106 satisfy the following system @xmath107 w_{n{\bf p}}^{\alpha \beta}(t ) % \nonumber\\ = \sum_{\alpha ' \beta'}\int\frac{d{\bf q}}{(2\pi)^2 } v_n^ { \alpha \beta,\alpha ' \beta'}({\bf p},{\bf q } ) w_{n{\bf q}}^{\alpha ' \beta'}(t ) + m_{n{\bf p}}^{\alpha \beta}(t).\end{aligned}\ ] ] note that @xmath108 .
the source term @xmath109 can be calculated from eq .
( [ j ] ) as @xmath110 \nonumber\\ \times \frac{1}{2}\bigl[p_{n+\alpha}(t)p_{n-\alpha}(t ) + p_{n+\beta}(t)p_{n-\beta}(t)\bigr],\end{aligned}\ ] ] where we used @xmath111 . in the absence of auger scattering ( @xmath112 )
, we have @xmath113 , i.e. interactions do not contribute to the fwm signal ( in the ideal 2d case ) . using that @xmath114
is related to its fourier transform as @xmath115 , we obtain from eq .
( [ chi ] ) that @xmath116 using eqs ( [ chi - auger ] ) and ( [ k - auger ] ) , the fwm polarization can be easily expressed in terms of amplitudes @xmath117 . in the following ,
we consider the case when the spectral width of the pulse is smaller than the ll separation . in this case
, we have @xmath118 , and the fwm polarization takes the form ( restoring pauli blocking contribution ) @xmath119 with @xmath120,\ ] ] and @xmath121 \nonumber\\ % & & + \mu^3 p_n(t ) \int_{-\infty}^{t}dt^{\prime}{\cal e}_{2}(t ' ) \int_{-\infty}^{t^{\prime}}dt '' { \cal e}_{1}(t '' ) e^{-i\omega_0(t+t'')+i(\omega_n+e_{n0}^{00})(t'-t '' ) -\gamma|t'-t''|},\end{aligned}\ ] ] where @xmath122 and @xmath123 is the pump - induced polarization on @xmath39th ll .
we consider the case when the central laser frequency is tuned to interband transition between @xmath10 ll s . in this case , two exciton excited by the pump to @xmath10 ll , can auger - scatter only to @xmath5 and @xmath12 ll s , i.e. , @xmath124 ( see fig .
1 ) . system ( [ w - eq - auger ] ) then takes the form : @xmath125 = m_{1p}^{01}(t ) , % \nonumber\end{aligned}\ ] ] where the source term and inter - pair coulomb matrix elements are given by @xmath126 \frac{p_1 ^ 2(t)}{2 } , \nonumber\\ & & v_1^{00}({\bf p},{\bf q})=-4\sin^2({\bf p\times q}l^2/2 ) e^{-|{\bf p - q}|^2l^2/2}v_{|{\bf p - q}|}\bigl[l_1(|{\bf p - q}|^2l^2/2)\bigr]^2 , \nonumber\\ & & v_1^{11}({\bf p},{\bf q})=-4\sin^2({\bf p\times q}l^2/2 ) e^{-|{\bf p - q}|^2l^2/2}v_{|{\bf p - q}|}l_2(|{\bf p - q}|^2l^2/2 ) , \nonumber\\ & & v_1^{1 - 1}({\bf p},{\bf q})= -e^{-|{\bf p - q}|^2l^2/2}v_{|{\bf p - q}| } \biggl [ 4\sin^2({\bf p\times q}l^2/2)l_2(|{\bf p - q}|^2l^2/2 ) \nonumber\\ & & \mbox{\hspace{2.5 in } } + \bigl[1-l_2(|{\bf p - q}|^2l^2/2)\bigr]^2 \biggr ] , \nonumber\\ & & v_1^{01}({\bf p},{\bf q})= -e^{-|{\bf p - q}|^2l^2/2}v_{|{\bf p - q}| } \biggl [ 2\sin^2({\bf p\times q}l^2/2 ) \bigl[l_2(|{\bf p - q}|^2l^2/2 ) + \bigl[l_1(|{\bf p - q}|^2l^2/2)\bigr]^2\bigr ] \nonumber\\ & & \mbox{\hspace{1 in } } + \bigl[1-l_1(|{\bf p - q}|^2l^2/2)\bigr ] \bigl[l_1(|{\bf p - q}|^2l^2/2)-l_2(|{\bf p - q}|^2l^2/2)\bigr ] \biggr ] , \nonumber\\ & & \tilde{v}_1^{01}({\bf p},{\bf q})= 2\cos({\bf p\times q}l^2 ) e^{-|{\bf p - q}|^2l^2/2}v_{|{\bf p - q}| } \frac{|{\bf p - q}|^2l^2}{4}l_1 ^ 1(|{\bf p - q}|^2l^2/2 ) , % \nonumber\end{aligned}\ ] ] with @xmath127 , while the single - exciton energies @xmath128 have the following form @xmath129 , \nonumber\\ & & e_{1p}^{-1 - 1}=-\sqrt{\frac{\pi}{2}}\frac{e^2}{l } \phi\bigl(\frac{1}{2},1,-\frac{p^2l^2}{2}\bigr ) , \nonumber\\ & & e_{1p}^{11}=-\sqrt{\frac{\pi}{2}}\frac{e^2}{l } \biggl[\phi\bigl(\frac{1}{2},1,-\frac{p^2l^2}{2}\bigr ) -2\phi\bigl(\frac{3}{2},1,-\frac{p^2l^2}{2}\bigr ) + \frac{15}{4}\phi\bigl(\frac{5}{2},1,-\frac{p^2l^2}{2}\bigr ) \nonumber\\ & & \mbox{\hspace{2 in } } -\frac{15}{4}\phi\bigl(\frac{7}{2},1,-\frac{p^2l^2}{2}\bigr ) + \frac{105}{64}\phi\bigl(\frac{9}{2},1,-\frac{p^2l^2}{2}\bigr ) \biggr ] , \nonumber\\ & & e_{1p}^{01}=-\sqrt{\frac{\pi}{2}}\frac{e^2}{l } \biggl[\phi\bigl(\frac{1}{2},1,-\frac{p^2l^2}{2}\bigr ) -\frac{3}{2}\phi\bigl(\frac{3}{2},1,-\frac{p^2l^2}{2}\bigr ) + \frac{15}{8}\phi\bigl(\frac{5}{2},1,-\frac{p^2l^2}{2}\bigr ) \nonumber\\ & & \mbox{\hspace{3.5 in } } -\frac{15}{16}\phi\bigl(\frac{7}{2},1,-\frac{p^2l^2}{2}\bigr ) \biggr ] , \nonumber\\ & & e_{1p}^{0 - 1}=-\sqrt{\frac{\pi}{2}}\frac{e^2}{l } \biggl[\phi\bigl(\frac{1}{2},1,-\frac{p^2l^2}{2}\bigr ) -\frac{1}{2}\phi\bigl(\frac{3}{2},1,-\frac{p^2l^2}{2}\bigr ) \biggr ] , \nonumber\\ & & e_{1p}^{1 - 1}=-\sqrt{\frac{\pi}{2}}\frac{e^2}{l } \biggl[\phi\bigl(\frac{1}{2},1,-\frac{p^2l^2}{2}\bigr )
-\phi\bigl(\frac{3}{2},1,-\frac{p^2l^2}{2}\bigr ) + \frac{3}{8}\phi\bigl(\frac{5}{2},1,-\frac{p^2l^2}{2}\bigr)\biggr ] , \nonumber\\ & & \tilde{e}_{1p}^{01}=-\sqrt{\frac{\pi}{2}}\frac{e^2}{l } \biggl[2\phi\bigl(\frac{3}{2},1,-\frac{p^2l^2}{2}\bigr ) -\frac{3}{2}\phi\bigl(\frac{5}{2},1,-\frac{p^2l^2}{2}\bigr ) \biggr ] , \nonumber\end{aligned}\ ] ] @xmath130 being the confluent hypergeometric function , and the pump - induced polarization is given by @xmath131 the first equation in systems ( [ eq - n=1 ] ) describes the time - evolution of two _ e - h _ pairs , excited to @xmath10 ll by the pump pulse [ fig . [ fig1](a ) ] , with and without inter - ll scattering ( second and third terms , respectively ) .
the third and second equations in ( [ eq - n=1 ] ) describe the similar time evolution of the state ( c ) in fig .
[ fig1 ] and of the state obtained from ( c ) by exchanging the holes in the valence band .
the last terms of the first three equations in ( [ eq - n=1 ] ) describe the auger scattering that relates the corresponding states to the state ( b ) in fig .
[ fig1 ] , described by the fourth equation .
note that only state ( b ) lacks the _ e - h _ symmetry , i. e. , does not transform into itself under replacement of electrons by hole and vice - versa in fig .
[ fig1 ] , as indicated by the source term in the rhs of the forth equation in ( [ eq - n=1 ] ) .
its amplitude , @xmath132 , plays the role of the source for the rest of the system ( [ eq - n=1 ] ) ( last terms ) and thus represents the sole source for the interaction - induced polarization . in the numerical calculations below
, we consider resonant excitation , i.e. , central frequency tuned at the transitions between ground state and @xmath10 mx with binding energy @xmath133 where @xmath134 is the binding energy of the @xmath5 mx .
the above equations for @xmath135 have been solved numerically using fourth - order runge - kutte routine .
fwm polarization was then calculated from eqs .
( [ fwm - gen]-[pb ] ) with @xmath10 , for gaussian pump and probe pulses with the duration @xmath136 separated by time delay @xmath17 .
dephasing due to electron - phonon interactions have been accounted for by mx width @xmath96 and by the width @xmath137 characterizing the dephasing of two - pair system .
we emphasize that in the case of interacting excitons , @xmath137 does not necessarily equal @xmath138 , as has been pointed out in experiment @xcite . in fig .
[ fig2 ] we show calculated time - integrated fwm signal , @xmath139 for parameter values @xmath140 ,
@xmath141 , @xmath142 versus time delay ( in units of @xmath134 ) .
the dot - dashed curve corresponds to pauli blocking contribution that comes from non - interacting excitons .
it should be emphasized that , in strong magnetic field , the _ only _ interaction - induced contribution comes from auger scattering , so the hartree - fock contribution is suppressed by a small parameter @xmath143 .
it can be seen that auger scattering of magnetoexcitons strongly enhances the amplitude of fwm signal . for negative time delays
, ti - fwm signal develops an exponential ( with decay - time @xmath144 ) tail that is characteristic for interacting excitons .
furthermore , ti - fwm signal exhibits two sets of oscillations superimposed on each other .
these oscillations represent , in fact , quantum beats between four - particle configurations contributing to fwm polarization .
we identify more the pronounced oscillations with the interference between states excited state ( a ) and state ( b ) in fig .
[ fig1 ] related to ( a ) by electron auger - scattering in conduction band .
these states are characterized by coulomb energies [ see eqs .
( [ eq - n=1 ] ) and ( [ en - n=1 ] ) ] @xmath145 and @xmath146 , so the oscillations period corresponds to the energy difference @xmath147 .
the weaker oscillations originate from the interference between states ( a ) and ( c ) in fig .
[ fig1 ] , where the state ( c ) is characterized by energy @xmath148 , so the period is determined by @xmath149 .
note that the state ( d ) , obtained from ( c ) by hole exchange in the valence band , is characterized by the energy @xmath150 and the corresponding period , determined by @xmath151 , is too long to be noticeable in fig .
it should be emphasized , however , that the above estimates involve zero - momentum energies of each of the pairs that constitute four - particle correlated state .
in fact , auger processes are accompanied by a momentum exchange between mx s , so only the total momentum of a four particle correlation is zero .
this momentum exchange leads to the damping of oscillations . since ( a ) and ( c )
are related via two auger processes ( in conduction and in valence band ) the additional momentum exchange during the hole auger - scattering in valence band leads to a much much stronger damping of ( a)-(c ) oscillations as compared to ( a)-(b ) oscillations .
calculated ti - fwm signal versus time delay ( in units of magnetoexciton energy ) shows oscillations corresponding to the interference between states ( a ) and ( b ) , and between states ( a ) and ( c ) in fig .
, width=384 ]
we investigated the role of inter - ll transitions in the coherent dynamics of quantum well excitons in strong magnetic field . while on the lowest ll , the suppression of inter - ll transitions results in the absence of exciton - exciton interactions , on higher ll levels the interactions become dominant due to resonant auger processes .
the coherent signature of exciton auger - scattering can be traced in the fwm polarization as quantum beats corresponding to the interference between optically - inactive four - particle correlated states .
author thanks i. e. perakis for useful discussions and n. primozich for help in numerical calculations .
this work was supported by national science foundation under grants dmr-0304036 and dmr-0305557 , and by army research laboratory under grant daad19 - 01 - 2 - 0014 . | we study theoretically the ultrafast nonlinear optical response of quantum well excitons in a perpendicular magnetic field .
we address the role of many - body correlations originating from the electron scattering between landau levels ( ll ) . in the linear optical response , the processes involving inter - ll transitions are suppressed provided that the magnetic field is sufficiently strong .
however , in the _ nonlinear _ response , the auger processes involving inter - ll scattering of _ two _ photoexcited electrons remain unsuppressed .
we show that auger scattering plays the dominant role in the coherent exciton dynamics in strong magnetic field .
we perform numerical calculations for the third - order four - wave - mixing ( fwm ) polarization which incorporate the auger processes nonperturbatively . we find that inter - ll scattering leads to a strong enhancement and to oscillations of the fwm signal at negative time delays .
these oscillations represent quantum beats between optically - inactive two - exciton states related to each other via auger processes . |
statistical thermodynamics of simple liquids and their mixtures has been extensively studied , and thermodynamical and structural properties of such systems are well understood @xcite . in particular ,
an accurate equation of state of the lennard - jones fluid has been obtained @xcite .
the impressive development of the theory was possible thanks to the key contributors including prof .
tomas boublik and prof .
ivo nezbeda .
in contrast , the statistical thermodynamics of the so called soft matter systems is much less developed , and recently these systems draw increasing attention .
complex molecules , nanoparticles , colloid particles or polymers in various solvents interact with effective potentials that may have quite different forms . when the shape of the effective potential resembles the shape of interactions between atoms or simple molecules , then analogs of the gas - liquid and liquid - solid transitions occur @xcite . if , however , there are competing tendencies in the interactions , then instead of the gas - liquid transition or separation of the components , a self - assembly or a microsegregation may be observed @xcite .
the competing interactions can have quite different origin and form .
one important example of competing interactions is the so called short - range attraction ( sa ) , and long - range repulsion ( lr ) salr potential @xcite , consisting of a solvent - induced short - range attraction and long - range repulsion that is either of electrostatic origin , or is caused by polymeric brushes bound to the surface of the particles .
the attraction favours formation of small clusters .
because of the repulsion at large distances , however , large clusters are energetically unfavourable . for increasing concentration of the particles elongated clusters and a network
were observed in both experiment and theory @xcite .
competing interactions of a quite different nature are present in systems containing amphiphilic molecules such as surfactants , lipids or diblock copolymers @xcite .
amphiphilic molecules are composed of covalently bound polar and organic parts , and in polar solvents self - assemble into spherical or elongated micelles , or form a network in the sponge phase .
in addition , various lyotropic liquid crystal phases can be stable @xcite . despite of very different origin and shape of the interaction potentials ,
very similar patterns occur on the mesoscopic length scale in the systems interacting with the isotropic salr potential , and in the amphiphilic solutions with strongly anisotropic interactions @xcite . the particles interacting with the salr potential self - assemble into spherical or elongated clusters or form a network , whereas the amphiphiles self - assemble into spherical or elongated micells or form the sponge phase .
the distribution of the clusters or the micelles in space and the transitions between ordered phases composed of these objects are very similar .
the origin of the universal topology of the phase diagrams in the amphiphilic and salr systems was studied in ref.@xcite .
it has been shown by a systematic coarse - graining procedure that in the case of weak order the colloidal and the amphiphilic self - assembly can be described by the same landau - brazovskii functional @xcite .
the landau - brazovskii functional was first applied to the block - copolymers by leibler in 1980 @xcite .
later functionals of the same type were applied to microemulsions @xcite .
the landau - brazovskii -type functional , however , is appropriate only for weak order , where the average density and concentration are smooth , slowly varying functions on the mesoscopic length scale .
moreover , in derivation of the functional various assumptions and approximations were made .
further approximations are necessary in order to obtain solutions for the phase diagram , equation of state and correlation functions .
thus , the question of universality of the pattern formation on the mesoscopic length scale , particularly at low temperatures , is only partially solved .
we face two types of problems when we want to compare thermodynamic and structural properties in different self - assembling systems in the framework of statistical thermodynamics .
first , one has to introduce generic models with irrelevant microscopic details disregarded .
second , one has to make approximations to solve the generic models , or perform simulations .
it is not obvious a priori how the assumptions made in construction of the model and the approximations necessary for obtaining the solutions influence the results . in the case of simulations the simulation box should be commensurate with the characteristic size of the inhomogeneities that is to be determined .
it is thus important to introduce generic models for different types of self - assembly that can be solved exactly .
exact solutions can be easily obtained in one - dimensional models , but there are no phase transitions in one dimension for temperatures @xmath0 .
nevertheless , the ground state ( gs ) can give important information about energetically favorable ordered structures , and pretransitional ordering for @xmath0 can be discussed based on exact results for the equation of state , correlation function and specific heat .
a generic one - dimensional lattice model for the salr potential was introduced and solved exactly in ref.@xcite . in this model
the nearest - neighbors ( nn ) attract each other , and the third neighbors repel each other .
it is thus energetically favorable to form clusters composed of 3 particles separated by at least 3 empty sites .
the gs is governed by the repulsion - to - attraction ratio @xmath1 and by the chemical potential of the particles .
an interesting property of the gs is strong degeneracy at the coexistence of the ordered cluster phase with the gas or liquid phases . due to this degeneracy
the entropy per site does not vanish .
the collection of the microstates present for @xmath2 at the coexistence between the periodic and the gas or the liquid phases can be interpreted as a disordered cluster or bubble phase respectively . for @xmath0 pseudo phase transitions between the gas and the periodically distributed clusters , and between the periodically distributed clusters and the dense liquid phases were obtained .
the equation of state ( eos ) has a characteristic shape that is significantly different from the eos in simple fluids .
a one dimensional lattice model for amphiphiles in solvents attracting one of the two parts of the amphiphilic molecule was introduced in ref.@xcite .
the two models , one for the salr and the other one for the amphiphilic system , are defined on the same level of coarse - graining , therefore by comparing the exact results we can draw conclusions on similarities of self - assembly in these systems , and the origin of these similarities . in the model introduced in ref.@xcite
the gs is strongly degenerate at the phase coexistence between the pure solvent and periodically distributed bilayers , and the entropy per site does not vanish .
thus , the gs of the two models show remarkable similarity , despite quite different interaction potentials .
this suggests similar origin of the formation of disordered phases with mesoscopic inhomogeneities in various systems with competing interactions . in this work we solve exactly the model introduced in ref.@xcite by the transfer matrix method .
we describe the model and its ground state in sec.2 . the transfer matrix and exact expressions for the grand potential , density and correlation function
are given in sec.3 .
in sec.4 and 5 we present our results for the eos and the correlation function . in sec.6
we calculate the specific heat with fixed chemical potential in the grand canonical ensemble , and by using thermodynamic relations we obtain the specific heat with fixed concentration . to check our exact calculations we also perform monte carlo simulations and compute the specific heat with fixed concentration directly in the canonical ensemble .
our additional purpose is to verify the results of the simulations performed in a finite system by comparison with the exact results obtained in the thermodynamic limit .
we cross - check the exact and the simulation results for the specific heat , because in simulations of the 2- or 3-dimensional self - assembling systems the peak in the specific heat is interpreted as a signature of a phase transition@xcite .
it is thus worthwhile to compare the simulation and the exact results for this important quantity . in sec.7 our results for the mechanical , structural and thermal properties
are compared with the corresponding results obtained in ref.@xcite for the salr model .
the amphiphilic molecules consist of a hydrophilic head and a hydrophobic tail , therefore the interactions between them depend on orientations . in the case of a two - component mixture of a polar solvent and amphiphilic molecules , for example lipids ,
we assume that the solvent molecules attract the polar head , and effectively repel the hydrophobic tail of the amphiphilic molecule .
we neglect orientational degrees of freedom of the solvent molecules . in a one - dimensional model
the continuum of different orientations of amphiphiles is reduced to just two orientations ( fig.1 ) .
we assume that the molecules occupy lattice sites , and the lattice constant @xmath3 is of order of the length of the amphiphilic molecule in this model . if the solvent molecules are much smaller than the amphiphilic molecules , we assume that the site is occupied by a cluster of several solvent molecules . we assume nearest - neighbor interactions .
the absolute value of the energy of two clusters of solvent molecules that occupy the nearest - neighbour sites , @xmath4 , is taken as the energy unit .
we assume that the interaction between the cluster of solvent molecules and the amphiphilic molecule in the favorable ( unfavorable ) orientation is @xmath5 ( @xmath6 ) , and the interaction between two amphiphilic molecules in the favorable and unfavorable orientation is@xmath7 and @xmath8 respectively .
the orientations of two amphiphilic molecules are favorable when they are oriented either head - to - head or tail - to - tail .
the neighborhood of the polar head and the hydrophobic tail is unfavorable .
the energies of different pairs of occupied sites are shown in fig .
[ fig_energies ] .
the model is similar to a lattice model of ternary oil - water - surfactant mixtures introduced in ref.@xcite and to a continuous model of binary mixtures with amphiphiles @xcite .
different values of the parameters @xmath9 may correspond to different particular mixtures . in this work
we are interested in general aspects of the amphiphilic self - assembly , especially in similarities between ordering on the mesoscopic length scale in the amphiphilic and colloidal systems , and in origin of these similarities .
for this reason we shall not try to fit the model parameters to any particular mixture .
a representative example of a system described by this model is a mixture of lipids and water .
we should stress that water is a complex liquid @xcite , and micro - heterogeneities are present in aqueous solutions of polar molecules @xcite .
however , on the mesoscopic length scale of tens or hundreds of nanometers the ordering of the water molecules plays a subdominant role . we introduce the microscopic densities
@xmath10 with @xmath11 denoting the cluster of solvent molecules , and the amphiphile with the head on the left and on the right respectively .
@xmath12 when the site @xmath13 is in the state @xmath14 and @xmath15 otherwise .
multiple occupancy of the lattice sites is excluded .
we further restrict our attention to the liquid phase and assume close - packing , @xmath16 up to a state - independent constant the hamiltonian of an open system , with the chemical - potential contribution included , can be written in the form @xmath17/b=
\frac{1}{2}\sum_{x=1}^l\sum_{x'=1}^l\hat\rho_i(x)v_{ij}(x , x')\hat\rho_j(x')-\sum_{x=1}^l\mu\hat\rho_1(x),\ ] ] where the summation convention for repeated indices is used , @xmath18 is the system size , @xmath19 , @xmath20 is the chemical potential of the solvent , and the chemical potential of amphiphiles , @xmath21 , is independent of orientations of the molecules .
we assume periodic boundary conditions , @xmath22 . according to the above discussion of interactions the interaction potential @xmath23 is @xmath24 \label{bfv}\end{aligned}\ ] ] and @xmath25 , @xmath26 for @xmath27 . in the liquid phase we can neglect density fluctuations ( eq.([1 ] ) ) , hence @xmath28 and there are two independent densities . in a disordered phase @xmath29 , and @xmath30 with @xmath31 denoting the average amphiphile concentration . at @xmath2
the stable structure corresponds to the global minimum of the hamiltonian .
apart from the solvent - rich and amphiphile - rich phases we find stability of the periodic phase where amphiphilic bilayers are separated by layers of solvent . in the amphiphile - rich phase
the molecules are oriented head - to - head and tail - to - tail when @xmath32 . in the periodic phase a solvent - occupied site
is followed by @xmath33 sites occupied by properly oriented amphiphilic molecules .
the coexistence lines obtained by equating @xmath34 in these phases are @xcite @xmath35 the @xmath36 ground state is shown in fig . [ fig_gs ] .
the solvent - amphiphile coexistence occurs for small values of @xmath37 .
all the three phases coexist at the triple line @xmath38 and @xmath39 .
when @xmath40 the periodic structure of solvent - separated bilayers may be present for some range of @xmath41 and the sequence of the stable phases for increasing @xmath41 at fixed @xmath37 and @xmath42 is : amphiphiles - bilayers - solvent . at the solvent - periodic phase coexistence
the separation @xmath43 between the bilayers can be arbitrary , because @xmath34 is independent of @xmath43 for @xmath41 given by eq.([gsceox])b @xcite .
thus , the ground state is strongly degenerate and the entropy per site does not vanish .
similar degeneracy occurs at the periodic - amphiphile phase coexistence . at the periodic - amphiphile coexistence
the separation between the solvent occupied sites is @xmath44 with arbitrary @xmath45 , because when eq.([gsceox])c holds , @xmath34 is independent of @xmath45 @xcite .
note that the arbitrary separation between the bilayers at the coexistence between the solvent and the periodic phase signals vanishing surface tension between the two phases .
similar degeneracy of the ground state was found earlier for the lattice model of microemulsion @xcite .
the very low surface tension at the coexistence between the microemulsion and the water - rich phases was attributed to the amphiphilic nature of surfactant molecules .
( panel a ) and in the plane @xmath46 for @xmath47 ( panel b ) .
the surfaces in panel a represent coexistence between the phases , while the black dashed line represents the triple points where all the three phases coexist .
the triple point for @xmath48 is at @xmath49 . on the b panel
a schematic illustration of the three phases is shown in the insets inside the region of stability of each phase . at the coexistence with the solvent - rich phase the layers of solvent , @xmath43 ,
can have an arbitrary length , and at the coexistence with the amphiphile - rich phase the layers of amphiphilic molecules can have the thickness @xmath44 with arbitrary @xmath45 .
the dashed line corresponds to @xmath50 for which the eos , the correlation function and the specific heat are calculated in the following sections . ]
in this section we introduce the transfer matrix and develop exact expressions for the grand thermodynamic potential , the average density of each component and for the correlation function .
the elements of the transfer matrix @xmath51 are given by @xmath52 where @xmath53 is 1 for @xmath54 and 0 otherwise , and @xmath55 with @xmath56 denoting the boltzmann constant .
the partition function for a system with periodic boundary conditions is @xmath57 \right\}.\ ] ] where @xmath58 denotes the microscopic state at the site @xmath13 , and is transverse to the columnar vector @xmath59 . at each lattice site
there can be one of the 3 microscopic states @xmath60 , @xmath61 , or @xmath62 . in terms of the transfer matrix @xmath63
takes the form @xmath64 where @xmath65 is the eigenvalue of the transfer matrix .
if we denote @xmath66 , the partition function for the system size @xmath67 takes the even simpler asymptotic form @xmath68 in the thermodynamic limit the grand potential in @xmath69 units , @xmath70 , is given by the exact formula @xmath71 where @xmath72 is 1 dim .
pressure in @xmath73 units .
the average density of the @xmath14-th state is independent of @xmath13 because of the translational invariance , and is given by @xmath74 \right\}.\end{aligned}\ ] ] if we change the basis of * t * with a help of the invertible matrix * p * such that @xmath75 is diagonal , then the average density in thermodynamic limit is given by the following expression : @xmath76 the correlation function @xmath77 between two sites in the same state @xmath14 separated by @xmath45 sites is given by @xmath78 where @xmath79 { \bf t}^{l - n}[{\hat{\rho}}_i(n+1){\hat{\rho}}_i(1)].\end{aligned}\ ] ] we change the basis to the one in which * t * is diagonal , take the thermodynamic limit and obtain the exact formula , @xmath80 where @xmath81 from ( [ rr ] ) and ( [ g ] ) we obtain the correlation function @xmath82 eq .
( [ corr_func ] ) can be further simplified for @xmath83 .
in such a case we can neglect the smallest components of the sum in eq.([corr_func ] ) . if the second largest ( in the absolute value ) eigenvalue @xmath84 is a pure real number , then @xmath85 , but
if the @xmath84 has a non - zero imaginary part , then we have to take into account also the eigenvalue @xmath86 , complex conjugate to @xmath87 .
let us introduce the notations @xmath88 and @xmath89 in terms of these parameters the correlation function for large separations between the particles in an infinite system takes the asymptotic form @xmath90
we choose stronger interactions between the amphiphilic than between the solvent molecules , @xmath50 . for such interaction parameters
the periodic phase is present on the gs ( see the dashed line in fig.2b ) . in fig.[eosm ] we show the concentration @xmath91 ( average density of the amphiphile ) as a function of the reduced chemical potential difference @xmath41 . note the rounded steps for @xmath92 and @xmath93 .
the steps occur for the values of @xmath41 corresponding to the gs phase transitions between the periodic and the amphiphile - rich or solvent - rich phases respectively . between the steps the plateaus for the three densities ,
@xmath94 occur . for increasing @xmath95
the @xmath96 lines become smoother , but the inflection points exist up to @xmath97 . , for @xmath50 at @xmath98 as a function of the reduced chemical potential difference @xmath41 . ] in eqs.([omega ] ) and ( [ avr ] ) the pressure and the average density are expressed in terms of @xmath95 and the reduced chemical potential difference @xmath41 . by eliminating @xmath41 we can obtain the dependence of the amphiphile density on @xmath72 .
we present @xmath99 in fig.[eos ] . , for @xmath50 at @xmath98 . as the energy unit
we choose the interaction @xmath69 between the two solvent molecules , and the length unit is the lattice constant @xmath3 .
we assume that @xmath3 is of order of the size of the amphiphilic molecules . in the case of lipids @xmath100 . ]
note that although there are no phase transitions in the strict sense in 1d , for low @xmath95 there is a rapid change in @xmath91 between @xmath101 and @xmath102 for a very small @xmath72 interval near @xmath103 , almost constant amphiphile density between @xmath103 and @xmath104 , and again a rapid change of @xmath91 between @xmath105 and nearly @xmath106 for @xmath104 .
this behavior suggests the pseudo - phase transitions between the amphiphile - rich and the periodic pseudo - phase with the density @xmath107 ( see fig.2b ) and next between the periodic pseudo - phase and the solvent - rich pseudo - phase .
for increasing temperature the changes of the slope of the @xmath99 line for increasing @xmath72 become smaller .
this result should be contrasted with the pressure - concentration dependence shown in fig.[cgsmall ] for the interaction parameters such that the periodic phase is not stable at the gs . , for @xmath108 @xmath109 and @xmath110 .
as the energy unit we choose the interaction @xmath69 between the two solvent molecules , and the length unit is the lattice constant @xmath3 .
we assume that @xmath3 is of order of the size of the amphiphilic molecules . in the case of lipids @xmath100 . ]
in the case of the periodic boundary conditions the system is translationally invariant , and the assembly into bilayers should be reflected in the shape of the correlation function . when the bilayers are formed , then the correlation function for the solvent
, @xmath111 should be negative for two subsequent values of @xmath13 , where the properly oriented amphiphilic molecules should appear with larger probability than the solvent molecule .
the exact results for @xmath111 , given in eq.([corr_func ] ) , are shown in fig.[cfm ] for @xmath112 , i.e. for the gs stability of the periodic phase , for a few rather high temperatures .
we can see the oscillatory decay of correlations , with the period 3 as in the case of the concentration in the gs periodic phase .
the decay length decreases with increasing temperature . only for short distances @xmath113 for two subsequent values of @xmath13 , however . in figs.[cft1 ] and [ cft2 ]
we show @xmath111 for very low @xmath95 and a few values of @xmath41 close to the gs coexistence between the periodic and the solvent- or amphiphile - rich phases .
very large correlation length , 3 orders of magnitude larger than the molecular size , can be seen inside the gs stability of the periodic phase .
the correlation length @xmath114 ( see eq.([xi ] ) ) and the period of the damped oscillatory decay ( see eq.([corr_func_ag ] ) ) are shown in fig.[xi_l ] .
we can see that @xmath115 beyond the gs stability of the periodic phase , i.e. for @xmath116 and @xmath117 . moreover , for @xmath118 the period of the damped oscillations is @xmath119 . for @xmath120 and @xmath121 .
] for @xmath120 and @xmath122 ( a ) , @xmath123 ( b ) , @xmath124 ( c ) , with @xmath124 being the value of the chemical potential at the coexistence between the periodic and the solvent - rich phases at the gs . ] for @xmath120 and @xmath125 ( a ) , @xmath126 ( b ) , @xmath127 ( c ) , with @xmath127 being the value of the chemical potential at the coexistence between the periodic and the amphiphile - rich phases at the gs . ] and the period of the oscillatory decay of correlations ( eqs.([xi ] ) and ( [ corr_func_ag ] ) ) for @xmath120 and @xmath128 . for @xmath127 and @xmath129
the periodic phase coexists with the amphiphile and the solvent respectively at @xmath2 . ] and the period of the oscillatory decay of correlations for @xmath130 at @xmath131 . for @xmath132 the amphiphile and the solvent coexist at @xmath133 . for @xmath134 the monotonic decay of correlations between the solvent molecules
is obtained . at @xmath135 the period of the oscillatory decay jumps from zero to infinity for @xmath136 . for @xmath137 the correlation function for the solvent exhibits an oscillatory decay for some range of @xmath41 . in the case of @xmath138
the oscillatory decay occurs for @xmath139 .
note the logarithmic scale on the vertical axis . ]
let us focus on the structure for the interaction parameters corresponding to the absence of the periodically distributed bilayers in the gs .
we choose @xmath140 and @xmath141 .
the correlation length @xmath114 and the period @xmath142 of the damped oscillatory decay for @xmath140 and @xmath141 , and different temperatures are shown in fig.[xi_l_nn ] .
for @xmath143 the eigenvalue @xmath87 is a pure real number for any value of the chemical potential , and the correlation function decays monotonically .
the change of the @xmath114 slope around @xmath144 at @xmath145 indicates the point where @xmath87 changes sign .
the period of the oscillatory decay jumps from zero to infinity at this point .
both cases correspond to the monotonic decay of correlations , but at this point ( @xmath144 and @xmath145 ) an oscillatory decay for @xmath137 and a range of @xmath41 around @xmath144 begins .
this is because for @xmath137 there are two discontinuities of the derivative of @xmath114 ( e.g. at @xmath146 and @xmath147 for @xmath138 ) . between the two points of discontinuity of @xmath148
, the @xmath87 is a complex number , and hence for this interval of @xmath41 the correlation function has an oscillatory decay .
note that for @xmath149 the periodic phase is not stable on the gs , and counterintuitively the oscillatory decay of @xmath111 occurs at higher @xmath95 .
however , the correlation length is smaller than the period of the damped oscillations .
similar change from the monotonic to the oscillatory decay of correlations for increasing temperature has been obtained for the salr system in ref.@xcite .
in this section we consider the specific heat for a fixed concentration .
fixed concentration imposes a global constraint on the microstates , therefore exact analytical calculations directly in the canonical ensemble are less easy . to overcome this difficulty
, we first calculate the specific heat for fixed chemical potential using the exact results of sec.3 . in order to compute the specific heat for fixed concentration we use the thermodynamic relations . for comparison
we perform monte carlo ( mc ) simulations in the canonical ensemble . the specific heat of a mixture with fixed number of all particles and fixed chemical potential difference between the components , @xmath41 , is given by @xmath150 where the exact expression for @xmath151 is given in eq.([omega ] ) .
we compute @xmath152 using the following relations , @xmath153 and @xmath154 in order to verify the exact calculations based on the grand canonical ensemble , we performed mc simulations in the canonical ensemble with the sampling procedure based on the metropolis algorithm [ 35 ] .
the sampling is made with two kind of mc steps : ( i ) the exchange of lipid and water molecules positions ( ii ) the change of lipid molecule orientation .
the specific heat per lattice site is computed using the fluctuation formula : @xmath155\ ] ] where here @xmath156 is the average over the microstates in the canonical ensemble .
the simulations were performed for @xmath157 .
we verified that the finite size effects are negligible for @xmath158 ( see fig.[cv22]a ) .
as a function of @xmath91 for @xmath50 and different temperatures .
top : per unit volume ( @xmath152 ) .
bottom : per amphiphilic molecule ( @xmath159 ) .
symbols denote the results of the monte carlo simulations for @xmath160 , and lines represent the analytical results in thermodynamic limit ( eqs .
( [ cmu])- ( [ rho_mu ] ) and ( [ omega ] ) ) . ,
title="fig : " ] as a function of @xmath91 for @xmath50 and different temperatures .
top : per unit volume ( @xmath152 ) .
bottom : per amphiphilic molecule ( @xmath159 ) .
symbols denote the results of the monte carlo simulations for @xmath160 , and lines represent the analytical results in thermodynamic limit ( eqs .
( [ cmu])- ( [ rho_mu ] ) and ( [ omega ] ) ) . ,
title="fig : " ] in order to compare thermal properties of this model and the salr model of ref.@xcite , we should note that in the case of the colloidal self - assembly the specific heat was calculated per colloid particle , and the solvent was disregarded .
the specific heat per unit volume calculated here is a different quantity , therefore in fig.[cv22]b we present @xmath159 representing the specific heat per amphiphilic molecule .
note that in this model the specific heat of the pure solvent ( i.e. for @xmath161 ) vanishes , since the energy does not fluctuate when all the sites are occupied by the solvent . in this respect
the solvent is analogous to the disregarded solvent in the salr model .
thus , to compare the thermal effects of the amphiphilic and the colloidal self - assembly , we shall compare @xmath159 obtained here and the specific heat per colloid particle calculated in ref.@xcite .
let us first discuss the specific heat for @xmath50 .
for such interactions three phases are present on the gs : the solvent - rich for @xmath161 , the periodic array of bilayers for @xmath102 and the amphiphile - rich for @xmath101 .
the formation of the periodic phase at @xmath133 is indicated by the peak for @xmath102 that becomes narrower for decreasing temperature ( fig.[cv22]a ) . in fig . [ cv22]b an increase of @xmath159 for @xmath162
can be observed for the range of @xmath91 that increases with increasing @xmath95 .
this behaviour may be associated with an equilibrium between bilayers and isolated amphiphilic molecules . in the salr system
the qualitative behavior of the specific heat is similar .
the peak for the density corresponding to the stability of the periodic phase that becomes narrower for decreasing temperature , and another maximum for a very small density are both present @xcite . for @xmath108 ,
@xmath109 only the solvent - rich and the amphiphile - rich phases are stable at the gs .
the absence of the periodic phase leads to a lack of the peak in the specific heat for @xmath102 . as in the previous case an increase of @xmath159 for @xmath162
can be observed .
the @xmath159 in this case is also similar to the specific heat in the salr system for the interactions such that only the two homogeneous phases are present on the gs @xcite .
note that from figs.[cv22 ] and
[ cv21 ] it follows that the dependence of @xmath152 on @xmath163 for fixed @xmath91 is nontrivial , and differs qualitatively from @xmath164 in the lattice gas or ising model . in the latter models
a single maximum of @xmath164 occurs , whereas in this model two maxima separated by a minimum are present .
the simulation snapshots indicate formation of solvent - separated bilayers at low @xmath95 for @xmath165 . between the two maxima of @xmath164 amphiphiles ,
oriented head - to - head or tail - to - tail , form larger domains separated by domains of solvent . at the second maximum random positions and orientations of the amphiphiles appear .
the two maxima of @xmath164 are present even for @xmath166 such that only the pure solvent and pure amphiphiles are present at the gs .
in such a case phase - separated solvent and amphiphiles , or random positions and orientations are present for low @xmath95 or for high @xmath95 respectively , whereas between the two maxima of @xmath164 domains of solvent and properly oriented amphiphiles are formed . as a function of @xmath91 for @xmath108 and @xmath109 and different temperatures . left : per unit volume ( @xmath152 ) .
right : per amphiphilic molecule ( @xmath159 ) .
lines represent the analytical results in thermodynamic limit ( eqs .
( [ cmu])- ( [ rho_mu ] ) and ( [ omega ] ) ) . ,
title="fig : " ] as a function of @xmath91 for @xmath108 and @xmath109 and different temperatures . left : per unit volume ( @xmath152 ) .
right : per amphiphilic molecule ( @xmath159 ) .
lines represent the analytical results in thermodynamic limit ( eqs .
( [ cmu])- ( [ rho_mu ] ) and ( [ omega ] ) ) . ,
title="fig : " ]
we have solved exactly the simple lattice model for amphiphilic mixtures introduced in ref.@xcite .
the ground - state shows that the model predicts the key properties of aqueous solutions of amphiphilic molecules such as lipids .
it also helps to understand the relation between the degeneracy of the ground state and the low surface tension .
for @xmath167 we have obtained oscillatory decay of correlations for the range of @xmath41 corresponding to the gs stability of the periodic phase .
the oscillatory decay of correlations indicates alternating layers of the solvent and the properly oriented amphiphilic molecules . for small @xmath95 and for the range of the chemical potential @xmath41 corresponding to the gs stability of the periodic phase
the correlation length is very large , a few orders of magnitude larger than the molecular size .
thus , the correlation function is consistent with a quasi long - range order .
the periodic ordering at low temperatures is reflected in the shape of the @xmath99 line too .
the average density @xmath107 is nearly independent of pressure for a large pressure interval .
these two results are consistent with a formation of solvent - separated bilayers in domains of a size of order of micrometers . for increasing @xmath95 the correlation length decreases , and
the line @xmath99 becomes smoother .
finally , the specific heat assumes a maximum for the concentration of amphiphiles @xmath102 corresponding to the gs stability of the periodic phase .
the maximum becomes very narrow for low @xmath95 .
when the periodic phase is not present on the gs ( see fig.2 ) , the maximum of the specific heat per amphiphilic molecule for @xmath102 is absent .
we have found strong similarity of the @xmath2 phase diagrams for the present model for amphiphilic self - assembly ( fig.2 ) and for the lattice model for colloidal self - assembly ( fig.1 in ref .
the phases with oscillatory density in the salr systems or oscillatory concentration in the amphiphilic mixtures occur when the repulsion in the case of the colloids or attraction between the solvent and properly oriented amphiphilic molecules are sufficiently strong .
when the above interactions are weak , only two phases are present in the ground state , namely the gas and liquid in the salr case , and the pure water and amphiphile phases in the present model .
it is interesting that the ground state has the same kind of degeneracy at the phase coexistence between the periodic phase and the pure solvent for both , the amphiphilic and the colloidal self - assembly @xcite . in both models
the ground state is strongly degenerate , and the surface tension between the homogeneous and the periodic phase vanishes for @xmath2 , although in the case of the colloidal self - assembly the particles have neither shape nor interaction anisotropy .
arbitrary number of arbitrarily small droplets of the coexisting phases can be present at the phase coexistence .
this degeneracy of the ground state means that the macroscopic separation of the two phases at @xmath168 is not possible , since the formation of an interface does not lead to any increase of the grand potential . at the same time , because of the microscopic size of the droplets , one can interpret the degenerate ground state as a disordered phase .
the region of the @xmath2 phase diagram corresponding to the stability of this phase is of zero measure , however , in contrast to the remaining , ordinary phases .
note that the ultra - low surface tension is a generic property of systems with competing tendencies in the interactions that lead to stability of periodic structures , and is not limited to amphiphilic molecules .
the exact results for the present model and for the model of colloidal self - assembly show that the low surface tension is a more general property of systems with competing interactions , and is not limited to amphiphilic molecules .
this confirms the observation of universality of the periodic ordering on the mesoscopic length scale that was derived under the assumption of weak ordering @xcite .
the effect of the inhomogeneous density or concentration in the above systems on the equation of state has been studied in the framework of the statistical thermodynamics only very recently @xcite .
we have found strong similarity between the pressure - density isotherm in the salr system @xcite and the pressure - concentration isotherm in our model of the amphiphilic mixture .
the characteristic feature of these lines is the plateau in the density or concentration as a function of pressure .
similar shapes have also the density - chemical potential and the concentration - chemical potential lines .
the plateau occurs when the density or concentration takes the value corresponding to the periodic distribution of the clusters or the bilayers . for the corresponding range of @xmath41 and @xmath95 the correlation functions in both models exhibit exponentially damped oscillatory decay with a very large correlation length .
the thermal properties are also similar . in both models
the specific heat assumes a maximum for the density or concentration corresponding to the gs periodic phase .
in addition , the specific heat per particle in the salr case or per amphiphilic molecule in this model increase for decreasing number of particles or amphiphilic molecules . this behavior is associated with the equilibrium between the clusters and isolated particles or between the bilayers ( or micelles ) and isolated amphiphiles .
the main difference between the self - assembly in the amphiphilic and colloidal systems concerns the periodic ordering of the pure amphiphiles into the lamellar structure that is absent in the dense phase in the colloidal system .
this work is dedicated to prof .
tomas boublik and prof .
ivo nezbeda .
a part of this work was realized within the international phd projects programme of the foundation for polish science , cofinanced from european regional development fund within innovative economy operational programme `` grants for innovation '' .
ac and pr acknowledge the financial support by the national science center grant 2012/05/b / st3/03302 .
jp acknowledges the financial support by the national science center under contract decision no .
dec-2013/09/n / st3/02551 . | we consider a lattice model for amphiphiles in a solvent with molecules chemically similar to one part of the amphiphilic molecule .
the dependence of the interaction potential on orientation of the amphiphilic molecules is taken into account explicitly .
the model is solved exactly in one dimension by the transfer - matrix method . in particular , pressure as a function of concentration ,
correlation function and specific heat are calculated .
the model is compared with the recently introduced lattice model for colloidal self - assembly , where the particles interact with the isotropic short - range attraction and long - range repulsion ( salr ) potential .
similarities between the amphiphilic and the colloidal self - assembly are highlighted . |
[ [ algebraic - lambda - calculi . ] ] algebraic @xmath0-calculi .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + two algebraic versions of the @xmath0-calculus arise independently in distinct contexts : the algebraic @xmath0-calculus ( @xmath1 ) @xcite and the linear algebraic @xmath0-calculus ( @xmath2 ) @xcite .
the former has been introduced in the context of linear logic as a fragment of the differential @xmath0-calculus @xcite : the algebraic structure allows to gather in a non deterministic manner different terms , _
i.e. _ each term represents one possible execution .
the latter has been introduced as a candidate @xmath0-calculus for quantum computation : in @xmath2 , a linear combination of terms reflects the phenomenon of superposition , _
i.e. _ the capability for a quantum system to be in two or more states at the same time .
[ [ four - languages - with - different - behaviours . ] ] four languages with different behaviours .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in both languages , functions which are linear combinations of terms are interpreted pointwise : @xmath3 , where `` @xmath4 '' denotes the scalar multiplication .
the two languages differ on the treatment of the arguments . in @xmath2 , in order to deal with the algebraic structure , any function is considered as a linear map : @xmath5 , reflecting the fact that any quantum evolution is a linear map .
it reflects a call - by - value behaviour in the sense that the argument is evaluated until one has a base term . in the opposite
, @xmath1 has a call - by - name evolution : @xmath6 $ ] , without any restriction on @xmath7 . as a consequence ,
the evolutions are different as illustrated by the following example . in @xmath2 ,
@xmath8 while in @xmath1 , @xmath9 .
because they were designed for different purposes , another difference appears between the two languages : the way the algebraic part of the calculus is treated . in @xmath2 , the algebraic structure is captured with a rewrite system , whereas in @xmath1 terms are considered up to algebraic equivalence .
the two choices call - by - value versus call - by - name , algebraic equality versus algebraic reduction allow one to construct four possible calculi .
we name them @xmath10 , @xmath11 , @xmath12 , and @xmath13 , see figure [ tab : linalg ] where they are presented according to their evolution policy and the way they take care of the algebraic part of the language .
inspired by @xmath2 and @xmath1 , the operational semantics of the four languages we introduce slightly differ from the original ones for focusing on the particularities of the calculi we are interested in : reduction strategy and handling of algebraic structure in _
programs_. a first modification is that in all four languages , we avoid reduction under lambda abstractions . as a consequence , contrary to @xmath1 ,
the @xmath0-abstraction is not linear anymore : @xmath14 .
this restriction is a common restriction : reducing under @xmath0 could be considered as optimising the program .
also note that since we consider @xmath15 with algebraic rewrite rules instead of the equalities used in @xmath1 , we need two extra rules : @xmath16 and @xmath17 . these rules were not needed with equalities , since @xmath18 . concerning @xmath19 and @xmath20 ,
restrictions originally imposed in @xmath2 on the rewrite system to ensure confluence are replaced by restrictions which make @xmath19 and @xmath20 actual call - by - value languages .
for example , the rule @xmath21 when @xmath22 is closed - normal form is replaced by only asking @xmath23 to be a value . notice that even in the original language @xmath2 , waiving the restrictions makes sense when confluence can be ensured by other means , see _ e.g.
_ @xcite .
[ [ contribution - relation - between - the - four - languages - through - simulation . ] ] contribution : relation between the four languages through simulation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + although these languages behave differently , we show in this paper that they simulate each other .
this result connects works done in linear logic @xcite and works on quantum computation @xcite .
we show that call - by - value algebraic @xmath0-calculi simulate call - by - name ones and _ vice versa _ by extending the continuation passing style ( cps ) @xcite to the algebraic case .
we also provide simulations between algebraic equality and algebraic reduction in both directions .
the simulations we prove are summed up in figure [ fig : relation ] .
the solid arrows stand for theorems that do not require confluence in their hypothesis whereas the dashed arrows stands for theorems that do .
a preliminary version of this work was presented in @xcite .
[ [ consistency . ] ] consistency .
+ + + + + + + + + + + + without restrictions on the set of terms , both algebraic reductions and algebraic equalities cause problems of consistency , albeit differently .
let @xmath24 . in a system with algebraic reduction
, the term @xmath25 reduces to @xmath26 , but also reduces to @xmath27 and hence to @xmath28 , breaking confluence . to solve this issue ,
several distinct techniques can be used to make an algebraic calculus confluent . in @xcite , restrictions on reduction rules are introduced , e.g. @xmath29 if @xmath28 is closed normal . in @xcite , type systems are set up to forbid diverging terms such as @xmath30 . in a system with algebraic equalities , if @xmath28 and @xmath7 are any terms , the term @xmath28 reduces to @xmath31 , therefore to @xmath7 . in @xmath1 a restriction to positive scalars is proposed to solve the problem .
however such a solution does not work in a system with algebraic reduction ( _ cf .
_ section [ sec : consistency ] ) . in this paper
we do not make a choice _ a priori _ , instead we show that the simulations between the four calculi are correct , providing a general enough methodology to work in a large variety of restrictions on the language .
therefore , we do not force one specific method to make the calculi consistent , leaving the choice to the reader .
[ [ plan - of - the - paper . ] ] plan of the paper .
+ + + + + + + + + + + + + + + + + + in section [ sec : alglam ] , we define the set of terms and the rewrite systems we consider in the paper . in section
[ sec : consistency ] , we discuss the confluence of the algebraic rewrite systems .
section [ sec : sim ] is concerned with the actual simulations . in section [ sec : redeq ] we consider the correspondence between algebraic reduction and algebraic equality whereas in section [ subsec : lintoalg ] and [ subsec : algtolin ] we consider the distinction call - by - name versus call - by - value . in section [ subsec : compose ] , we show how the simulations can compose to obtain the correspondence between any two of the four languages . in section
[ sec : concl ] we conclude by providing some paths for future work .
most of the omitted and sketched proofs are fully developed in the appendix .
the languages @xmath2 and @xmath1 share the same syntax , defined as follows : @xmath32 where @xmath33 ranges over a ring , the _ ring of scalars_. we use the notation @xmath34 as a shorthand for @xmath35 .
note that we could have asked for a semiring instead ; in fact we shall see in section [ sec : conf ] that the analysis we develop here can be adapted to semirings of scalars .
we provide a complete formalisation of the rewrite rules and show how they relate to each other .
we summarise in figure [ tab : rw ] all the rewrite rules that are being used .
the rules are grouped with respect to their intuitive meaning .
we use the usual notation regarding rewrite systems : given a rewrite system @xmath36 , we write @xmath37 for its reflexive and transitive closure .
that is , @xmath38 is valid if @xmath39 or if there exists a rewrite sequence @xmath40 linking @xmath41 and @xmath42 .
we write @xmath43 for the symmetric closure of @xmath36 , that is , the relation that satisfies @xmath44 if and only if @xmath45 or @xmath46 . the original languages @xmath2 and @xmath1 made particular assumptions both on the reduction strategy and the handling of algebraic structure under the reduction . in this paper , we consider separately the distinction call - by - name / call - by - value and the distinction algebraic equality / algebraic reduction .
we develop therefore four languages : a call - by - value language @xmath20 with algebraic equality , a call - by - value language @xmath19 with algebraic reduction , a call - by - name language @xmath47 with algebraic equality and a call - by - name language @xmath15 with algebraic reduction .
these four languages are summarised in figure [ tab : linalg ] .
[ def : lang ] we use the following notations for the rewrite systems obtained by combining the rules described in figure [ tab : rw ] : @xmath48 we define the following four languages and their associated rewrite systems : [ cols="^,^",options="header " , ]
in this section we show that the four languages @xmath19 , @xmath15 , @xmath20 , and @xmath47 are locally confluent . a rewrite system @xmath36 is locally confluent if whenever @xmath49 and @xmath50 there is @xmath51 such that @xmath52 and @xmath53 . in comparison ,
a rewrite system @xmath36 is confluent if whenever @xmath38 and @xmath54 there is @xmath51 such that @xmath52 and @xmath53 .
notice that confluence implies local confluent whereas the inverse is not true .
we first concentrate on the algebraic rules . for each of these calculi
, we use the reductions describing the algebraic structure : @xmath55 and @xmath56 correspond to an oriented rewriting description whereas @xmath57 and @xmath58 correspond to a description by equalities ( since every rewrite rule can be reversed , _ cf . _ definition [ def : lang ] ) .
[ locallyconfluence ] the rewrite systems @xmath55 , @xmath56 , @xmath57 and @xmath58 are locally confluent . for @xmath56 and @xmath55
, we give a semi - automatised proof in the interactive theorem prover coq @xcite .
the interested reader can find the proof in @xcite which is sketched in the appendix .
since for any rewrite system @xmath36 , its symmetric closure @xmath43 is trivially locally confluent , both @xmath57 and @xmath58 are locally confluent
. the rewrites systems considered in this paper are also locally confluent in the presence of the @xmath59-rewrite rules .
[ lem : localconf - llinred ] the four languages in figure [ tab : linalg ] are locally confluent .
the local confluence of the algebraic fragment is proven in lemma [ locallyconfluence ] .
the beta - reduction is confluent using a straightforward extension of the confluence of lambda calculus .
finally , the beta - reduction and the algebraic fragments commute , making each rewrite system locally confluent . in this section , we show that the algebraic fragments are confluent modulo associativity and commutativity .
concerning the full languages , we show that they are either not confluent or trivially confluent ( in the sense that any term is reducing to any other ) . as a consequence ,
we introduce a generic notion of language fragment to describe confluent and consistent sub - languages .
in particular , fragments are used in simulations theorems in section [ sec : sim ] for abstractly representing confluent sub - languages .
[ [ the - algebraic - fragment . ] ] the algebraic fragment .
+ + + + + + + + + + + + + + + + + + + + + + + it is clear that neither @xmath55 nor @xmath56 is strongly normalising : with both systems one can go back and forth between @xmath22 and @xmath60 .
they are however strongly normalising `` modulo associativity and commutativity '' in the sense that any rewrite sequence consists eventually of terms that are equal modulo associativity and commutativity . on the contrary ,
the rewrite systems @xmath57 and @xmath58 are not . in order to formalise this ,
let us denote @xmath61 the system generated by @xmath62 and @xmath63 the rewrite system obtained by taking off the rules @xmath64 and @xmath65 where @xmath36 stands for @xmath55 or @xmath56 .
hence , @xmath66 stands for the system generated by @xmath67 and @xmath68 for the system generated by @xmath69 .
[ def : acnorm ] let @xmath36 be either @xmath56 or @xmath55 and @xmath70 be a reduction sequence ( finite or not ) characterised by the set of terms @xmath71 and the set of rules @xmath72 used to go from @xmath73 to @xmath74 , where @xmath75 stands for a fixed rule in @xmath36 .
we say that the reduction is _ ac - finite _ if the set of indices @xmath76 such that @xmath77 is finite .
the _ ac - length _ of the rewrite sequence is the cardinal of this set of indices .
the rewrite system @xmath36 is _ ac - strongly - normalising _ ( ac - sn ) if for any term @xmath28 , there exists a number @xmath78 such that the ac - length of any rewrite sequence starting at @xmath28 is less than @xmath78 . a term @xmath28 is _ ac - normal _ with respect to a rewrite system @xmath36 if any rewrite sequence starting with @xmath28 consists only of rules ac .
[ th : acnorm ] the systems @xmath55 and @xmath56 are ac - sn .
we use the technique described in @xcite .
an auxiliary measure is defined on terms by @xmath79 , @xmath80 , @xmath81 , @xmath82 , @xmath83 and @xmath84 .
this measure is preserved by rules ac and strictly decreasing on the other algebraic rules .
local confluence plus strong normalisation implies confluence ( see for example @xcite ) .
[ cor : algconf ] the rewrite systems @xmath55 and @xmath56 are confluent , modulo ac .
although we have proved that the four languages under consideration are locally confluent , neither @xmath19 nor @xmath15 is confluent : in each one , the term @xmath25 rewrites both to @xmath26 and @xmath28 , where @xmath24 .
regarding @xmath20 and @xmath47 , without restriction both are trivially confluent since for all terms @xmath28 and @xmath7 , @xmath28 reduces to @xmath7 : @xmath85 .
hence , with the algebraic equality , both languages can simulate any rewrite system . for getting back consistency , it is of course possible to modify the rewrite systems as in @xcite but it would break the correspondence between call - by - value and call - by - name . in this paper
we propose instead to restrict the set of terms . in the literature ,
there have been two methods : with algebraic equalities , vaux @xcite considers non - negative scalars ( semiring ) on a language with algebraic equality .
the restriction on scalars is enough for getting unicity of normal forms .
although this solves the consistency problem for the languages with algebraic equality , it does not give confluence for the languages with algebraic reduction .
indeed , consider the critical pair @xmath86 , @xmath87 .
the term @xmath88 can only produce an even number of @xmath28 s : we can not close the pair . with algebraic reductions , other papers @xcite use type systems for retrieving strong normalisation on a language with algebraic reduction .
this technique could be directly adapted to our setting since it is possible to have subject reduction in these cases .
the simulations theorems that we develop in this paper are correct in a general untyped setting ( and in fact trivially true when we simulate a language with algebraic reduction with a language with algebraic equality as remarked above ) , but also true if one restrict the scalars to a semiring ( as done in @xcite ) , or if we restrict the terms to any typed setting , provided that the languages @xmath19 and @xmath15 satisfy subject reduction and that the cps translations preserve typability .
thus , in this paper we do not restrict the calculi _ a priori _ , instead , we propose a notion of _ language fragments _ to parametrise the simulation results .
the definition of fragment is general enough to capture many settings : various typed systems , but also the restrictions to a given set of terms such as the set of ac - sn terms or taking scalars from a semiring .
we define formally a fragment in the following way : [ def : subset ] a fragment @xmath89 of @xmath19 ( resp .
@xmath15 ) is a language defined on a subset of terms closed under @xmath90-reduction ( resp .
@xmath91-reduction ) .
the rewrite system of @xmath89 is inherited from the one of @xmath19 ( resp .
@xmath15 ) . the definition of a fragment in the presence of algebraic equalities should be treated carefully .
indeed , note that the algebraic equalities are defined as @xmath92 if and only if @xmath93 or @xmath94 . as a consequence , for any subset @xmath89 of terms closed under @xmath95-reduction , if @xmath28 is in @xmath89 then for any @xmath7 ( in @xmath89 or not ) , @xmath96 since @xmath97 .
we therefore need to define the algebraic equality with respect to the particular subset of terms under consideration .
[ def : subeq ] a fragment @xmath89 of @xmath20 ( resp .
@xmath47 ) is a fragment of @xmath19 ( resp .
@xmath15 ) together with an algebraic equality defined as @xmath98 ( resp .
@xmath99 ) if and only if @xmath100 and @xmath101 or @xmath102 ( resp . @xmath103 or @xmath104 ) .
the @xmath59-reduction is not modified . when referring to a fragment of @xmath20 ( resp .
@xmath47 ) , we use the abuse of notation @xmath58 ( resp .
@xmath57 ) instead of @xmath105 ( resp .
@xmath106 ) for the restricted rewrite system , when the fragment under consideration is clear .
the core of the paper is concerned with the mutual simulations of the four languages .
the first class of problems relates algebraic reduction with algebraic equality . if simulating a language with algebraic reduction with a language with algebraic equality is not specially difficult , going in the opposite direction is not possible in general .
indeed , if @xmath107 is possible in @xmath20 , ( where @xmath24 ) it is difficult to see how one could make @xmath26 go to @xmath28 in @xmath19 without further hypotheses . in this section ,
we show that a fragment of a language with algebraic equality can be simulated by the corresponding fragment with algebraic reduction provided that the latter is _ confluent _ ( theorems [ th : eqred ] and [ th : eqred2 ] ) .
the second class of problems is concerned with call - by - value and call - by - name . in this paper , the simulations of call - by - name by call - by - value and its reverse
are treated using continuation passing style ( cps ) , extending the techniques described in @xcite to the algebraic case ( theorems [ th : sim ] , [ th : simeq ] , [ th : sim2 ] and [ th : sim2eq ] ) .
the results are summarised in figure [ fig : relation ] .
solid arrows correspond to results where no particular hypothesis on the language is made .
dashed arrows correspond to results where confluence is required . as the relation @xmath90 is contained in @xmath108 and the relation @xmath91
is contained in @xmath109 , the first simulation theorems are trivial .
[ th : redeq ] for any term @xmath28 if @xmath110 , then @xmath111 .
[ th : redeq2 ] for any term @xmath28 if @xmath112 , then @xmath113 .
the simulations going in the other direction are only valid in the presence of confluence . in the following two theorems ,
the algebraic equality is defined with respect to the considered fragment ( see definition [ def : subeq ] . )
[ th : eqred ] for any term @xmath28 in a confluent fragment of @xmath19 , if @xmath114 , then @xmath115 , with @xmath116 . first note that a value can only reduce to another value .
this follows by direct inspection of the rewriting rules .
we proceed by induction on the length of the reduction . * if @xmath117 , then choose @xmath118 and note that @xmath119 . *
assume the result true for @xmath120 : there is a value @xmath121 such that @xmath115 and @xmath116 .
let @xmath122 .
case distinction : * * @xmath123 , then @xmath124 which implies @xmath125 . * * @xmath126 , then either @xmath127 , and then this case is analogous to the previous one , or @xmath102 . due to the confluence of the subset , there exists a term @xmath23 such that @xmath128 and @xmath129 , implying that @xmath23 is a value , thus @xmath130
. then we have @xmath130 and @xmath116 , so @xmath131 , closing the case .
[ th : eqred2 ] for any term @xmath28 in a confluent fragment of @xmath15 , if @xmath132 , then @xmath133 , with @xmath134 .
similar to the previous theorem . to prove the simulation of @xmath19 with @xmath15 and the simulation of @xmath20 with @xmath47
, we introduce an algebraic extension of the continuation passing style used to prove that call - by - name simulates call - by - value in the regular @xmath0-calculus @xcite .
let @xmath135\!]}:\lambda_{{\ensuremath{\lambda_{{\it lin}}}}}\to\lambda_{{\ensuremath{\lambda_{{\it alg}}}}}$ ] be the following encoding where @xmath136 and @xmath137 are fresh variables . @xmath138\ ! ] } & = & { \lambda f\,\left(f\right)~x } , & { [ \![{0}]\ ! ] } & = & 0,\\ { [ \![{\lambda x\,m}]\ ! ] } & = & \lambda f\,(f)~\lambda x\,{[\![{m}]\ ! ] } , & { [ \![{(m)~n}]\ ! ] } & = & { { \lambda{f}\,({}})~\lambda{g}\,({{[\![{m}]\!]}})~\lambda{h}\,(({g})~{h})~{f}}{{[\![{n}]\!]}},\\ { [ \![{\alpha.m}]\ ! ] } & = & \lambda f\,(\alpha.{[\![{m}]\!]})~f , & { [ \![{m+n}]\!]}&=&\lambda f\,({[\![{m}]\ ! ] } + { [ \![{n}]\!]})~f .
\end{array}\ ] ] let @xmath139 be the encoding for values defined by @xmath140 , @xmath141 , @xmath142\!]}$ ] , @xmath143 , @xmath144 .
note that this encoding is compatible with substitution ( proof by induction on @xmath28 ) : [ 2:lem : substitution ] @xmath145}]\!]}={[\![{m}]\!]}[x:=\psi(b)]$ ] with @xmath146 a base term . using this encoding
, we can simulate @xmath19 with @xmath15 , as formalised in the following theorem .
the sketch of the proof is developed in the second part of this section .
[ th : sim ] for any term @xmath28 if @xmath147 where @xmath148 is a value , then @xmath149\!]})~\lambda x\,x { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$}}}}}}}\psi(v)$ ] .
[ ex:1 ] for any terms @xmath28 and @xmath7 , let @xmath150 .
let @xmath151 be the term @xmath152 , and let @xmath153 and @xmath154 be two values . then @xmath155 and @xmath156 .
we consider the simulation @xmath19 to @xmath15 .
the translation @xmath157\!]}$ ] is @xmath158\!]})\ , \lambda g\ , ( { [ \![{u+v}]\!]})\lambda h\ , ( ( g)\,h)\,f$ ] , where @xmath159\!]}$ ] is the term @xmath160\!]}$ ] , with @xmath161\!]}$ ] being @xmath162 , @xmath163\!]}$ ] being @xmath164\!]}+{[\![{v}]\!]})\,f$ ] , and @xmath165\!]}$ ] being @xmath166 .
we now rewrite @xmath167\!]})\,\lambda z\,z$ ] in @xmath15 .
@xmath168\!]})\ , \lambda g\ , ( { [ \![{u+v}]\!]})\lambda h\,((g)\,h)\,\lambda z\,z \\ & = & ( \lambda f\ , ( f)\,\lambda x\ , { [ \![{{\left\langle x , x\right\rangle}}]\!]})\ , \lambda g\ , ( { [ \![{u+v}]\!]})\lambda h\ , ( ( g)\,h)\,\lambda z\ , z \\ & { { \ensuremath{{\to_{a\cup\beta } } } } } & ( \lambda g\ , ( { [ \![{u+v}]\!]})\ , \lambda h\ , ( ( g)\,h)\,\lambda z\ , z)\ , \lambda x\ , { [ \![{{\left\langle x , x\right\rangle}}]\ ! ] } \\ & { { \ensuremath{{\to_{a\cup\beta } } } } } & ( { [ \![{u+v}]\!]})\,\lambda h\ , ( ( \lambda x\ , { [ \![{{\left\langle x , x\right\rangle}}]\!]})\,h)\,\lambda z\ , z \\ & { { \ensuremath{{\to_{a\cup\beta } } } } } & ( { [ \![{u}]\ ! ] } + { [ \![{v}]\!]})\ , \,\lambda h\ , ( ( \lambda x\ , { [ \![{{\left\langle x , x\right\rangle}}]\!]})\,h)\,\lambda z\ , z\\ & { { \ensuremath{{\to_{a\cup\beta } } } } } & ( { [ \![{u}]\!]})\ , \,\lambda h\ , ( ( \lambda x\ , { [ \![{{\left\langle x , x\right\rangle}}]\!]})\,h)\,\lambda z\ , z + ( { [ \![{v}]\!]})\ , \,\lambda h\ , ( ( \lambda x\ , { [ \![{{\left\langle x , x\right\rangle}}]\!]})\,h)\,\lambda z\ , z \\ & { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$ } } } } } } } & ( \lambda h\ , ( ( \lambda x\ , { [ \![{{\left\langle x , x\right\rangle}}]\!]})\,h)\,\lambda z\ , z)\ , \psi(u ) + ( \lambda h\ , ( ( \lambda x\ , { [ \![{{\left\langle x , x\right\rangle}}]\!]})\,h)\,\lambda z\ , z)\ , \psi(v ) & ( * ) \\ & { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$ } } } } } } } & ( ( \lambda x\ , { [ \![{{\left\langle x , x\right\rangle}}]\!]})\,\psi(u))\,\lambda z\ , z + ( ( \lambda x\ , { [ \![{{\left\langle x , x\right\rangle}}]\!]})\,\psi(v))\,\lambda z\ , z \\ & { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$ } } } } } } } & ( { [ \![{{\left\langle x , x\right\rangle}}]\!]}[x:= \psi(u)])\,\lambda z\ , z + ( { [ \![{{\left\langle x , x\right\rangle}}]\!]}[x:= \psi(v)])\,\lambda z\ , z \\ & { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$ } } } } } } } & ( { [ \![{{\left\langle u , u\right\rangle}}]\!]})\,\lambda z\ , z + ( { [ \![{{\left\langle v , v\right\rangle}}]\!]})\,\lambda z\ , z \quad \textrm{(lemma \ref{2:lem : substitution } ) } \\ & { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$ } } } } } } } & ( \lambda z\ , z ) \ , \psi({{\left\langle u , u\right\rangle } } ) + ( \lambda z\ , z ) \ , \psi({{\left\langle v , v\right\rangle}})&(**)\\ & { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$ } } } } } } } & \psi({{\left\langle u , u\right\rangle } } ) + \psi({{\left\langle v , v\right\rangle}})\\ & = & \psi({{\left\langle u , u\right\rangle } } + { { \left\langle v , v\right\rangle } } ) \\ \end{array}\ ] ] similarly , one can relate fragments of @xmath47 to fragments of @xmath20 as follows . [
th : simeq ] for any two fragments @xmath169 of @xmath20 and @xmath170 of @xmath47 such that @xmath171 , @xmath149\!]})~\lambda x\,x\in s_{\it a}$ ] , and for any term @xmath28 in @xmath169 , if @xmath114 where @xmath148 is a value , then @xmath149\!]})~\lambda x\,x { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{{\tiny $ = $ } $ \ast$}}}}}}}\psi(v)$ ] .
again , the sketch of the proof is developed later in the section . as we already noted several times in this paper , without restricting the languages , theorem [ th : simeq ] would be trivial .
any term reducing to any other one , the desired reduction would be of course valid without restriction .
this theorem shows that if the calculi are restricted to fragments , the result is still true .
one example of such fragments is found by taking the restriction of scalars to non - negative elements , as in @xcite .
once a term is encoded it can be reduced either by @xmath172 or by @xmath173 ( respectively @xmath174 or @xmath175 ) without distinction , and still obtain the same result .
we state this fact as a corollary : [ cor : ind ] ( 1 ) for any term @xmath28 , if @xmath147 where @xmath148 is a value , then @xmath149\!]})~\lambda x\,x { { \ensuremath{{{\to_{\ell\cup\beta}^{\textrm{$\ast$}}}}}}}\psi(v)$ ] ; ( 2 ) for any fragment @xmath89 of @xmath20 such that @xmath176 , @xmath149\!]})~\lambda x\,x\in s$ ] , and for any term @xmath28 in @xmath89 , if @xmath114 where @xmath148 is a value , then @xmath149\!]})~\lambda x\,x { { \ensuremath{{{\to_{\ell\cup\beta}^{\textrm{{\tiny $ = $ } $ \ast$}}}}}}}\psi(v)$ ] .
it suffices to check the proofs of theorems [ th : sim ] and [ th : simeq ] to verify that all the reductions @xmath172 are done by rules common in both languages .
[ ex:1-indifference ] note that in example [ ex:1 ] one could have as well rewrite with @xmath90 which illustrates the indifference property ( corollary [ cor : ind ] ) .
now we proceed to prove theorems [ th : sim ] and [ th : simeq ] . the proof in @xcite
can be extended to the algebraic case .
[ [ an - administrative - operation . ] ] an administrative operation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + we define a convenient infix operation ( @xmath177 ) capturing the behaviour of translated terms . for example , if @xmath146 is a base term , _
i.e. _ a variable or an abstraction , then its translation into @xmath15 is @xmath178\!]}={\lambda f\,\left(f\right)~\psi(b)}$ ] .
if we apply this translated term to a certain @xmath179 , we obtain @xmath180 .
we define @xmath181 and get that @xmath182\!]})~k{{\ensuremath{{\to_{a\cup\beta}}}}}b : k$ ] .
this fact will be generalised to @xmath149\!]})~k{{\ensuremath{{\to_{a\cup\beta}}}}}m : k$ ] in lemma [ lem : lemma2 ] .
let @xmath183 be the infix binary operation defined by : @xmath184\!]})\,\lambda\,h\,((g)\,h)\,k \\
\end{array}\ ] ] [ lem : lemma2 ] if @xmath179 is a base term , then for any @xmath28 , @xmath149\!]})~k{{\ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$}}}}}}}m : k$ ] .
structural induction on @xmath28 .
we give the case @xmath185 , as an example .
first an intermediate result is needed : for any @xmath28 , @xmath186\!]})~\lambda h\,((g)~h)~k$ ] @xmath91 @xmath187 .
this can be proved by structural induction on @xmath28
. then @xmath188\!]})~k=({{\lambda{f}\,({}})~\lambda{g}\,({{[\![{m'}]\!]}})~\lambda{h}\,(({g})~{h})~{f}}{{[\![{n}]\!]}})~k$ ] which @xmath91-reduces to @xmath189\!]})~\lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k$ ] .
note that @xmath190\!]})~\lambda h\,((g)~h)~k$ ] is a base term , so by the induction hypothesis the above term reduces to @xmath191\!]})~\lambda h\,((g)~h)~k$ ] which by the previous intermediate result , @xmath91-reduces to @xmath192 .
the following lemmas and its corollary state that the @xmath193 operation preserves reduction .
[ lem : lemma3alg ] if @xmath102 then @xmath194 base term , @xmath195 .
induction on the possible rule applied from @xmath102 .
we give one simple case as an example .
. then @xmath197 .
[ lem : lemma3 ] if @xmath198 then @xmath194 base term , @xmath199 . if @xmath102 , then use lemma [ lem : lemma3alg ] . if @xmath200 , then we prove it by induction on the possible rule applied ( either @xmath201 , @xmath202 or one of @xmath203 ) .
we give the case of the @xmath201-reduction as an example : @xmath204 , @xmath205-reducing to the term @xmath206\!]})\,\psi(b))\,k$ ] which @xmath205-reduces to @xmath207\!]}[x:=\psi(b)]\,k$ ] , equal by lemma [ 2:lem : substitution ] to @xmath145}]\!]}\,k$ ] . by lemma [ lem : lemma2 ] , it @xmath172-reduces to @xmath208:k$ ] .
[ cor : lemma3 ] if @xmath113 then @xmath194 base terms , @xmath209 . by case distinction .
if @xmath198 , then by lemma [ lem : lemma3 ] , @xmath199 , which implies @xmath209 .
if @xmath127 , then by lemma [ lem : lemma3alg ] , @xmath210 , which also implies @xmath209 . finally , the @xmath193 operation also captures the translation of values in the following way : [ lem : valuestranslation ] for any value @xmath148 , @xmath211 we proceed by structural induction on @xmath148 .
let @xmath148 be a base term .
then @xmath212 . let @xmath213 .
then @xmath214 , which by the induction hypothesis , reduces to @xmath215 .
let @xmath216 .
then @xmath217 , which by the induction hypothesis , reduces to @xmath218 .
. then @xmath220 .
we discuss example [ ex:1 ] in the light of these results . the term @xmath221 is equal to the terms @xmath222 and @xmath223 .
the term @xmath224 is equal to the term @xmath225 which reduces to @xmath226 .
so we do have the rewrites requested by lemmas [ lem : lemma2 ] , [ lem : lemma3 ] and [ lem : valuestranslation ] .
now the proofs of theorems [ th : sim ] and [ th : simeq ] go as follows . from lemma [ lem : lemma2 ] , @xmath149\!]})~\lambda x\,x{{\ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$}}}}}}}m:\lambda x\,x$ ] and from lemma
[ lem : lemma3 ] , it @xmath172-reduces to @xmath227 . from lemma [ lem : valuestranslation ] , @xmath228 . from lemma [ lem : lemma2 ] ,
@xmath149\!]})~\lambda x\,x { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$}}}}}}}m:\lambda x\,x$ ] , and this implies that @xmath149\!]})~\lambda x\,x { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{{\tiny $ = $ } $ \ast$}}}}}}}m:\lambda x\,x$ ] . from corollary [ cor : lemma3 ] , this latter term @xmath174-reduces to @xmath227 . from lemma [ lem :
valuestranslation ] , @xmath228 , which implies that @xmath229 .
note that since @xmath149\!]})~\lambda x\,x\in s_{\it a}$ ] , @xmath230 is also in @xmath231 due to the closeness under @xmath57 of @xmath170 .
the same applies to @xmath232 , thus also to @xmath227 and finally to @xmath233 .
[ [ the - simulation - of - lambdato_it - alg - with - lambdato_it - lin . ] ] the simulation of @xmath234 with @xmath235 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to state that @xmath19 simulates @xmath15 , we use an algebraic extension of the continuation passing style encoding following again @xcite .
let @xmath236 be the following encoding where @xmath136 and @xmath137 are fresh variables .
@xmath237 this encoding satisfies two useful properties ( the first is a trivial result and the second follows by induction on @xmath28 ) .
[ 3:lem : base ] for all terms @xmath28 , the term @xmath238 is a base term .
[ 3:lem : substitution ] @xmath239\}\hspace{-.55em}|\hspace{.3em}}={\{\hspace{-.235em}| m\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]$ ] .
let @xmath240 be the encoding for values defined by @xmath241 , @xmath242 , @xmath243 @xmath244 , @xmath245 , @xmath246 .
simulation theorems , similar to theorems [ th : sim ] and [ th : simeq ] , can be stated as follows .
[ th : sim2 ] for any program @xmath28 ( _ i.e. _ closed term ) , if @xmath247 where @xmath148 is a value , then @xmath248 .
[ th : sim2eq][th : simeq2 ] for any two fragments @xmath170 of @xmath47 and @xmath169 of @xmath20 such that @xmath249 , @xmath250 , and for any program @xmath28 in @xmath170 , if @xmath132 where @xmath148 is a value , then @xmath251 . a result similar to corollary [ cor : ind ] can also be formulated .
it is proven in a similar manner .
[ cor : ind2 ] ( 1 ) for any program @xmath28 , if @xmath247 where @xmath148 is a value , then @xmath252 ; ( 2 ) for any fragment @xmath89 of @xmath47 such that @xmath176 , @xmath253 , and for any program @xmath28 in @xmath89 , if @xmath132 where @xmath148 is a value , then @xmath254 . before moving to the description of the proof of theorems [
th : sim2 ] and [ th : sim2eq ] , let us consider an example .
[ ex:2 ] we illustrate theorem [ th : sim2 ] using the term @xmath255 of example [ ex:1 ] which reduces to @xmath256 in @xmath19 and to @xmath257 in @xmath15 .
the translation @xmath258 is the term @xmath259 , where @xmath260 is @xmath261 .
@xmath262 is @xmath263 , @xmath264 is @xmath265 and @xmath266 is @xmath267 we now rewrite @xmath268 in @xmath15 .
@xmath269 ) \,\lambda z\,z \\
\textrm{(lemma \ref{3:lem : substitution } ) } & = & ( { \{\hspace{-.235em}| { \left\langle u+v , u+v\right\rangle}\}\hspace{-.55em}|\hspace{.3em}})\,\lambda z\,z \\ & { { \ensuremath{{\to_{\ell\cup\beta } } } } } & ( \lambda z\,z ) \ , \phi({{\left\langle u+v , u+v\right\rangle } } ) & ( { * } { * } { * } { * } ) \\ & { { \ensuremath{{\to_{\ell\cup\beta } } } } } & \phi({{\left\langle u+v , u+v\right\rangle } } ) \end{array}\ ] ] [ [ proof - of - the - simulation - theorems . ] ] proof of the simulation theorems .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in section [ subsec : lintoalg ] , the proofs of the simulations theorems were performed using an administrative operation `` @xmath177 '' and three intermediate results , as follows ( the term @xmath179 is taken as a base term ) .
( 1 ) prove that @xmath149\!]})\,k { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$}}}}}}}m : k$ ] ; ( 2 ) prove that if @xmath198 then @xmath199 ; ( 3 ) prove that if @xmath148 is a value , @xmath270 . for the simulation theorems of the present section , we use a similar procedure .
[ [ an - administrative - operation.-1 ] ] an administrative operation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + we keep the same notation for the administrative , infix operation defined for the purpose of the proof .
let @xmath271 be the infix binary operation defined by : @xmath272 the three lemmas needed for the proof of the simulation theorems now read as follow . [ lem : lemma2cps ] if @xmath179 is a base term , for any closed term @xmath28 @xmath273 .
the proof is done by structural induction on @xmath28 .
we follow the sketch of the proof of lemma [ lem : lemma2 ] , and give the case @xmath185 , as an example .
first we prove by induction on @xmath28 that @xmath274 . then @xmath275 .
note that @xmath276 is a base term , so by the induction hypothesis the above term reduces to @xmath277 which by the previous intermediate result , @xmath90-reduces to @xmath192 .
[ lem : lemma3cps ] if @xmath278 then @xmath194 base term , @xmath279 case by case on the rules of @xmath15 .
we give the case of the @xmath205-reduction as an example : @xmath280 which by lemma [ 3:lem : base ] , @xmath90-reduces to @xmath281)\,k$ ] .
this , by lemma [ 3:lem : substitution ] , is equal to @xmath282\}\hspace{-.55em}|\hspace{.3em}})\,k$ ] and this , by lemma [ lem : lemma2cps ] , @xmath173-reduces to @xmath283:k$ ] .
note that in the previous derivation , the reduction @xmath284
@xmath90 @xmath285 $ ] is valid since for any term @xmath7 , @xmath286 is a base term . [
lem : valuestranslation2 ] if @xmath148 is a value and @xmath179 is a base term , @xmath287 .
we discuss example [ ex:2 ] in the light of these results .
the term @xmath288 is equal to the terms @xmath222 .
the term @xmath289 is equal to the term @xmath290 which reduces to @xmath291 .
again , we have the rewrites requested by lemmas [ lem : lemma2cps ] , [ lem : lemma3cps ] and [ lem : valuestranslation2 ] .
we are now ready to prove the simulation theorems .
as advertised , these proofs reflect the exact same structures of the proofs of theorems [ th : sim ] and [ th : simeq ] .
from lemma [ lem : lemma2cps ] , @xmath292 , from lemma [ lem : lemma3cps ] it @xmath173-reduces to @xmath227 . from lemma
[ lem : valuestranslation2 ] , @xmath293 . from lemma [ lem : lemma2cps ] , @xmath294 , and
this implies that @xmath295 . a result equivalent to corollary [ cor : lemma3 ] can be shown as easily : if @xmath296 then for all base terms @xmath179 , @xmath297
this entails that @xmath232 @xmath175-reduces to @xmath227 . from lemma [ lem : valuestranslation2 ] , @xmath287 , which implies that @xmath298 .
note that since @xmath299 , @xmath230 is also in @xmath231 due to the closeness under @xmath58 of @xmath169 .
the same applies to @xmath232 , thus also to @xmath227 and finally to @xmath300 . in figure
[ fig : relation ] , some arrows are missing .
we are now showing that the already existing arrows `` compose '' well .
the first two simulations are @xmath301 and @xmath302 and do not require confluence . for any program
@xmath28 , if @xmath147 ( resp .
@xmath247 ) where @xmath148 is a value , then @xmath149\!]})~\lambda x\,x { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{{\tiny $ = $ } $ \ast$}}}}}}}\psi(v)$ ] ( resp .
@xmath303 ) . given that @xmath147 , by theorem [ th : sim ] , @xmath149\!]})~\lambda x\,x{{\ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$}}}}}}}\psi(v)$ ] , which by theorem [ th : redeq ] implies @xmath149\!]})~\lambda x\,x{{\ensuremath{{{\to_{a\cup\beta}^{\textrm{{\tiny $ = $ } $ \ast$}}}}}}}\psi(v)$ ] .
analogously , given that @xmath247 , by theorem [ th : sim2 ] , @xmath304 , which by theorem [ th : redeq2 ] implies that @xmath305 @xmath175 @xmath300 .
the other two simulations are @xmath306 and @xmath307 and they do require confluence . for any program @xmath28 in a confluent fragment of @xmath20 ( resp .
@xmath47 ) , if @xmath114 ( respectively @xmath132 ) then @xmath149\!]})~\lambda x\,x { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$}}}}}}}\psi(v')$ ] with @xmath116 ( respectively @xmath308 with @xmath134 ) . given that @xmath114 and that @xmath28 is in a confluent fragment , theorem [ th : eqred ] states that @xmath115 with @xmath116 .
in addition , theorem [ th : sim ] states that @xmath149\!]})~\lambda x\,x{{\ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$}}}}}}}\psi(v')$ ] .
the other result is similar using theorems [ th : eqred2 ] and [ th : sim2 ] .
in this paper we described four canonical algebraic lambda - calculi with vectorial structures , recapitulating the few existing means of writing such a language .
we show how each language can simulate the other , by taking care of marking where confluence is used or not .
as already shown by plotkin @xcite , if the simulation of call - by - value by call - by - name is sound , it fails to be complete for general ( possibly non - terminating ) programs . a known solution to this problem is developed in @xcite .
recent work @xcite shown that the technique can be adapted to the algebraic case to retrieve completeness .
the work @xcite develop a galois connection between call - by - name and call - by - value . a direction for study
is to build on this work to also get a galois connection in the algebraic case .
concerning semantics , the algebraic @xmath0-calculus admits finiteness spaces as a model @xcite .
what is the structure of the model of the linear algebraic @xmath0-calculus induced by the continuation - passing style translation in finiteness spaces ?
the algebraic lambda - calculus can be equipped with a differential operator . what is the corresponding operator in call - by - value through the translation ? [ [ acknowledgements ] ] acknowledgements + + + + + + + + + + + + + + + + many thanks to pablo arrighi , ali assaf and lionel vaux for fruitful discussions and suggestions .
this work is supported by the cnrs - ins2i peps project quand .
10 a. assaf and s. perdrix .
completeness of algebraic cps simulations . to appear in
dcm11_. http://membres-lig.imag.fr/perdrix/cps-completeness.html t. altenkirch and j. j. grattage . a functional quantum programming language . in _ proc .
lics05 _ , pp 249258 .
p. arrighi and a. daz - caro .
scalar system f for linear - algebraic @xmath0-calculus : towards a quantumphysical logic . in _ proc.qpl09_,vol .
270 - 2 of _ entcs _ , pp 219229 .
p. arrighi , a. daz - caro , and b. valiron .
a type system for the vectorial aspects of the linear - algebraic lambda - calculus . to appear in _ proc .
dcm11_. http://membres-liglab.imag.fr/diazcaro/vectorial.pdf p. arrighi and g. dowek . a computational definition of the notion of vectorial space . in _ proc .
wrla04 _ , vol .
117 of _ entcs _ , pp . 249261 .
p. arrighi and g. dowek .
linear - algebraic lambda - calculus : higher - order , encodings , and confluence . in _ proc .
rta08 _ , vol .
5117 of _ lncs _ , pp . 1731 .
p. buiras , a. daz - caro , and m. jaskelioff .
lower bounds for scalars in a typed algebraic @xmath0-calculus .
preprint at arxiv:1102.0749 , february 2011 . .
inria , 8.2 edition .
a. daz - caro , s. perdrix , c. tasson , and b. valiron .
equivalence of algebraic @xmath0-calculi . in
_ informal pro .
hor10 _ , pp 611 .
a. daz - caro and b. petit .
sums in linear algebraic lambda - calculus .
preprint at arxiv:1011.3542 ,
november 2010 .
t. ehrhard .
, 12(5):579623,2003 .
t. ehrhard .
finiteness spaces . , 15(4):615646 , 2005 .
t. ehrhard . afinitenessstructureonresourceterms .
in_proc.lics10_,pp.402410 . t.ehrhardandl.regnier .
, 309(1):141,2003 .
m. j. fischer .
@xmath0-calculus schemata .
, 7(1):104109 , 1972 .
m. pagani and s. ronchi della rocca .
solvability in resource lambda calculus . in _ proc .
fossacs10 _ , vol .
6014 of _ lncs _ , pp . 358373 .
m. pagani and p. tranquilli .
parallel reduction in resource lambda - calculus . in _ proc .
aplas09 _ ,
5904 of _ lncs _ , pp . 226242 .
g.d.plotkin.callbyname,callbyvalueandthe@xmath0-calculus._tcs_,1(2):125-159,1975 .
areflexiononcall - by - value . 19(6):916941 , 1997 .
c. tasson .
algebraic totality , towards completeness . in _ proc .
tlca09 _ , vol .
5608 of _ lncs _ , pp 325340 . .
cambridge university press , 2003 .
b. valiron .
semantics of a typed algebraic lambda - calculus . in _ proc .
dcm10 _ , vol .
26 of _ eptcs _ , pp . 147158 .
b. valiron .
coq proof .
http://www.monoidal.net/vectorial-alglin-coqproof-v2.tgz .
b. valiron .
orthogonality and algebraic @xmath0-calculus . in _ proc .
qpl10 _ , pp .
169175 .
l. vaux . on linear combinations of @xmath0-terms . in _ proc .
4533 of _ lncs _ , pp . 374388 .
l. vaux . the algebraic lambda calculus .
, 19(5):10291059 , 2009 .
* if @xmath93 and @xmath309 , using only algebraic rules , then this has been already proven in lemma [ locallyconfluence ] .
* if @xmath93 and @xmath309 , using only beta - reduction , this is a trivial extension of the confluence of lambda calculus . *
if @xmath93 by an algebraic rule and @xmath309 by beta reduction , then in @xmath19 a term of the form @xmath310 has to be a subterm of @xmath28 , since @xmath28 beta - reduces . note that @xmath311 can not reduce since it is under a lambda and @xmath146 can not reduce since it is a base term .
then the beta - reduction and the algebraic - reduction are independent in @xmath19 , and so this result is trivial . in @xmath15 a term of the form
@xmath312 has to be a subterm of @xmath28 .
note that @xmath311 can not reduce since it is under a lambda and @xmath7 can not reduce since it is an argument . then again the beta - reduction and the algebraic - reduction are independent in @xmath15 , and so this result is trivial . structural induction on @xmath28 .
. cases : * * @xmath314 . then @xmath208=y$ ] , and so @xmath145}]\!]}={\lambda f\,\left(f\right)~y}={\lambda f\,\left(f\right)~x}[y / x]={[\![{m}]\!]}[x:=\psi(b)]$ ] . * * @xmath315
. then @xmath145}]\!]}={\lambda f\,\left(f\right)~\lambda y\,{[\![{n}]\!]}}={\lambda f\,\left(f\right)~x}[\lambda y\,{[\![{n}]\!]}/x]={[\![{m}]\!]}[x:=\psi(b)]$ ] . * @xmath316
. then @xmath145}]\!]}={[\![{m}]\!]}[x:=\psi(b)]={[\![{m}]\!]}$ ] .
* @xmath317 .
analogous to previous case .
* @xmath318 . then @xmath319}]\!]}&=&{[\![{\lambda y\,(n[x:=b])}]\!]}\\ & = & { \lambda f\,\left(f\right)~\lambda y\,{[\![{n[x:=b]}]\!]}}\\ & & \textrm{by the induction hypothesis}\\ & = & { \lambda f\,\left(f\right)~\lambda y\,{[\![{n}]\!]}[x:=\psi(b)]}\\ & = & ( { \lambda f\,\left(f\right)~\lambda y\,{[\![{n}]\!]}})[x:=\psi(b)]\\ & = & { [ \![{m}]\!]}[x:=\psi(b ) ] \end{aligned}\ ] ] * @xmath320 . then @xmath321}]\ ! ] } & = & { [ \![{((n_1)~n_2)[x:=b]}]\!]}\\ & = & { [ \![{(n_1[x:=b])~n_2[x:=b]}]\!]}\\ & = & { { \lambda{f}\,({}})~\lambda{g}\,({{[\![{n_1[x:=b]}]\!]}})~\lambda{h}\,(({g})~{h})~{f}}{{[\![{n_2[x:=b]}]\!]}}\\ & & \textrm{by the induction hypothesis}\\ & = & { { \lambda{f}\,({}})~\lambda{g}\,({{[\![{n_1}]\!]}[x:=\psi(b)]})~\lambda{h}\,(({g})~{h})~{f}}{{[\![{n_2}]\!]}[x:=\psi(b)]}\\ & = & { { \lambda{f}\,({}})~\lambda{g}\,({{[\![{n_1}]\!]}})~\lambda{h}\,(({g})~{h})~{f}}{{[\![{n_2}]\!]}}[x:=\psi(b)]\\ & = & { [ \![{(n_1)~n_2}]\!]}[x:=\psi(b)]\\ & = & { [ \![{m}]\!]}[x:=\psi(b ) ] \end{aligned}\ ] ] * @xmath322 . then @xmath323}]\ ! ] } & = & { [ \![{(\alpha.n)[x:=b]}]\!]}\\ & = & { [ \![{\alpha.(n[x:=b])}]\!]}\\ & = & { \lambda f\,({\alpha.{[\![{n[x:=b]}]\!]}}\,f)}\\ & & \textrm{by the induction hypothesis}\\ & = & { \lambda f\,({\alpha.{[\![{n}]\!]}[x:=\psi(b)]}\,f)}\\ & = & ( { \lambda f\,({\alpha.{[\![{n}]\!]}}\,f)})[x:=\psi(b)]\\ & = & { [ \![{\alpha.n}]\!]}[x:=\psi(b)]\\ & = & { [ \![{m}]\!]}[x:=\psi(b ) ] \end{aligned}\ ] ] * @xmath324
. then @xmath323}]\ ! ] } & = & { [ \![{(n_1+n_2)[x:=b]}]\!]}\\ & = & { [ \![{n_1[x:=b]+n_2[x:=b]}]\!]}\\ & = & { \lambda f\,({({[\![{n_1[x:=b]}]\!]}+{[\![{n_2[x:=b]}]\!]})}\,f)}\\ & & \textrm{by the induction hypothesis}\\ & = & { \lambda f\,({({[\![{n_1}]\!]}[x:=\psi(b)]+{[\![{n_2}]\!]}[x:=\psi(b)])}\,f)}\\ & = & ( { \lambda f\,({({[\![{n_1}]\!]}+{[\![{n_2}]\!]})}\,f)})[x:=\psi(b)]\\ & = & { [ \![{n_1+n_2}]\!]}[x:=\psi(b)]\\ & = & { [ \![{m}]\!]}[x:=\psi(b ) ] \end{aligned}\ ] ] structural induction on @xmath28 . * @xmath313 . then @xmath325\!]})~k=({\lambda f\,\left(f\right)~x})~k { { \ensuremath{{\to_{a\cup\beta}}}}}(k)~x = x : k$ ] .
* @xmath326 . then @xmath327\!]})~k=({\lambda f\,\left(f\right)~\lambda x\,{[\![{n}]\!]}})~k$ ] and by definition of @xmath139 this is equal to @xmath328 .
* @xmath317 . then @xmath329\!]})~k = ( 0)~k { { \ensuremath{{\to_{a\cup\beta}}}}}0 = 0:k$ ] . * @xmath330 . then @xmath331\!]})~k = ( \lambda f\ , ( { [ \![{m'}]\!]}+{[\![{n}]\!]})~f)~k$ ] which @xmath91-reduces to @xmath189\!]}+{[\![{n}]\!]})~k { { \ensuremath{{\to_{a\cup\beta}}}}}({[\![{m'}]\!]})~k+({[\![{n}]\!]})~k $ ] which @xmath91-reduces by the induction hypothesis to @xmath332 . * @xmath333 . then @xmath334\!]})~k = ( \lambda f\ , ( \alpha .
{ [ \![{n}]\!]})~f)~k { { \ensuremath{{\to_{a\cup\beta}}}}}(\alpha . { [ \![{n}]\!]})~k$ ] which @xmath91-reduces to @xmath335\!]})~k)$ ] and this , by the induction hypothesis , @xmath91-reduces to @xmath336 . * @xmath185
. then @xmath188\!]})~k=({{\lambda{f}\,({}})~\lambda{g}\,({{[\![{m'}]\!]}})~\lambda{h}\,(({g})~{h})~{f}}{{[\![{n}]\!]}})~k$ ] which @xmath91-reduces to @xmath189\!]})~\lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k$ ] .
notice that the term @xmath190\!]})~\lambda h\,((g)~h)~k$ ] is a value , so by the induction hypothesis the above term reduces to @xmath191\!]})~\lambda h\,((g)~h)~k$ ] .
we do a second induction , over @xmath311 , to prove that @xmath191\!]})~\lambda h\,((g)~h)~k{{\ensuremath{{\to_{a\cup\beta}}}}}(m')~n : k$ ] . * * if @xmath337 , then @xmath191\!]})~\lambda h\,((g)~h)~k=((m_1)~m_2)~n : k=(m')~n : k$ ] . * * if @xmath311 is a base term , then @xmath191\!]})~\lambda h\,((g)~h)~k$ ] is equal to @xmath338\!]})~\lambda h\,((g)~h)~k)~\psi(m'){{\ensuremath{{\to_{a\cup\beta}}}}}({[\![{n}]\!]})~\lambda h\,((\psi(m'))~h)~k$ ] which by the main induction hypothesis @xmath91-reduces to @xmath339 , and this is equal to @xmath192 . * * if @xmath340 , then the term @xmath191\!]})~\lambda h\,((g)~h)~k$ ] is equal to @xmath341\!]})~\lambda h\,((g)~h)~k = \alpha.(m_1:\lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k)$ ] which by the second induction hypothesis @xmath91-reduces to @xmath342 . * * if @xmath343 , then @xmath191\!]})~\lambda h\,((g)~h)~k = m_1+m_2 : \lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k$ ] which is equal to @xmath344\!]})~\lambda h\,((g)~h)~k+m_2 : \lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k$ ] which @xmath91-reduces by the second induction hypothesis to @xmath345 . * * if @xmath346 then @xmath186\!]})~\lambda h\,((g)~h)~k= 0 : \lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k = 0= ( 0)~n : k = ( m')~n : k$ ] case by case on the rules @xmath56 . rules @xmath347 : : + * @xmath348 , with @xmath146 being a base term . then @xmath349 .
* @xmath350 , with @xmath146 base term .
then @xmath351 @xmath352 .
* @xmath353 , with @xmath146 a base term . then @xmath354 .
rules @xmath355 : : + * @xmath356 , with @xmath148 being a value
. then @xmath357 .
* @xmath358 , with @xmath148 being a value
. then @xmath359 .
* @xmath360 , with @xmath148 a value .
then @xmath361 .
rules @xmath362 and @xmath89 : : + * @xmath363
. then @xmath364 . * @xmath365
. then @xmath366 . * @xmath367 . then @xmath368 . * @xmath369
. then @xmath370 . * @xmath371
. then @xmath372 .
. then @xmath374 which @xmath55-reduces to @xmath375 . * @xmath376
. then @xmath377 . * @xmath378
. then @xmath379 . * @xmath380
. then @xmath381 .
rules @xmath382 and @xmath383 : : + * @xmath384
. then @xmath385 . * @xmath386 . then @xmath387 .
rules @xmath203 and @xmath202 : : assume @xmath388 , and assume that for all @xmath179 base term , @xmath389 .
we show that the result also holds for each contextual rule .
+ * @xmath390
. then @xmath391 .
* @xmath392 , analogous to previous case . * @xmath393 . then @xmath394 . * @xmath395 .
case by case : * * @xmath396 .
then @xmath397 which @xmath55-reduces by the induction hypothesis to @xmath398 . * * @xmath399 .
then @xmath400 . *
. then @xmath402 which @xmath55-reduces by the induction hypothesis to @xmath403 . * * @xmath213
. then @xmath404 which @xmath55-reduces by the induction hypothesis to @xmath405 .
* @xmath406 case by case : * * @xmath407 .
absurd since a base term can not reduce .
* * @xmath408 .
case by case on the possible @xmath56-reductions of @xmath28 : * * * @xmath409 with @xmath410
. then @xmath411 which by the induction hypothesis @xmath55-reduces to @xmath412 . * * * @xmath413 and @xmath414
. then @xmath415 . * * * @xmath416 and @xmath417 .
then @xmath418 . * * * @xmath419 and @xmath420
. then @xmath421 . * * * @xmath422 and @xmath346
. then @xmath423 . * * * @xmath424 and @xmath346
. then @xmath425 . * * @xmath426 .
case by case on the possible @xmath56-reductions of @xmath28 : * * * @xmath427 with @xmath428 . then @xmath429 which by the induction hypothesis @xmath55-reduces to @xmath430 . * * * @xmath431 with @xmath432 .
analogous to previous case .
* * * @xmath433 and @xmath434 . then @xmath435 and this @xmath55-reduces to @xmath436 . * * * @xmath437 and @xmath438 .
analogous to previous case . *
* * @xmath439
. then @xmath440 . * * * @xmath441 , @xmath442 and @xmath443
. then @xmath444 . * * * @xmath441 , @xmath445 and @xmath446 .
analogous to previous case . *
* * @xmath447 and @xmath448 .
analogous to previous case . *
* @xmath317 .
absurd since @xmath26 does not reduce .
* * @xmath449 .
then the term @xmath450 is equal to @xmath451\!]})~\lambda h\,((g)~h)~k$ ] , which by the induction hypothesis @xmath55-reduces to @xmath191\!]})~\lambda h\,((g)~h)~k$ ] .
we do a second induction , over @xmath311 , to prove that @xmath191\!]})~\lambda h\,((g)~h)~k{{\ensuremath{{\to_{a}}}}}(m')~n : k$ ] . *
* * if @xmath452 , then @xmath191\!]})~\lambda h\,((g)~h)~k$ ] is equal to @xmath453 .
* * * @xmath311 can not be a base term since from @xmath454 it is not possible to arrive to a base term using only @xmath56 . * * * if @xmath455 , then @xmath191\!]})~\lambda h\,((g)~h)~k= \alpha.m'_1:\lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k=\alpha.(m'_1:\lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k)$ ] which @xmath55-reduces by the induction hypothesis to @xmath456 . * * * if @xmath457 , then the term @xmath191\!]})~\lambda h\,((g)~h)~k$ ] is equal to @xmath458\!]})~\lambda h\,((g)~h)~k$ ] which is equal to @xmath459\!]})~\lambda h\,((g)~h)~k+m'_2 : \lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k$ ] which @xmath55-reduces by the induction hypothesis to @xmath460 . * * * if @xmath346 then @xmath191\!]})~\lambda h\,((g)~h)~k$ ] is equal to @xmath461\!]})~\lambda h\,((g)~h)~k = 0= ( 0)~n : k = ( m')~n : k$ ] case by case on the rules of @xmath2 .
rule @xmath201 : : @xmath462\!]})\,\psi(b))\,k\\ & { { \ensuremath{{\to_{\beta_n}}}}}&{[\![{m}]\!]}[x:=\psi(b)]\,k\\ ( \textrm{lemma~\ref{2:lem : substitution}})&=&{[\![{m[x:=b]}]\!]}\,k\\ ( \textrm{lemma~\ref{lem : lemma2 } } ) & { { \ensuremath{{{\to_{a\cup\beta}^{\textrm{$\ast$ } } } } } } } & m[x:=b]:k \end{aligned}\ ] ] algebraic rules : : if @xmath102 , then by lemma [ lem : lemma3alg ] @xmath195 which implies that @xmath199 . rules
@xmath203 and @xmath202 : : if @xmath388 , then we use lemma [ lem : lemma3alg ] to close the case .
assume @xmath463 , and assume that for all @xmath179 base term , @xmath464 .
we show that the result also holds for each contextual rule .
+ * @xmath465
. then @xmath466 .
* @xmath467 , analogous to previous case .
. then @xmath469 . *
case by case : * * @xmath396 .
then @xmath397 which @xmath91-reduces by the induction hypothesis to @xmath398 .
* * @xmath399
. then @xmath400 . *
* @xmath401 .
then @xmath402 which @xmath91-reduces by the induction hypothesis to @xmath403 . * * @xmath213
. then @xmath404 which @xmath91-reduces by the induction hypothesis to @xmath405 . * @xmath471 case by case : * * @xmath407 .
absurd since a base term can not reduce . *
* @xmath408 .
the only possible @xmath472-reduction from @xmath28 is @xmath409 with @xmath473
. then @xmath411 which by the induction hypothesis @xmath91-reduces to @xmath412 .
* * @xmath426 .
case by case on the possible @xmath472-reductions of @xmath28 : * * * @xmath427 with @xmath474 .
then @xmath429 which by the induction hypothesis @xmath91-reduces to @xmath430 .
* * * @xmath431 with @xmath475 .
analogous to previous case . *
* @xmath317 .
absurd since @xmath26 does not reduce . *
* @xmath449 .
then the term @xmath450 is equal to @xmath451\!]})~\lambda h\,((g)~h)~k$ ] , which @xmath91-reduces , by the induction hypothesis , to @xmath191\!]})~\lambda h\,((g)~h)~k$ ] .
we do a second induction , over @xmath311 , to prove that @xmath191\!]})~\lambda h\,((g)~h)~k$ ] @xmath91-reduces to @xmath192 . *
* * if @xmath452 , then @xmath191\!]})~\lambda h\,((g)~h)~k$ ] is equal to @xmath453 . * * * if @xmath311 is a base term , then the term @xmath191\!]})~\lambda h\,((g)~h)~k$ ] is equal to @xmath338\!]})~\lambda h\,((g)~h)~k)~\psi(m')$ ] which @xmath91-reduces to @xmath476\!]})~\lambda h\,((\psi(m'))~h)~k$ ] which , by lemma [ lem : lemma2 ] , @xmath91-reduces @xmath477 . * * * if @xmath455 , then @xmath191\!]})~\lambda h\,((g)~h)~k= \alpha.m'_1:\lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k=\alpha.(m'_1:\lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k)$ ] which @xmath91-reduces by the induction hypothesis to @xmath456 . * * * if @xmath457 , then the term @xmath191\!]})~\lambda h\,((g)~h)~k$ ] is equal to @xmath458\!]})~\lambda h\,((g)~h)~k$ ] which is equal to @xmath459\!]})~\lambda h\,((g)~h)~k+m'_2 : \lambda g\,({[\![{n}]\!]})~\lambda h\,((g)~h)~k$ ] which @xmath91-reduces by the induction hypothesis to @xmath460 . * * * if @xmath346 then @xmath191\!]})~\lambda h\,((g)~h)~k$ ] is equal to @xmath461\!]})~\lambda h\,((g)~h)~k = 0= ( 0)~n : k = ( m')~n : k$ ] structural induction on @xmath28 .
. then @xmath478\}\hspace{-.55em}|\hspace{.3em}}={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}=x[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]={\{\hspace{-.235em}| x\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]$ ] . * @xmath316
. then @xmath479\}\hspace{-.55em}|\hspace{.3em}}=y={\{\hspace{-.235em}| y\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]$ ] . *
analogous to previous case .
* @xmath480 . then @xmath481\}\hspace{-.55em}|\hspace{.3em}}&=&{\{\hspace{-.235em}| \lambda y\,(m'[x:=n])\}\hspace{-.55em}|\hspace{.3em}}\\ & = & { \lambda f\,\left(f\right)~\lambda y\,{\{\hspace{-.235em}| m'[x:=n]\}\hspace{-.55em}|\hspace{.3em}}}\\ & & \textrm{by the induction hypothesis}\\ & = & { \lambda f\,\left(f\right)~\lambda y\,{\{\hspace{-.235em}| m'\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]}\\ & = & ( { \lambda f\,\left(f\right)~\lambda y\,{\{\hspace{-.235em}| m'\}\hspace{-.55em}|\hspace{.3em}}})[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]\\ & = & { \{\hspace{-.235em}| m\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em } } ] \end{aligned}\ ] ] * @xmath320 . then @xmath482\}\hspace{-.55em}|\hspace{.3em } } & = & { \{\hspace{-.235em}| ( ( n_1)~n_2)[x:=n]\}\hspace{-.55em}|\hspace{.3em}}\\ & = & { \{\hspace{-.235em}| ( n_1[x:=n])~n_2[x:=n]\}\hspace{-.55em}|\hspace{.3em}}\\ & = & \lambda f\,({\{\hspace{-.235em}| n_1[x:=n]\}\hspace{-.55em}|\hspace{.3em}})~\lambda g\,((g)~{\{\hspace{-.235em}| n_2[x:=n]\}\hspace{-.55em}|\hspace{.3em}})~f\\ & & \textrm{by the induction hypothesis}\\ & = & \lambda f\,({\{\hspace{-.235em}| n_1\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}])~\lambda g\,((g)~{\{\hspace{-.235em}| n_2\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}])~f\\ & = & ( \lambda f\,({\{\hspace{-.235em}| n_1\}\hspace{-.55em}|\hspace{.3em}})~\lambda g\,((g)~{\{\hspace{-.235em}| n_2\}\hspace{-.55em}|\hspace{.3em}})~f)[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]\\ & = & { \{\hspace{-.235em}| ( n_1)~n_2\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]\\ & = & { \{\hspace{-.235em}| m\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em } } ] \end{aligned}\ ] ] * @xmath483 . then @xmath482\}\hspace{-.55em}|\hspace{.3em } } & = & { \{\hspace{-.235em}| ( \alpha.m')[x:=n]\}\hspace{-.55em}|\hspace{.3em}}\\ & = & { \{\hspace{-.235em}| \alpha.(m'[x:=n])\}\hspace{-.55em}|\hspace{.3em}}\\ & = & \lambda f\ , ( \alpha.{\{\hspace{-.235em}| m'[x:=n]))\}\hspace{-.55em}|\hspace{.3em } } ) ~f\\ \textrm{by the induction hypothesis}\\ & = & \lambda f\ , ( \alpha.{\{\hspace{-.235em}| m'\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}])~f\\ & = & \lambda f\ , ( \alpha.{\{\hspace{-.235em}| m'\}\hspace{-.55em}|\hspace{.3em}})~f[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]\\ & = & { \{\hspace{-.235em}| \alpha.m'\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]\\ & = & { \{\hspace{-.235em}| m\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em } } ] \end{aligned}\ ] ] * @xmath324 . then @xmath482\}\hspace{-.55em}|\hspace{.3em } } & = & { \{\hspace{-.235em}| ( n_1+n_2)[x:=n]\}\hspace{-.55em}|\hspace{.3em}}\\ & = & { \{\hspace{-.235em}| n_1[x:=n]+n_2[x:=n]\}\hspace{-.55em}|\hspace{.3em}}\\ & = & \lambda f\ , ( { \{\hspace{-.235em}| n_1[x:=n]\}\hspace{-.55em}|\hspace{.3em}}+{\{\hspace{-.235em}| n_2[x:=n]\}\hspace{-.55em}|\hspace{.3em}})~f\\ & & \textrm{by the induction hypothesis}\\ & = & \lambda f\ , ( { \{\hspace{-.235em}| n_1\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]+{\{\hspace{-.235em}| n_2\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}])~f\\ & = & \lambda f\ , ( ( { \{\hspace{-.235em}| n_1\}\hspace{-.55em}|\hspace{.3em}}+{\{\hspace{-.235em}| n_2\}\hspace{-.55em}|\hspace{.3em}})~f [ x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]\\ & = & { \{\hspace{-.235em}| n_1+n_2\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em}}]\\ & = & { \{\hspace{-.235em}| m\}\hspace{-.55em}|\hspace{.3em}}[x:={\{\hspace{-.235em}| n\}\hspace{-.55em}|\hspace{.3em } } ] \end{aligned}\ ] ] structural induction on @xmath28 . * @xmath326 . then @xmath484 and by definition of @xmath240 this is equal to @xmath485 . * @xmath317
. then @xmath486 . *
then @xmath487 which @xmath90-reduces by the induction hypothesis to @xmath332
. * @xmath333 .
then @xmath488 which @xmath90-reduces to @xmath489 and this , by the induction hypothesis , @xmath90-reduces to @xmath336
. * @xmath185 . then @xmath490 which @xmath90-reduces to @xmath491 .
note that @xmath276 is a value , so by the induction hypothesis the above term reduces to @xmath277 .
we do a second induction , over @xmath311 , to prove that @xmath492 . * * if @xmath337 , then @xmath493 . * * if @xmath311 is a base term , @xmath494 which @xmath90-reduces to @xmath495 . * * if @xmath340 , then @xmath496 which @xmath173-reduces by the induction hypothesis to @xmath497 . * * if @xmath343 , then @xmath498 which is equal to @xmath499 which @xmath173-reduces by the induction hypothesis to @xmath345 . * * if @xmath346 then @xmath500 case by case on the rules of @xmath501 . rule @xmath201 : : @xmath502\,k\\ ( \textrm{lemma~\ref{3:lem : substitution}})&=&{\{\hspace{-.235em}| m[x:=n]\}\hspace{-.55em}|\hspace{.3em}}\,k\\ ( \textrm{lemma~\ref{lem : lemma2cps } } ) & { { \ensuremath{{{\to_{\ell\cup\beta}^{\textrm{$\ast$ } } } } } } } & m[x:=n]:k \end{aligned}\ ] ] rules @xmath503 : : * let @xmath504 . @xmath505 . *
let @xmath506 .
@xmath507 * let @xmath508 .
@xmath509 rules @xmath362 and @xmath89 : : + * @xmath510
. then @xmath511 . * @xmath512
. then @xmath513 .
. then @xmath515 . * @xmath516
. then @xmath517 . * @xmath518
. then @xmath519 .
. then @xmath374 which @xmath90-reduces to @xmath375 . * @xmath521
. then @xmath522 .
. then @xmath524 . * @xmath525
. then @xmath526 .
rules @xmath382 and @xmath383 : : + * @xmath527 . then @xmath528 . *
. then @xmath530 .
rules @xmath203 : : assume @xmath531 , and that for all @xmath179 base term , @xmath532 . we show that the result also holds for each contextual rule .
+ * @xmath533
. then @xmath534 . * @xmath535 , analogous to previous case . *
. then @xmath537 . * @xmath538 case by case : * * @xmath407 .
absurd since a base term can not reduce .
* * @xmath408 .
case by case on the possible @xmath91-reductions of @xmath28 : * * * @xmath409 with @xmath539
. then @xmath411 which by the induction hypothesis @xmath90-reduces to @xmath412 . * * * @xmath413 and @xmath414
. then @xmath540 . * * * @xmath416 and @xmath417 .
then @xmath541 . * * * @xmath419 and @xmath420
. then @xmath542 . * * * @xmath422 and @xmath346
. then @xmath543 . * * * @xmath424 and @xmath346
. then @xmath544 . * * @xmath426 .
case by case on the possible @xmath91-reductions of @xmath28 : * * * @xmath427 with @xmath545 . then @xmath429 which by the induction hypothesis @xmath90-reduces to @xmath430 . * * * @xmath431 with @xmath546 .
analogous to previous case . *
* * @xmath433 and @xmath434 . then @xmath435 which @xmath90-reduces to @xmath436 . * * * @xmath437 and @xmath438 . analogous to previous case . * * * @xmath439
. then @xmath547 . * * * @xmath441 , @xmath442 and @xmath443 .
then @xmath548 . * * * @xmath441 , @xmath445 and @xmath446 .
analogous to previous case . *
* * @xmath447 and @xmath448 .
analogous to previous case . *
* @xmath317 .
absurd since @xmath26 does not reduce .
* * @xmath449 . then @xmath450 is equal to @xmath549 , which by the induction hypothesis @xmath90-reduces to @xmath277
. we do a second induction , over @xmath311 , to prove that @xmath492 . *
* * if @xmath452 , then @xmath550 . * * * if @xmath311 is a base term , then @xmath277 is equal to @xmath551 . * * * if @xmath455 , then @xmath552 is equal to @xmath553 which @xmath173-reduces by the induction hypothesis to @xmath456 . * * * if @xmath457 , then @xmath554 which is equal to @xmath555 which @xmath173-reduces by the induction hypothesis to @xmath460 . *
* * if @xmath346 then @xmath500
the proof of the local confluence of the algebraic fragments of @xmath19 and @xmath15 are sufficiently monotonous so that one can ask a proof assistant to do them . for this purpose
we use the library locconf setting up some convenient tactics .
the interested reader can find the whole set of files in @xcite : hypothesis s_0_1_dec : ~ s1 = s0 .
hypothesis s_0_lunit : forall a , s0 + a = a. hypothesis s_0_lelim : forall a , s0 * a = s0 .
hypothesis s_1_lunit : forall a , s1 * a = a. hypothesis s_rdistrib : forall a b c , a*(b+c ) = ( a*b)+(a*c )
. hypothesis s_ldistrib : forall a b c , ( a+b)*c = ( a*c)+(b*c ) .
hypothesis s_add_assoc : forall a b c , ( a+b)+c = a+(b+c ) .
hypothesis s_mul_assoc : forall a b c , ( a*b)*c = a*(b*c ) .
hypothesis s_add_commut : forall a b , a+b = b+a
. hypothesis s_mul_commut : forall a b , a*b = b*a . notation " a + b " : = ( tadd a b ) : term_scope .
notation " a ' * * ' b " : = ( tmul a b ) ( at level 35 ) : term_scope . notation " @ a " : = ( tvar a ) ( at level 10 ) : term_scope . notation " a ; b " : = ( tapply a b ) ( at level 30 ) : term_scope .
notation " \ a " : = ( tlambda a ) ( at level 40 ) : term_scope .
( * * elementary rules * ) definition r_t0_runit : = forall t , r ( t + t0 ) t. definition r_s0_anni : = forall t , r ( s0 * * t ) t0 .
definition r_s1_unit : = forall t , r ( s1 * * t ) t. definition r_t0_anni : = forall a , r ( a * * t0 ) t0 .
definition r_mul_abs : = forall a b t , r ( a * * ( b * * t ) ) ( ( a * s b ) * * t ) .
definition r_ma_dist : = forall a s t , r ( a * * ( s + t ) ) ( a * * s + a * * t ) .
( * * factorization * ) definition r_add_fact : = forall a b t , r ( a * * t + b * * t ) ( ( a + s b ) * * t ) .
definition r_add_fact1 : = forall a t , r ( a * * t + t ) ( ( a + s s1 ) * * t ) .
definition r_add_fact11 : = forall t , r ( t + t ) ( ( s1 + s s1 ) * * t ) .
( * * assoc . and commut .
of addition * ) definition r_add_com : = forall s t , r ( s + t ) ( t + s ) .
definition r_add_rassoc : = forall r s t , r ( ( r + s ) + t ) ( r + ( s + t ) ) .
definition r_add_lassoc : = forall r s t , r ( r + ( s + t ) ) ( ( r + s ) + t ) .
( * * congruence * ) definition r_cong_mul : = forall a s t , r s t - > r ( a**s ) ( a**t ) .
definition r_cong_ladd : = forall u s t , r s t - > r ( s+u ) ( t+u ) .
definition r_cong_radd : = forall u s t , r s t - > r ( u+ s ) ( u+t ) .
definition r_cong_lapp : = forall u s t , r s t - > r ( s;u ) ( t;u ) .
definition r_cong_rapp : = forall u s t , is_value u - > r s t - > r ( u;s ) ( u;t ) .
( * * linearity of application * ) definition r_add_app_ldist : = forall r s t , is_value
t - > r ( ( r + s);t ) ( r;t + s;t ) .
definition r_mul_app_ldist : = forall a r s , is_value s - > r ( ( a**r);s ) ( a**(r;s ) ) .
definition r_t0_app_ldist : = forall s , is_value s - > r ( t0;s ) t0 .
definition r_add_app_rdist : = forall r s t , is_base t - > r
( t;(r + s ) ) ( t;r + t;s ) .
definition r_mul_app_rdist : = forall a r s , is_base s - > r ( s;(a**r ) ) ( a**(s;r ) ) .
definition r_t0_app_rdist : = forall s , is_base s - > r ( s;t0 ) t0 .
( * * elementary rules * ) definition r_t0_runit : = forall t , r ( t + t0 ) t. definition r_s0_anni : = forall t , r ( s0 * * t ) t0 .
definition r_s1_unit : = forall t , r ( s1 * * t ) t. definition r_t0_anni : = forall a , r ( a * * t0 ) t0 .
definition r_mul_abs : = forall a b t , r ( a * * ( b * * t ) ) ( ( a * s b ) * * t ) .
definition r_ma_dist : = forall a s t , r ( a * * ( s + t ) ) ( a * * s + a * * t ) .
( * * factorization * ) definition r_add_fact : = forall a b t , r ( a * * t + b * * t ) ( ( a + s b ) * * t ) .
definition r_add_fact1 : = forall a t , r ( a * * t + t ) ( ( a + s s1 ) * * t ) .
definition r_add_fact11 : = forall t , r ( t + t ) ( ( s1 + s s1 ) * * t ) .
( * * assoc . and commut .
of addition * ) definition r_add_com : = forall s t , r ( s + t ) ( t + s ) .
definition r_add_rassoc : = forall r s t , r ( ( r + s ) + t ) ( r + ( s + t ) ) .
definition r_add_lassoc : = forall r s t , r ( r + ( s + t ) ) ( ( r + s ) + t ) .
( * * congruence * ) definition r_cong_mul : = forall a s t , r s t - > r ( a**s ) ( a**t ) .
definition r_cong_ladd : = forall u s t , r s t - > r ( s+u ) ( t+u ) .
definition r_cong_radd : = forall u s t , r s t - > r ( u+ s ) ( u+t ) .
definition r_cong_lapp : = forall u s t , r s t - > r ( s;u ) ( t;u ) .
( * * linearity of application * ) definition r_add_app_ldist : = forall r s t , r ( ( r + s);t ) ( r;t + s;t ) .
definition r_mul_app_ldist : = forall a r s , r ( ( a**r);s ) ( a**(r;s ) ) .
definition r_t0_app_ldist : = forall s , r ( t0;s ) t0 . | we examine the relationship between the _ algebraic @xmath0-calculus _ , a fragment of the differential @xmath0-calculus and the _ linear - algebraic @xmath0-calculus _ , a candidate @xmath0-calculus for quantum computation .
both calculi are algebraic : each one is equipped with an additive and a scalar - multiplicative structure , and their set of terms is closed under linear combinations .
however , the two languages were built using different approaches : the former is a call - by - name language whereas the latter is call - by - value ; the former considers algebraic equalities whereas the latter approaches them through rewrite rules . in this paper , we analyse how these different approaches relate one to the other . to this end , we propose four canonical languages based on each of the possible choices : call - by - name versus call - by - value , algebraic equality versus algebraic rewriting .
we show that the various languages simulate one another . due to subtle interaction between beta - reduction and algebraic rewriting , to make the languages consistent some additional hypotheses such as confluence might be required .
we carefully devise the required properties for each proof , making them general enough to be valid for any sub - language satisfying the corresponding properties . |
vortices of persistent current in superconductors have been viewed for a long time as a testing ground for various models of classical and quantum phase transitions .
this is due to relatively strong interactions between vortices coupled with a high degree of control over the vortex arrays .
for example , inter - vortex interactions in the presence of a periodic external potential ( created experimentally by modulation of the superconducting film thickness @xcite ) made vortices a convenient target for investigation of commensurability transitions.@xcite theoretical studies of the effects of randomness on vortex structure and vortex dynamics have led to the notion of collective pinning,@xcite with importance stretching well beyond the physics of superconductivity.@xcite later , the discovery of high - temperature superconductors triggered studies of the vortex lattice melting transition and glassy behavior@xcite in layered superconductors as well as structural transitions of vortices interacting with columnar defects.@xcite because vortices in continuous superconductors have normal cores , vortex motion is dissipative@xcite and much of the above - mentioned work used classical statistical mechanics to address collective phenomena in vortex arrays.@xcite the effects of quantum fluctuations of vortices in continuous films become important only close to the supeconductor - insulator transition , which requires special tuning of the films normal - state resistance.@xcite in arrays of josephson junctions , however , vortices do not have cores , allowing for the study of quantum fluctuations .
an array of small superconducting islands connected by josephson junctions , where the island charging energies were sufficiently large for quantum effects to be important,@xcite was used in attempts to observe a quantum kosterlitz - thouless ( kt ) transition @xcite and mott transition@xcite in a 1d array .
while these works advanced the nanofabrication techniques needed to produce highly regular arrays of small josephson junctions , the measurement results were ambiguous .
the current - voltage ( @xmath0-@xmath1 ) characteristics gave inconclusive evidence for a kt transition in single - line 1d arrays.@xcite the observations of ref .
were later interpreted to be consistent with a purely classical commensurability transition rather than the quantum mott transition.@xcite the suppression of quantum effects in these experiments stemmed from the low inductance of the continuous superconducting wires , which were necessary to make the josephson junction arrays .
we should note , also , that in the majority of experiments the evidence for the various classical and quantum transitions mentioned above was based on signatures in @xmath0-@xmath1 characteristics .
this method is limited to addressing highly averaged quantities , and relies on substantial deviations of the investigated system from equilibrium ( needed , for example , to overcome the static pinning of vortices ) .
recent developments in superconducting qubit techniques offer the possibility of overcoming the described limitations of previous experimental studies of many - body physics of vortices .
typical superconducting qubit experiments address the superconducting system using microwaves.@xcite the system is only weakly perturbed by the microwave excitation ; hence , this spectroscopic approach allows one to probe the system close to equilibrium . on the circuit element side ,
the development of the fluxonium qubit,@xcite which combines a josephson junction with a superinductor@xcite ( i.e. , an element exhibiting high inductance and low capacitance ) opens avenues for studying quantum effects in superconducting arrays . in this work ,
we study theoretically a one - dimensional array consisting of superinductors and a chain of small josephson junctions as shown in fig .
[ fig01 ] .
two parameters characterize such an array : the ratios of the josephson ( @xmath2 ) and inductive ( @xmath3 ) energies ( see sec . [
sec : model ] ) , combined into a characteristic length @xmath4 and the ratio @xmath5 of charging and josephson energies , which controls the quantum fluctuations . despite its simplicity , this model allows for a variety of phase transitions of a classical or quantum nature ( depending on the ratio @xmath5 ) as a function of the applied magnetic flux @xmath6 per plaquette of the array .
classically , for fixed @xmath7 , there is a critical magnetic field @xmath8 at which the system undergoes a transition from a homogeneous state with no persistent currents through the inductors to a state with static persistent currents . as a function of the characteristic length @xmath9 , eq .
( [ x_ell ] ) , we discuss two regimes , cf .
[ fig01](b ) : a `` type - i '' regime ( @xmath10 ) featuring a second - order transition to a state of staggered persistent currents as shown in fig .
[ fig_vortices](a ) , and a `` type - ii '' regime ( @xmath11 ) , in which the transition takes place by the sequential ( first - order ) introduction of localized vortices of persistent currents in the lattice plaquettes , fig .
[ fig_vortices](b ) .
these vortices correspond to _ kinks _ , meaning discontinuous jumps of height @xmath12 , in the node phase ( or node flux ) @xmath13 , see fig .
[ fig01](a ) .
repulsive interactions between kinks lead to a series of pinned commensurate phases with increasing @xmath6 . in the presence of quantum fluctuations , the number of kinks fluctuates and the initial transition at @xmath8 turns into a kt transition .
this is followed by commensurate - incommensurate transitions between classical pinned phases and quantum liquid phases of floating crystalline cells of a kink lattice .
each of the phases carries a `` fingerprint '' in the microwave absorption spectrum , as the nature of low - energy excitations is sensitive to the types of phases and the transitions between them .
the microwave spectra also carry information about the crossover from classical to quantum critical behavior in the vicinity of the transition .
yet another advantage of the spectroscopic approach is that it is a linear response to a weak perturbation . in this work
, we will highlight the signatures of phase transitions that can be measured using microwave photons .
the paper is organized as follows : in sec .
[ sec : model ] , we formulate the mathematical model for the circuit under consideration .
we also introduce and discuss two methods , capacitative and inductive , for coupling the circuit to an external resonator , and give general forms for the radiation absorption rate in the two cases . in sec .
[ sec : type - i ] , we describe the type - i regime , with @xmath9 comparable to a lattice spacing . in this limit
, the low energy excitations are gapped plasmon oscillations , where the node phases @xmath13 undergo small fluctuations .
the plasmon excitations soften at the critical magnetic flux @xmath8 , leading to a quantum ising transition between a phase with the magnetic fluxes @xmath14 ( cf .
[ fig01 ] ) everywhere and a staggered phase @xmath15 , with order parameter @xmath16 .
section [ sec : type - ii ] discusses the type - ii ( @xmath17 ) regime . here
, the low - energy excitations are associated with the addition or removal of kinks . as the magnetic field
is increased from @xmath18 , the cost of creating a kink is reduced , leading to a visible peak in the absorption spectrum below the plasmon continuum . at a critical field , this energy cost vanishes , leading to a proliferation of kinks and the formation of a series of gapped and gapless crystalline phases of kinks .
each of these has an observable signature in the excitation spectrum for adding or removing kinks .
finally , quantum effects such as broadening of peaks in the excitation spectrum and the appearance of phases of incommensurate quantum liquids are discussed .
a discussion of our analysis and its results is presented in sec .
[ sec : discussion ] . .
( b ) phase diagram for the classical ground state , cf .
@xcite for the phase diagram of similar classical models . at inductances @xmath19
such that @xmath20 , the system is in the `` type - i '' regime that features a single ising transition from a homogeneous phase into a phase of staggered persistent currents . for large
@xmath19 or @xmath17 ( `` type - ii '' regime ) , the system , as a function of external flux @xmath6 , develops subsequent phases of commensurate lattices of vortices or _
kinks_. the rational numbers describe the kink density in the node flux configurations @xmath21 .
darker shaded regions contain phases with higher denominators that complete the devil s staircase .
the line separating the homogeneous and inhomogeneous phases is given by eq .
( [ phiec_ising ] ) in the type - i and by eq .
( [ x_phiec ] ) in the type - ii regime . ] ) ordered phase in the type - i regime .
( b ) a vortex of static persistent currents decaying over the length @xmath9 , stable in the type - ii regime . ]
the circuit in fig . [ fig01](a ) is described by the lagrangian @xmath22 & = \frac{\hbar^2}{2e_c}\sum_{j=0}^{n-1}\big(\dot{\phi}_{j}-\dot{\phi}_{j-1}\big)^2 - v[\phi ] \ , \label{x_lagrangian}\end{aligned}\ ] ] where @xmath23 with @xmath24 the josephson junction capacitance , and we formally set @xmath25 , assuming periodic boundary conditions .
the potential in ( [ x_lagrangian ] ) takes the form @xmath26 & = \sum_{j=0}^{n-1}\big\ { \frac{e_l}{2 } \phi_j^2 - e_j\big[\cos\big(\theta_j-\phi_e\big)-1\big ] \big\ } \label{x_potential}\end{aligned}\ ] ] where @xmath27 , @xmath19 is the inductance of the inductors and @xmath28 with @xmath29 denoting the magnetic flux per plaquette and @xmath30 being the flux quantum . furthermore , we define @xmath31 , the phase difference across the josephson junction between nodes @xmath32 and @xmath33 , which in the following we denote as _ link _ @xmath33 .
the potential @xmath34 $ ] is not invariant under @xmath35 , as would be expected for a superconducting system .
this is because we neglect the phase slip processes that allow the inductors to relax to the true ground state of the system .
superinductors such as those used in a fluxonium qubit are engineered so that this is a valid approximation.@xcite at large @xmath24 , the physics is dominated by the ( meta)stable configurations @xmath21 that minimize the static potential @xmath34 $ ] .
the magnetic flux @xmath6 and the characteristic length scale @xmath9 , defined in eq .
( [ x_ell ] ) , are the only parameters in this `` classical '' regime . throughout our theoretical analysis
, we assume the limit of large systems , @xmath36 .
note that , due to the inductive potential , a configuration with local flux @xmath13 is physically distinct from a state with flux @xmath37 with integer @xmath38 . in the ground state , each variable @xmath13 will thus take values between @xmath39 and @xmath40 .
furthermore , minimal configurations @xmath21 necessarily satisfy @xmath41 , which physically corresponds to current conservation . by coupling the circuit in fig .
[ fig01](a ) to microwaves , signatures of the various phases and phase transition become observable in absorption spectroscopy .
additional interest in such experiments may arise because of well - defined peaks in the absorption spectrum below the plasma gap .
we suggest two , in a certain way complementary , schemes of coupling@xcite the circuit to a microwave resonator : ( a ) inductive coupling and ( b ) capacitive coupling , see fig .
[ fig_antennas ] .
in the situation of fig .
[ fig_antennas](a ) , we add an antenna capacitively coupled to two neighboring nodes @xmath42 and @xmath43 , which will couple the microwave mode to the josephson junction at link @xmath43 .
the hamiltonian for the the microwave mode ( with creation and annihilation operators @xmath44 and @xmath45 ) and its coupling to the circuit takes the form @xmath46 herein , @xmath47 is the frequency of the microwave mode and @xmath48 is the coupling constant , which is determined by the impedance of the microwave mode and the coupling capacitance.@xcite @xmath49 is the cooper pair number operator for the antenna josephson junction at link @xmath43 .
using fermi s golden rule , the absorption rate for driven radiation with frequencies @xmath50 incident on the antenna has the general form @xmath51 with the sum being over the excited states @xmath52 with energy @xmath53 relative to the ground state . for the inductive scheme , fig .
[ fig_antennas](b ) , we add inductances @xmath54 into the former `` ground line '' such that the system remains translationally symmetric .
the antennas themselves may couple inductively to one or several of these inductances . here
we suppose it solely couples to one link , say @xmath43 . the inductances @xmath54 should be chosen small such that the length @xmath55 is not only large itself ( in comparison with unity ) but also @xmath56 ( implying @xmath57 ) , where @xmath9 is the characteristic length defined in eq .
( [ x_ell ] ) . in this limit ,
as far as the ground state and excitation energies are concerned , the inductances @xmath54 merely lead to a renormalized characteristic length @xmath58 . with good accuracy , we may thus set @xmath59 and ignore the effects of the inductances @xmath54 in the analysis of the model ( [ x_lagrangian ] ) . using inductively coupled antennas , the microwave modes ( described by @xmath45 and @xmath44 ) effectively couple to the josephson phase differences @xmath60 instead of cooper pair numbers @xmath61 , cf .
( [ eq : antenna_hamiltonian ] ) .
currents induced by the antenna at link @xmath43 decay only over the ( large ) distance @xmath62 , eq .
( [ ella ] ) .
this leads to the following form of the absorption rate : @xmath63 with @xmath64 where @xmath65 with @xmath66 and @xmath67 is the fourier transform of @xmath60 .
the last approximation is valid in the long - wavelength limit .
the coupling constant @xmath68 in eq .
( [ absorptionrate_l ] ) is determined by the impedance of the microwave mode and the coupling inductances .
in contrast to the capacitive coupling , which _ locally _ couples the microwave mode to one josephson link , the inductive scheme has a much _ longer range _
@xmath62 , eq . ( [ ella ] ) .
we will specify the spectral absorption rates for the capacitively coupled antenna , @xmath69 , and the inductively coupled antenna , @xmath70 , for the type - i regime in sec .
[ sec : type - i ] and type - ii regime in sec .
[ sec : type - ii ] . in the following
, we will use units in which @xmath71 .
for @xmath20 , as one increases the external flux @xmath6 , the system undergoes a second - order phase transition from a low - flux homogeneous configuration @xmath72 , corresponding to zero persistent current through the inductors , into a staggered order @xmath73 with an alternating persistent current through the inductors ; see fig .
[ fig_vortices](a ) . this state
is classically connected to the state with kink density @xmath74 at larger @xmath9 in the type - ii regime , cf .
the phase diagram in fig .
[ fig01](b ) . in this type - i regime , close to @xmath8 , the relevant low - energy excitations are plasmons , quantized small fluctuations in @xmath13 .
mean - field theory , which assumes that these fluctuations are small , predicts that the transition happens at the critical flux @xmath8 given by @xmath75 technically , it is more convenient to work with the field @xmath31 , which represents the phase differences over the josephson links . for @xmath76 , the field @xmath60 fluctuates around the ordered configuration @xmath77 . for @xmath78 , mean - field theory for the phase differences @xmath60
gives the plasmon spectrum : @xmath79 where , in the continuum approximation , the wave numbers are @xmath80 .
the lowest energy states occur for momentum @xmath81 .
for momenta @xmath82 close to @xmath40 , the spectrum is simplified to @xmath83 herein , @xmath84 is the plasmon gap with the second line showing that mean - field theory predicts the gap closing as a square root as @xmath85 . at the transition , the softened low - energy plasmon modes have a linear dispersion relation @xmath86 , where @xmath87 is the sound velocity . on the ordered side , for @xmath76 , plasmons describe the fluctuations of the order parameter @xmath16 . a similar mean - field theory calculation yields for momenta @xmath82 with @xmath88 a soft - mode dispersion relation of the same form as ( [ plasmon - spectrum - soft ] ) , with the gap larger by a factor of @xmath89 , @xmath90 but otherwise behaves as a function of @xmath91 in the same way as the gap ( [ plasmon - gap ] ) on the disordered side . in absorption spectroscopy ,
microwave photons excite plasmons so that we expect manifestations of the critical point at @xmath92 in the spectral absorption rate @xmath69 , eq .
( [ absorptionrate ] ) , or @xmath70 , eq .
( [ absorptionrate_l ] ) , for capacitive or inductive coupling to the microwave resonator . specifically , we may write the cooper pair number operator @xmath93 and its conjugate , the phase difference @xmath94 , in terms of the plasmon modes , @xmath95 where @xmath96 is the annihilation operator for a plasmon at wave number @xmath82 . inserting eq .
( [ eq : plasmon_number ] ) into eqs .
( [ absorptionrate ] ) , we find that at @xmath78 the spectral absorption rate from the ground state is given by @xmath97 for the capacitively coupled antenna , where @xmath98 is the plasmon density of states . for inductive coupling , inserting eq .
( [ eq : theta ] ) into eq .
( [ absorptionrate_l ] ) , we find @xmath99 for the softest plasmons with momentum @xmath82 near @xmath40 , i.e. @xmath88 . comparing eqs .
( [ fc_plasmons ] ) and ( [ fl_plasmons ] ) , we infer that the inductive - coupling scheme leads to stronger response at low energies @xmath100 and thus to higher - contrast results close to the transition at @xmath8 .
the plasmon spectrum ( [ plasmon - spectrum - soft ] ) implies the density of states @xmath101 where @xmath102 denotes the heaviside step function . in the frequency region @xmath103
, the absorption rate features a square - root singularity , @xmath104 at the critical field @xmath8 , where the gap closes , @xmath105 , the van - hove singularity disappears and eq .
( [ dos_plasmons ] ) becomes @xmath106 as plasmons have become soft acoustic modes .
mean - field theory is valid as long as fluctuations are small .
close to the critical flux @xmath8 , plasmon fluctuations become significant as the modes at @xmath107 soften . introducing @xmath108
, we thus write an effective theory for the `` slow '' field @xmath109 , keeping only the quadratic leading order in `` discrete gradients '' @xmath110 so that @xmath111 $ ] .
the euclidean action then reads @xmath112 \big\ } \label{action}\end{aligned}\ ] ] with @xmath113 at zero temperature . outside a ginzburg region close to the critical field @xmath8
, one can treat the action ( [ action ] ) in the saddle - point approximation and the mean - field results from the preceding section become accurate . in order to determine the ginzburg region , we expand the cosine - potential to fourth order in @xmath109 and then employ a continuum approximation , @xmath114 and @xmath115 .
rescaling coordinates and fields so that they become dimensionless and so the prefactors of the quadratic terms are @xmath116 , we find that close to @xmath8 the prefactor of the @xmath117 term is small if @xmath118 for external fluxes @xmath6 satisfying eq .
( [ ginzburg ] ) , the mean - field results of the preceding sections are valid . very close to @xmath8 , this ginzburg criterion breaks down as quantum fluctuations become strong .
the quantum critical behavior , which is due to the non - linearity of the josephson current - phase relationship , leads to an excitation spectrum that is considerably different from mean - field theory and should correspond to a @xmath119-dimensional quantum ising chain.@xcite as a result , e.g. , the gap @xmath120 in the plasmon spectrum , cf .
( [ plasmon - gap ] ) , is expected to close at @xmath8 as @xmath121 i.e. with critical exponent @xmath122 instead of @xmath123 in the mean - field prediction ( [ plasmon - gap ] ) .
furthermore , the @xmath124-term in the action ( [ action ] ) will become renormalized , effectively shifting the critical flux @xmath8 to a higher value inside the ginzburg region . at criticality , @xmath92 ,
the system is a liquid with spectrum @xmath125 and plasmon density of states of the form of eq .
( [ critical_f ] ) , but in the fluctuation regime , the renormalized sound velocity @xmath126 has to be considered an effective phenomenological parameter . while parameters entering the observable quantities @xmath127 and @xmath128 , eqs .
( [ fc_plasmons ] ) and ( [ fl_plasmons ] ) , will be effective ones , the qualitative threshold behavior should still be described in terms of eqs .
( [ dos_vh ] ) and ( [ critical_f ] ) .
observability of quantum critical behavior requires the system size @xmath129 to be larger than the correlation length at the boundaries of the quantum critical region as given by eq .
( [ ginzburg ] ) .
this leads to the condition @xmath130 for typical parameters@xcite for @xmath2 and @xmath131 ( and @xmath132 in the type - i regime ) , the right - hand side of this estimate is of order unity .
here we consider the limit @xmath11 , corresponding to large inductances @xmath19 .
this regime is realizable with superinductors as demonstrated in fluxonium qubits.@xcite the most interesting physical effects are due to the proliferation of _ kinks _ in the node phases @xmath13 corresponding to current vortices as shown in fig . [ fig_vortices](b ) .
these vortices pick up currents over many plaquettes of the order of @xmath9 and therefore are stable already at small external fluxes @xmath6 . thus
, phase transitions associated with kink proliferation preempt the instability driven by fluctuations of plasmons , which remain gapped for all external magnetic fluxes @xmath6 . as a result ,
the ground state and excitations are fundamentally different from the type - i regime .
we begin with the study of the phase diagram and the absorption spectrum in the `` classical '' limit . by `` classical '' , we mean that the capacitances are large enough ( and hence @xmath131 is small enough ) to make effects due to spontaneous quantum phase slips negligible but still allow for induced phase slips by microwave absorption .
quantum effects due to finite capacitances alter the classical picture and excitation spectrum and will be investigated in sec .
[ sec : quantum ] .
the classical phase diagram in fig .
[ fig01](b ) is obtained by finding the configurations @xmath21 of node fluxes that minimize the potential @xmath34 $ ] , eq .
( [ x_potential ] ) , cf . refs . .
they are found from solving the set of equations given by @xmath133\ .
\nonumber\end{aligned}\ ] ] summation over @xmath33 of eq .
( [ dv ] ) yields the constraint @xmath41 , which corresponds to zero net current to ground . at zero external flux @xmath6
, the ground state of the system is given by the homogeneous configuration @xmath72 , which remains a _
local _ minimum of @xmath34 $ ] for non - zero @xmath6 as long as @xmath134 . in the limit of large @xmath17 ,
( [ dv ] ) admits non - trivial solutions already at small @xmath135 that contain _ kinks_,@xcite static local configurations of @xmath13 with a jump of the order of @xmath12 across one link . in order to study such soliton solutions ,
we employ villain s approximation@xcite , in which we expand @xmath136 with integer @xmath137 such that @xmath138 is small . for a _ single _ kink , with one jump of @xmath139 over the link between sites @xmath140 and @xmath141 and @xmath142 for all @xmath143 , villain s approach yields the configuration @xmath144 assuming that @xmath145 . from eq .
( [ x_singlekink ] ) , we see that @xmath9 , eq .
( [ x_ell ] ) , determines the scale of the width of a kink . inserting typical fluxonium values ,
this width is @xmath146 , @xcite which sets the smallest system size necessary to observe the physics under discussion . as the external flux @xmath6
is increased , we may expect the kink solution @xmath147 , eq .
( [ x_singlekink ] ) , to become more favorable than the homogeneous configuration @xmath72 because of the josephson junction s preference for finite flux gradients at finite @xmath6 , cf .
( [ x_potential ] ) .
in fact , for the difference @xmath148-v[0]$ ] between the potential energies of the single kink and homogeneous configuration we find @xmath149 where @xmath150 is the critical external flux in the type - ii regime , @xmath151 , cf .
[ fig01](b ) . at flux
@xmath8 , configurations @xmath21 with kinks become energetically favorable over the homogeneous configuration .
the critical flux @xmath8 is analogous to the critical magnetic field @xmath152 in type - ii superconductors , when vortices begin proliferating.@xcite its smallness in @xmath153 reflects that the vortex is able to pick up currents over the ( large ) length scale @xmath9 , cf .
[ fig_vortices](b ) . for @xmath76 , the density of kinks in the ground state grows continuously as a function of @xmath6 .
each kink , when nucleating individually , brings an energy gain of @xmath154 . on the other hand
, this individual gain has to be balanced with the interaction energy between two kinks . for the repulsive potential for two kinks at links @xmath155 and @xmath33
we find @xmath156 which decays only at large distances on the scale of @xmath9 , eq .
( [ x_ell ] ) . the competition between kink - kink repulsion @xmath157 , eq .
( [ x_j ] ) , and the single - kink energy @xmath158 , eq .
( [ delta ] ) , which plays the role of a chemical potential , completely determines the ground state as a function of the external flux @xmath6 at @xmath17 .
this interplay may be effectively described in terms of a classical spin chain model .
specifically , we associate with each josephson link @xmath33 a pseudo - spin that distinguishes whether there is a kink across it ( `` spin - up '' ) or not ( `` spin - down '' ) , see fig .
[ fig04 ] .
the effective spin hamiltonian then has the form @xmath159 where @xmath160 and we introduce pauli matrices @xmath161 , @xmath162 , and @xmath163 . configuration for a commensurate kink density @xmath164 at @xmath165 .
( b ) the same state represented in the effective spin ladder model for the coupled josephson links .
links with upward orientied spin feature a kink in @xmath13 . ]
the ground states of an ising chain with infinite - range convex interaction potentials were systematically studied as a function of the `` chemical potential '' @xmath158 by bak and bruinsma@xcite and aubry@xcite . in the language of our model , they showed that in the limit of very large systems @xmath166 , the kink density @xmath167 depends on @xmath6 in the form of a devil s staircase:@xcite this means that @xmath168 only takes values in the rational numbers and rises monotonically and continuously with @xmath6 such that for each rational @xmath169 , there is a finite interval in @xmath6 in which @xmath170 . for our effective model ( [ x_clhamiltonian ] ) , this interval has the width @xmath171 figure [ fig02](a ) illustrates the devil s staircase for fixed @xmath9 , showing as a function of @xmath6 the kink density up to commensurability order @xmath172 .
a ground state configuration @xmath21 with kink density @xmath173 is periodic,@xcite @xmath174 .
for example the ground state configuration for @xmath175 is @xmath176 , with `` primitive cell '' @xmath177 . for the non - unit fraction @xmath178
, a primitive cell contains a non - trivial basis : @xmath179 . for the classical ground state as a function of the external flux @xmath6 ( for @xmath180 fixed ) .
the plot shows all densities with periodicity @xmath181 ; higher - order commensurate phases are hidden behind the vertical dashed lines .
the dashed curve is obtained in the continuum approximation ( [ x_kinkdensity ] ) .
( b ) classical excitation levels by adding ( red , falling ) or removing ( blue , rising ) a kink . for fixed @xmath6 ,
the lines mark the frequencies at which the spectral function exhibits a @xmath182-peak .
the upper ( lower ) dashed curve shows the minimal energy for adding ( removing ) a kink in the continuum approximation , cf .
( [ minenergy ] ) and ( [ e_annihilation ] ) . for excitations close to @xmath8 ,
see fig .
[ fig03 ] . ]
figure [ fig02](a ) also illustrates that close to @xmath8 , the dependence of the kink density on @xmath6 appears smooth so that we may try to effectively describe it using a continuum approximation .
physically such a description seems reasonable because close to @xmath8 , kinks are sparse such that the large distances between neighboring kinks `` wash out '' the discrete structure of the underlying lattice .
thus , close to @xmath8 , the ground state density @xmath168 of the classical model ( [ x_clhamiltonian ] ) is related to the external flux @xmath6 by @xmath183 } \big]\ .
\label{x_kinkdensity}\end{aligned}\ ] ] solving this relation for @xmath168 leads to the dashed curve in fig [ fig02](a ) . for @xmath184 , eq . ( [ x_kinkdensity ] )
would predict the relation @xmath185 , which in particular implies @xmath74 at half flux quantum , @xmath186 , corresponding to the staggered order . however , discreteness effects are visibly superposed on the continuum model , mostly due to large intervals in @xmath6 that hold the same kink density , cf . eq .
( [ x_width ] ) . very close to @xmath8 such that @xmath187\gg 1 $ ] , we explicitly find the dependence @xmath188 \right\}^{-1}\ , \label{x_kinkdensity_close}\end{aligned}\ ] ] which in particular shows that at @xmath85 the kink density grows with infinite slope .
the transitions between the various commensurate phases of the model ( [ x_clhamiltonian ] ) are first order.@xcite at zero temperature , quantum fluctuations due to the capacitative term in eq .
( [ x_lagrangian ] ) facilitate equilibration and prevent hysteresis effects if @xmath6 is varied sufficiently slowly in an experiment .
realistic capacitive energies @xmath189 lead to extended phases of incommensurate order , phases of `` floating primitive cells '' , around the transitions which we will discuss in sec .
[ sec : quantum ] .
in the regime of kink proliferation , the elementary excitations by photon absorption are the creation of an additional kink or the annihilation of a kink already in existence in the ground state .
the typical energy scale associated with such excitations is given by @xmath190 .
this scale could possibly already be below the plasmon gap @xmath191 if @xmath192 . at the critical flux @xmath8 , eq .
( [ x_phiec ] ) , when the first kinks nucleate in the ground state , the spectrum for kink excitations extends to zero energy . here , and in the vicinity of @xmath8 , these are therefore the relevant low - energy excitations . on either side of @xmath92 , the energies correspond to isolated low - frequency absorption peaks , see fig .
[ fig03 ] . in this section ,
we discuss the `` classical '' absorption spectrum , i.e. we neglect spontaneous creation / annihilation of kinks due to quantum phase slips and only allow for such processes in the context of photon absorption .
we discuss how quantum effects alter this picture in sec .
[ sec : quantum ] . in a setup with a single _
capacitively - coupled _ antenna at link @xmath43 [ fig .
[ fig_antennas](a ) ] , the system may be excited by _ locally _ introducing a kink at the link @xmath43 of the antenna . in a single absorption act associated with kink creation or annihilation at link
@xmath43 , the system responds by boosting the phase difference @xmath193 over the @xmath43-th josephson junction from @xmath194 to @xmath12 or vice versa , whereas phase differences at other links acquire negligible changes @xmath195 . for this reason
, we may estimate the matrix element in the absorption rate @xmath127 , eq .
( [ absorptionrate ] ) , in the limit of a single fluxonium qubit@xcite , where the energy difference @xmath196 between the excited and the ground state is determined by the external flux @xmath6 and the effective potential a kink at @xmath43 feels in the presence of kinks at other sites .
the matrix element entering eq .
( [ absorptionrate ] ) is then given by @xmath197 herein , the parameter @xmath198 with dimensions of energy , is the amplitude of a quantum phase slip.@xcite in determining matrix elements and absorption rates , we assume the typical limit @xmath199 and @xmath200 for microwave frequencies @xmath50 . as a result , we obtain for the absorption rate in the case of capacitive coupling @xmath201 with @xmath202 .
it features a single peak at frequency @xmath196 , which corresponds to the energy cost for adding or removing a kink from the ground state at link @xmath43 the spectrum of energies @xmath203 is determined in secs .
[ sec : addition ] and [ sec : annihilation ] , see also fig .
[ fig02](b ) .
the _ inductive coupling _ setup [ fig .
[ fig_antennas](b ) ] , as discussed in sec .
[ sec : inductive ] , couples the single antenna to links over the long range of @xmath62 . for realistic situations , we may assume that @xmath62 is much larger than the length @xmath9 , the relevant scale for the statics of kinks .
this allows the insertion of kinks at any link over a long range . if the antenna is situated at @xmath43 , the absorption rate @xmath128 , eq .
( [ absorptionrate_l ] ) , becomes @xmath204 where @xmath205 , as the matrix element in eq .
( [ absorptionrate_l ] ) is given by latexmath:[\ ] ] close to @xmath8 , and @xmath279 for @xmath280 .
if this inequality is not satisfied , especially close to @xmath8 , effects of the then strong effective hopping are more drastic and the single peak for kink annihilation fades away into the spectrum of a strongly - correlated quantum liquid , cf .
the next section .
away from @xmath8 , however , the peak should remain intact , owing to the exponential smallness of @xmath253 , eq .
( [ eq : x_phase_slip_amplitude ] ) . for fixed @xmath281 .
quantum phase slips , in this illustration with amplitude @xmath282 , melt all commensurate kink densities @xmath173 with @xmath283 , leading to extended incommensurate luttinger liquid ( ll ) phases , for which the spectral function has the form of eq .
( [ llspectrum ] ) . for general @xmath253 , eq .
( [ cic ] ) determines the maximum @xmath218 such that classical commensurate phases @xmath173 survive the quantum fluctuations . ]
quantum ising models beyond nearest neighbors , i.e. hamiltonians of the form of eq .
( [ x_hamiltonian ] ) , have recently @xcite been studied theoretically for power - law interactions , e.g. , in the context of rydberg atoms @xcite and experimentally using trapped ions.@xcite applying the theoretical results to our model ( [ x_hamiltonian ] ) , we predict the phase diagram as a function of external flux @xmath6 for fixed @xmath17 and @xmath253 to be the one shown in fig .
[ fig_quantum ] . below a critical flux @xmath8 ,
the system is , as in the classical limit , in a homogeneous phase with gapped excitation spectrum , cf .
[ sec : broadenedpeaks ] , but then undergoes a kosterlitz - thouless ( kt ) transition into a gapless phase of `` floating '' kinks .
this phase is a luttinger liquid and the excitation spectrum for adding or removing kinks is analogous to the spectrum of adding or removing spinless fermions in a one - dimensional system with strong repulsion . upon further increasing @xmath6 ,
the system undergoes a commensurate - incommensurate ( cic ) transition into a `` pinned '' phase featuring a rational kink density @xmath173 and a gapped excitation spectrum as in fig .
[ fig02](b ) .
quantum effects broaden the `` classical '' peaks according to eq .
( [ broadenedpeak ] ) .
after another cic transition , the system becomes again a luttinger liquid .
an ( even ) number of cic transitions ( depending on @xmath253 ) may follow before the system undergoes a second kt transition from a liquid into another gapped homogeneous phase , which in a final quantum ising ( qi ) transition turns into the staggered order corresponding to kink density @xmath74 .
let us have a closer look into the phases that feature a finite kink density .
for this purpose , let us consider a classical ground state with kink density @xmath173 and estimate its stability to quantum fluctuations .
whereas the adjacent ground state for increasing ( decreasing ) @xmath6 has only one extra ( fewer ) kink , practically all the kinks need to be rearranged in order to minimize the potential energy.@xcite .
we can interpret this state as the original state of period @xmath218 with the addition of @xmath218 defects , in analogy to the two domain walls introduced in an antiferromagnet by flipping one spin.@xcite for the kink density @xmath74 , this analogy is perfect .
the quantum @xmath284 term introduces effective hopping of the defects by virtual quasi - simultaneous adding and removing of kinks , similar to the single - kink hopping discussed in sec .
[ sec : broadenedpeaks ] .
the classical energies of the virtually occupied intermediate states are of order @xmath285 , cf .
[ fig02](b ) , giving a hopping amplitude @xmath286 . the kinetic energy gain from such hopping of @xmath218 defects , which is @xmath287 ,
then lowers the energy of the states with defects relative to the original period-@xmath218 ground state and , as a result , takes away a strip of width @xmath288 from both ends of the interval the period-@xmath218 ground state has classically occupied in @xmath6 . in fig .
[ fig_quantum ] , e.g. , the @xmath289 phase had to cede almost one half of its classical interval in @xmath6 , cf .
( [ x_width ] ) , to the strips to its left and its right . in these strips ,
because of the mobility of the kinks , the commensurate configuration has melted and been replaced by an incommensurate state of floating `` defective '' primitive cells .
this state is gapless and physically described by a luttinger liquid .
creation of a kink at a link @xmath33 is here analogous to injecting a spinless electron at a site @xmath290 into a ( strongly - coupled ) quantum wire.@xcite therefore , using the well - known luttinger liquid results , we find for absorption at low energies the spectral function @xmath291 within the incommensurate phase , the luttinger parameter @xmath292 varies@xcite as a function of @xmath6 . at the cic transition involving a commensurate kink density @xmath173
, it takes the value @xmath293 , cf . refs . .
if the width of the incommensurate strip exceeds the width of the interval of the classically commensurate state , i.e. if @xmath294 the classical state with period @xmath218 is entirely unstable and merges into an incommensurate phase of floating defects between two commensurate phases of shorter period @xmath218 .
the estimate ( [ cic ] ) shows explicitly that classical ground states of kink densities with smaller denominator @xmath218 are more immune to quantum fluctuations .
in fact , the classical phase @xmath175 , the `` most immune '' pinned phase , would be destroyed entirely only if @xmath253 were so large that it would also destabilize the luttinger liquid into a gapped homogeneous quantum paramagnet.@xcite based on typical fluxonium parameters @xcite , we predict that classical phases of pinned densities @xmath173 with @xmath295 survive also in the presence the quantum fluctuations and systems of size larger than @xmath296 would be needed to observe signatures of luttinger liquid physics such as an absorption spectrum according to eq .
( [ llspectrum ] ) . upon decreasing of the external flux to a critical value @xmath8
, the lowest-@xmath6 incommensurate phase destabilizes into a homogeneous phase in a kt transition@xcite .
we note that the value of the critical flux @xmath8 , because of quantum fluctuations , will be slightly larger than the classical value from eq .
( [ x_phiec ] ) .
close to this transition , theory @xcite predicts a critical @xmath297 .
as the external flux @xmath6 approaches half a flux quantum , the system undergoes a similar kt transition from the rightmost ( see fig .
[ fig_quantum ] ) incommensurate phase ( with the same @xmath298 ) into a homogeneous quantum paramagnet .
eventually , there is a quantum ising transition@xcite to the antiferromagnet - like @xmath299 phase in the vicinity of @xmath186 .
the model circuit , fig . [ fig01](a ) , we have discussed has a surprisingly rich equilibrium phase diagram , cf . figs . [ fig01](b ) and [ fig_quantum ] , despite its relatively simple structure .
this arises due to the combination of the nonlinear properties of josephson junctions and the long - range interactions introduced by the coupling to a common ground . as a function of the circuit parameters
@xmath300 , @xmath5 , and the external field @xmath6 , the model exhibits equilibrium phase transitions of the kosterlitz - thouless , commensurate - incommensurate , and ising classes .
we have shown that circuit qed realizations of this model enable access to low - energy excitations . in particular , they allow to identify the quantum phase transitions in linear response by the absorption of microwaves using a capacitively ( c ) or inductively ( l ) coupled antenna . here
we summarize our predictions for the absorption rate @xmath301 in the various characteristic regimes of parameters , particularly @xmath9 and @xmath6 . in this limit ,
the elementary excitations of the system are plasmons .
the absorption rates @xmath69 and @xmath70 are given by eq .
( [ fc_plasmons ] ) and ( [ fl_plasmons ] ) , respectively , with the density of states of plasmon excitations given by eq .
( [ dos_plasmons ] ) . at the critical external flux @xmath92 , classically given by eq .
( [ phiec_ising ] ) , the spectrum is gapless with a uniform density of states [ eq .
( [ critical_f ] ) ] . in the proximity of @xmath8 , where the ginzburg criterion ( [ ginzburg ] ) is violated and quantum fluctuations are strong , the gap in the plasmon spectrum grows linearly as a function of the distance to @xmath8 , [ eq . ( [ gap_linear ] ) ] , while outside the ginzburg region , the classical square root dependence on the external flux [ eqs .
( [ plasmon - gap ] ) and ( [ plasmon - gap - ordered ] ) ] sets in . this regime , with @xmath302 , is closer to the the parameters realized in fluxonium qubits.@xcite the ground state and excitation spectrum at large flux , @xmath76 with @xmath8 given by eq .
( [ x_phiec ] ) , are significantly different from the small - flux regime , @xmath78 , where the ground state is homogeneous .
the elementary excitations are associated with the addition or removal of localized kinks ( or vortices ) .
the absorption spectrum is different for the capacitive [ eq . ( [ fckink ] ) ] or inductive [ eq . ( [ fl ] ) ] coupling of the antenna , the former adding or removing kinks locally , the latter over an extended range . in the realistic quantum picture , the lowest excited states in the regime @xmath78 appear above a gap of order @xmath303 and form a very flat quantum band [ eq .
( [ broadenedpeak ] ) ] .
the band slightly broadens the classical peak in the absorption rate [ eq .
( [ spectrumleft ] ) and fig .
[ fig03](b.1 ) ] . as @xmath85 ,
the gap closes and , upon undergoing a kosterlitz - thouless ( kt ) transition at @xmath8 , the system enters the phase of a `` floating '' crystal of kinks . the excitation spectrum here at @xmath76 is of luttinger - liquid type and at low energies is given by eq .
( [ llspectrum ] ) . as @xmath6
is increased further , depending on the strength of quantum fluctuations ( phase slips ) , the system undergoes a commensurate - incommensurate ( cic ) transition , after which the ground state carries a pinned classical kink density . here
, the excitation spectrum is essentially the classical spectrum of figs .
[ fig02](b ) and [ fig03](b.2 ) , where quantum fluctuations slightly broaden the @xmath182-peaks according to eq .
( [ bandwidth_right ] ) . the classical spectral function for kink addition
is given by eq .
( [ rho_continuum_close ] ) , for @xmath6 not too far from @xmath8 , and in general by eq .
( [ rho_discrete ] ) . for kink annihilation
, the spectral function is displayed in eq .
( [ rho_annihilation ] ) . for kink densities @xmath173 with @xmath304 or @xmath305 , which dominate the phase diagram
, there is only a single peak in the absorption spectrum related to kink annihilation , which is the lowest excitation energy of the system , cf .
( [ e_annih ] ) and figs . [ fig02](b ) and [ fig03](b.2 ) . in our theoretical analysis , we built on the availability of a `` superinductance '' such as in fluxonium qubits@xcite when envisioning experimental realizations .
as a result , the type - ii regime of small @xmath3 could be reached without considering the effects of additional parasitic capacitances to ground that would be unavoidable in a realization using ordinary electromagnetic inductance .
such additional parasitic capacitances would suppress quantum fluctuations and thus enhance the classical behavior of the system . in the type - i regime
, it decreases the effective @xmath131 for the low energy plasmon modes , narrowing the quantum critical region for the ising transition . in the type - ii regime ,
capacitance to ground decreases the quantum phase slip rate @xmath253,@xcite which helps to stabilize the classical pinned phases . throughout
, we have used periodic boundary conditions for theoretical convenience .
realistically , it is easier to create an array with open boundary conditions and so for small system sizes there will be edge effects
. these will extend over a length @xmath306 , eq .
( [ x_ell ] ) , as this is the scale for interactions in the system . yet for large systems whose size significantly exceeds at least the length @xmath9 , our results should be directly applicable .
we have also assumed the temperature is zero .
both finite temperature and the finite size of the system prevent the formation of long - range order and in principle mean that there will not be a true phase transition .
however , the equilibrium state and spectrum will retain signatures of the infinite system behavior on short enough length and timescales .
we note that there is an intermediate regime @xmath307 , where the classical model has an incomplete staircase of first order transitions , cf .
[ fig01](b ) and refs . .
in this parameter region , the type - ii regime commensurate and incommensurate phases of finite kink density for @xmath76 turn into the single ising phase of the type - i regime . the classical theory@xcite predicts a sequence of `` superdegenerate''@xcite and multicritical points as the transition at @xmath8 changes from first order at large @xmath9 to second order in the type - i regime . the construction of an effective model for this region that also allows one to analytically study the effect of quantum fluctuations appears difficult .
however , the study of the ground states and excitations may be amenable to numerical techniques such as quantum monte carlo as the problem is bosonic . in the type - ii regime at large @xmath9 , the low - energy behavior is well described in terms of localized kinks . using the locally - coupled capacitative antenna configuration ,
kinks can be selectively introduced or removed on individual sites .
if multiple kinks are introduced , they will interact over the large length scale @xmath9 according the hamiltonian ( [ x_clhamiltonian ] ) .
this introduces the possibility of investigating the many - body non - equilibrium physics of the system . in the presence of dissipation
, it may be possible to create a model system with a driven - dissipative steady - state of interacting kinks , an interesting addition to the set of non - equilbrium many body models that can be simulated with circuit qed systems.@xcite we would like to thank g. rastelli , i. pop , m. h. devoret , j. keeling , yu chen , c. neill , s. hacohen - gourgy , and v. ramasesh for interesting and helpful discussions .
this work was supported by nsf dmr1301798 and dmr1206612 , and aro w911nf1410011 . h. m. and r. t. b. acknowledge the yale prize postdoctoral fellowship .
99 o. daldini , p. martinoli , j. l. olsen , and g. berner , _ vortex - line pinning by thickness modulation of superconducting films _ , phys .
. lett . * 32 * , 218 ( 1974 ) ; p. martinoli , o. daldini , c. leemann , and b. van den brandt , _ josephson oscillation of a moving vortex lattice _ , _ ibid .
_ * 36*,382 ( 1976 ) ; o. daldini , c. leemann , and p. martinoli , _ coherent flux flow in superconducting films with periodically modulated thickness _ , solid state comm .
* 16 * , 509 ( 1975 ) ; p. martinoli , o. daldini , c. leemann , and e. stocker , _ a.c .
quantum interference in superconducting films with periodically modulated thickness _ , _ ibid .
_ * 17 * , 205 ( 1975 ) .
v. l. pokrovsky and a. l. talapov , _ ground state , spectrum , and phase diagram of two - dimensional incommensurate crystals _ ,
* 42 * , 65 ( 1979 ) and _ the theory of two - dimensional incommensurate crystals _ ,
jetp * 51 * , 134 ( 1980 ) .
d. r. nelson and v. m. vinokur , _ boson localization and pinning by correlated disorder in high - temperature superconductors _ , phys .
lett . * 68 * , 2398 ( 1992 ) and _ boson localization and correlated pinning of superconducting vortex arrays _
b * 48 * , 13060 ( 1993 ) .
a. van oudenaarden and j. e. mooij , _ one - dimensional mott insulator formed by quantum vortices in josephson junction arrays _ ,
* 76 * , 4947 ( 1996 ) ; a. van oudenaarden , b. van leeuwen , m. p. m. robbens , and j. e. mooij , _ one - dimensional mott localization of quantum vortices in josephson - junction arrays _ , phys
b * 57 * , 11684 ( 1998 ) .
d. b. haviland , k. andersson , and p. gren , _ superconducting and insulating behavior in one - dimensional josephson junction arrays _ , j. low temp
. phys . * 118 * , 733749 , ( 2000 ) ; k. andersson and d. b. haviland , _ escape from a zero - current state in a one - dimensional array of josephson junctions _ , phys .
b * 67 * , 092507 ( 2003 ) .
a. blais , r. s. huang , a. wallraff , s. m. girvin , and r. j. schoelkopf , _ cavity quantum electrodynamics for superconducting electrical circuits : an architecture for quantum computation _ ,
a * 69 * , 062320 ( 2004 ) .
n. a. masluk , i. m. pop , a. kamal , z. k. minev , and m. h. devoret , _ microwave characterization of josephson junction arrays : implementing a low loss superinductance _ ,
lett . * 109 * , 137002 ( 2012 ) ; m. t. bell , i. a. sadovskyy , l. b. ioffe , a. yu . kitaev , and m. e. gershenson _ quantum superinductor with tunable nonlinearity _ , _ ibid .
_ * 109 * , 137003 ( 2012 ) .
m. marchand , k. hood , and a. caill , _ nonconvex interactions and the occurrence of modulated phases _ , phys .
lett . * 58 * , 1660 ( 1987 ) and _ nonconvex interactions : a mechanism for the occurrence of modulated order in condensed matter _ , phys .
b * 37 * , 1898 ( 1988 ) . in order to avoid any confusion , we point out that although our interaction @xmath157 , eq . ( [ x_j ] ) , falls exponentially and so is certainly short - ranged in the sense of scaling , it is nevertheless non - zero at an arbitrarily large distance @xmath311 .
this implies the theoretical completeness of the devil s staircase in our model , even though the `` stairs '' for kink densities @xmath173 with large @xmath218 occupy exponentially small intervals in @xmath6 , cf .
( [ x_width ] ) .
p. richerme , c. senko , s. korenblit , j. smith , a. lee , r. islam , w. c. campbell , and c. monroe , _ quantum catalysis of magnetic phase transitions in a quantum simulator _ , phys .
* 111 * , 100506 ( 2013 ) . | we evaluate the microwave admittance of a one - dimensional chain of fluxonium qubits coupled by shared inductors . despite its simplicity
, this system exhibits a rich phase diagram .
a critical applied magnetic flux separates a homogeneous ground state from a phase with a ground state exhibiting inhomogeneous persistent currents . depending on the parameters of the array
, the phase transition may be a conventional continuous one , or of a commensurate - incommensurate nature .
furthermore , quantum fluctuations affect the transition and possibly lead to the presence of gapless `` floating phases '' .
the signatures of the soft modes accompanying the transitions appear as a characteristic frequency dependence of the dissipative part of admittance . |
[ introduction ] in this paper @xmath12 denotes the deligne - mumford compactification of the moduli space of smooth curves over @xmath13 of genus @xmath14 .
the boundary locus @xmath15 of @xmath12 consists of nodal curves with finite automorphism groups , which together with the smooth curves are referred to as _
stable _ curves . the locus of hyperelliptic curves will be denoted by @xmath16 , and the closure of the locus of trigonal curves will be denoted by @xmath17
. the main objects of our study will be families of genus @xmath11 stable curves , whose general members are smooth .
associated to any such _ flat and proper _
family @xmath18 are three basic invariants @xmath19 , @xmath20 and @xmath21 .
we define these in section [ definition ] as divisors on @xmath6 , but for most purposes one can think of them as integers by considering their respective degrees . the invariant @xmath20 counts , with appropriate multiplicities , the number of singular fibers of @xmath7 .
the self - intersection of the relative dualizing sheaf @xmath22 on @xmath7 defines @xmath21 , and its pushforward to @xmath6 is a rank @xmath11 locally free sheaf , whose determinant is @xmath19 . the basic relation @xmath23 and the positivity of the three invariants for non - isotrivial families force the _ slope _ @xmath24 of @xmath25 to fall into the interval @xmath26 ( cf .
[ slope - non - isotrivial ] ) .
in fact , cornalba - harris and xiao establish for this slope an exact upper bound of @xmath27 , which is achieved only for certain hyperelliptic families ( cf . theorem [ chx ] ) .
however , if the base curve @xmath6 passes through a _
general _ point of @xmath5 , mumford - harris - eisenbud give the better bound of @xmath28 ( cf .
theorem [ generalbound1 ] ) .
the families violating this inequality are entirely contained in the closure @xmath29 of the locus of @xmath30-sheeted covers of @xmath31 , for a suitably chosen @xmath30 .
in particular , for @xmath32 we recover the hyperelliptic locus @xmath16 , for @xmath33 - the trigonal locus @xmath17 , etc .
therefore , higher than the above `` generic '' ratio can be obtained only for families with special linear series , such as @xmath34 , @xmath35 , etc .
these observations clearly raise the following * question . * _ according to the possession of special linear series , is there a stratification of @xmath12 which would give successively smaller slopes @xmath36 ? what would be the successive upper bounds with respect to such a stratification ? _ the following result , whose proof will be given in the paper , answers this question for an exact upper bound for families with linear series @xmath35 .
* theorem i. * _ if @xmath37 is a trigonal nonisotrivial family with smooth general member , then the slope of @xmath25 satisfies : @xmath38 equality is achieved if and only if all fibers are irreducible , @xmath7 is a triple cover of a ruled surface @xmath39 over @xmath6 , and a certain divisor class @xmath40 on @xmath7 is numerically zero . _ to understand the importance of this result and the above question , consider mumford s alternative description of the basic invariants ( cf .
[ linebundles ] ) : @xmath19 , @xmath20 and @xmath21 are restrictions of certain rational divisor classes @xmath41 . specifically , @xmath42}$ ] , where @xmath43 the class of the boundary divisor @xmath44 of @xmath5 , and @xmath45 is freely generated by the hodge class @xmath4 and the boundary classes @xmath43 for @xmath46 ( cf .
thus , our question about a stratification of @xmath5 translates into a question about the relations among the fundamental classes of various subvarieties defined by geometric conditions in @xmath5 .
moreover , such a stratification would provide a link between the _ global _ invariant @xmath4 ( the degree of the hodge bundle on @xmath12 ) and the _ locally defined _
invariant @xmath3 of the singularities of our families . in the process of estimating the ratio @xmath47 we hope to understand the geometry of interesting loci in @xmath12 , and describe their rational picard groups
such a program for the hyperelliptic locus @xmath16 is completed by cornalba - harris ( cf .
theorem [ theoremchpic ] ) , who exhibit generators and relations for @xmath48 .
the typical examples of families with maximal ratio of @xmath49 are constructed as blow - ups of pencils of hyperelliptic curves , embedded in the same ruled surface .
similar examples for trigonal families yield the slope @xmath50 , but as theorem i shows , this ratio is _ not _ an upper bound .
this happens because of an `` extra '' _ maroni _ locus in @xmath17 ( cf .
while a general trigonal curve embeds in @xmath51 or in the blow - up @xmath52 of @xmath53 at a point , the remaining trigonal curves embed in other rational ruled surfaces and comprise a closed subset in @xmath17 , called the maroni locus .
the proof of theorem ii , stated below , implies that all trigonal families achieving the maximal bound lie entirely in the maroni locus , and moreover , their members are embedded in ruled surfaces `` as far as possible from the generic '' ruled surfaces @xmath54 and @xmath52 . the ratio @xmath50 , though not the `` correct '' maximum , plays a significant role in understanding the geometry of the trigonal locus , and in describing its rational picard group . in particular , in a linear relation
is established between the hodge class , the boundary classes on @xmath17 , and a canonically defined vector bundle @xmath8 of rank 2 on a ruled surface @xmath55 ( cf .
sect . 9 ) : * theorem ii .
* _ let @xmath56 denote the boundary class in @xmath57 corresponding to irreducible singular curves , and let @xmath58 be the remaining boundary classes .
for any trigonal non - isotrivial family with general smooth member , we have @xmath59 where @xmath60 is a quadratic polynomial in @xmath61 with linear coefficients in @xmath11 , and it is determined by the geometry of @xmath58 .
_ for example , @xmath62 corresponds to the boundary divisor @xmath63 , whose general member is the join in three points of two trigonal curves of genera @xmath61 and @xmath64 , respectively ( cf . fig .
[ delta - k , i ] ) .
recall that the vector bundle @xmath8 is called _ bogomolov semistable _ if its chern classes satisfy @xmath65 ( cf .
we show in section 9 the following * theorem iii . * _ for any trigonal nonisotrivial family @xmath1 with general smooth member , if @xmath8 is bogomolov semistable , then @xmath66 _ in the even genus case , the maroni locus is in fact a divisor on @xmath17 , whose class we denote by @xmath67 .
we further recognize the `` bogomolov quantity '' @xmath68 as counting roughly four times the number of maroni fibers in @xmath7 , and deduce * theorem iv . * _ for even @xmath11 , @xmath69 is freely generated by all boundary divisors @xmath56 and @xmath58 , and the maroni divisor @xmath67 .
the class of the hodge bundle on @xmath17 is expressed in terms of these generators as the following linear combination : @xmath70 _ consequently , the condition @xmath71 in theorem i can be interpreted as a relation among the number of irreducible singular curves and the `` maroni '' fibers in our family : @xmath72 , and hence maximal slope families are entirely contained in the maroni locus of @xmath17 ( cf . theorem [ maximalmaroni ] ) .
the stated theorems complete the program for the trigonal locus @xmath17 , which was outlined earlier in this section . an important interpretation of these results can be traced back to @xcite , where it is shown that the moduli space @xmath73 is of _ general type_. the @xmath30-gonal locus @xmath29 is realized in terms of the generating classes as : @xmath74=a\lambda - b\delta-{\mathcal}{e}$ ] for some @xmath75 , and an effective boundary combination @xmath76 .
restricting to a general curve @xmath77 , we have @xmath78 , and hence @xmath79 .
because of seshadri s criterion for ampleness of line bundles , in effect , we are asking for all positive numbers @xmath80 and @xmath81 such that the linear combination @xmath82 is ample on @xmath5 . the smaller the ratio @xmath83 is , the stronger result we obtain .
in other words , we are aiming at a maximal bound of @xmath84 , when we think of these classes as restricted to any curve @xmath85 . in view of this ,
part of this paper can be described as looking for all _ ample _ divisors on @xmath17 of the form @xmath82 with @xmath75 .
i then gives the necessary condition @xmath86 ( compare with @xcite ) .
some of the results can be applied to the study of the discriminant loci of a certain type of triple covers of surfaces .
the methods and ideas for the trigonal case are in principal extendable to more general families of @xmath30-gonal curves .
we refer the reader to sect .
13 for a general maximal bound for tetragonal curves ( for @xmath11 odd ) , and conjectures for the maximal and general bounds for any @xmath87-gonal and other families of stable curves
. acknowledgments this paper is based on my ph.d .
thesis at harvard university .
joe harris , my advisor , introduced me to the problem of finding a stratification of the moduli space with respect to a descending sequence of slopes of one parameter families .
i am very grateful to him for his advice and support throughout my work on the present thesis .
i would like to thank fedor bogomolov , david eisenbud , benedict gross , brendan hassett , david mumford , tony pantev and emma previato for the helpful discussions that i have had with them at different stages of the project , as well as kazuhiro konno for providing me with his recent results on trigonal families .
a source of inspiration and endless moral support has been my husband , edward frenkel , to whom goes my gratitude and love .
[ preliminaries ] let @xmath89 be a flat proper one - parameter family of stable curves of genus @xmath11 , where @xmath6 is a smooth projective curve .
assume in addition that the general member of @xmath7 is _ smooth _ ( cf .
[ family ] ) .
@xmath90 let @xmath91 be the relative dualizing sheaf of @xmath92 .
its pushforward @xmath93 is a locally free sheaf on @xmath6 of rank @xmath11 , and we set @xmath94 to be its determinant . the sheaf @xmath93 is known as the `` hodge bundle '' on @xmath6 , and @xmath19 - as the `` hodge class '' of @xmath6 . in a similar way , we set @xmath21 to be the self - intersection of @xmath22 : @xmath95 the definition of @xmath20 , on the other hand , is local and requires some notation .
let @xmath96 be any singular point of a fiber @xmath97 , @xmath98 .
since the general fiber of @xmath7 is smooth , the total space of @xmath7 near @xmath96 is given locally analytically by @xmath99 , where @xmath100 and @xmath101 are local parameters on @xmath97 , @xmath102 is a local parameter on @xmath6 near @xmath81 , and @xmath103 .
( this follows from the one - dimensional versal deformation space of the nodal singularity at @xmath96 . ) for any other point @xmath96 of @xmath7 we set @xmath104 .
now we can define @xmath105 by abuse of notation , we shall use the same letters for the line bundles @xmath106 and @xmath20 and for their respective degrees , e.g. @xmath107 . * remark 2.1 . *
it is possible to define the three basic invariants for a wider variety of families . in particular , dropping the assumption of smoothness of the general fiber roughly means that the base curve @xmath6 is contained entirely in the boundary locus of @xmath12 . since
such families are not discussed in our paper , we shall not give here these definitions .
the existence , however , of such invariants for any one - parameter family of stable curves will follow from the description of @xmath108 and @xmath109 as `` global '' classes in @xmath110 ( cf .
[ linebundles ] ) . * remark 2.2 . *
it is also possible to consider families whose special members are not stable , e.g. cuspidal , tacnodal and other types of singular curves .
one reduces to the above cases by applying _
semistable reduction _
@xcite ) .
another way to interpret the classes @xmath88 and @xmath21 is to think of them as rational divisor classes on @xmath5 .
in fact , mumford shows that such invariants , defined for _ any _ proper flat family @xmath112 and natural under base change , induce line bundles in pic@xmath111 .
here follows a rough sketch of the argument ( cf .
@xcite ) .
consider @xmath113 , the hilbert scheme parametrizing all closed subschemes of @xmath114 with hilbert polynomial @xmath115 for some @xmath116 and @xmath117 .
let @xmath118 be the locally closed smooth subscheme of @xmath119-canonical stable curves of genus @xmath11 .
then @xmath12 is the git - quotient of @xmath120 by @xmath121 .
let @xmath122 be the natural surjection , and let @xmath123 be the subgroup of isomorphism classes of line bundles on @xmath120 invariant under the action of @xmath121 .
consider also @xmath124 , the group of line bundles on the _ moduli functor_. an element @xmath125 of @xmath124 consists of the following data : for any proper flat family @xmath126 of stable curves a line bundle @xmath127 on @xmath128 natural under base change .
two such elements are declared isomorphic under the obvious compatibility conditions .
naturally , a line bundle on @xmath12 gives rise by pull - back to a line bundle on the moduli functor .
in fact , this map is an inclusion with a torsion cokernel , and @xmath124 is torsion free and isomorphic to @xmath123 : @xmath129 hence we may regard all these groups as sublattices of @xmath110 .
in particular , @xmath130 and any line bundle on the moduli functor can be thought of as a rational class on @xmath12 .
these identifications allow us to make the following * definition 2.1 . * in @xmath131
we define the line bundles @xmath132 and @xmath3 by @xmath133 where @xmath134 is the universal curve over @xmath120 , @xmath135 is the projection map , @xmath136 is the relative dualizing sheaf of @xmath137 , and @xmath138 is the divisor of singular curves on @xmath120 . as defined , @xmath139 and @xmath3 lie in @xmath110 , and as such they are only _ rational _ cartier divisors on @xmath12 . in @xcite one can find examples where @xmath4 does _ not _ descend to a line bundle on @xmath12 . on the other hand , it is obvious from which divisor on @xmath12 our @xmath3 comes : @xmath140 , where @xmath15 denotes the divisor on @xmath12 comprised of all singular stable curves . again , due to singularities of the total space of @xmath141 , @xmath15 is only a _ rational _ cartier divisor . in fact , the only locus of @xmath5 on which @xmath108 and @xmath109 are necessarily _ integer _ divisor classes is @xmath142 - the locus of automorphism - free curves .
we can further define the _ boundary _ classes @xmath143}$ ] in @xmath144 .
let @xmath44 be the @xmath145cartier divisor on @xmath146 whose general member is an irreducible nodal curve with a single node ( if @xmath147 ) , or the join of two irreducible smooth curves of genera @xmath61 and @xmath148 intersecting transversally in one point ( if @xmath149 ) . setting @xmath150 , we have @xmath151 and @xmath152 . as the following result of harer @xcite suggests that , in order to describe the geometry of the moduli space @xmath5 , it will be useful to study the divisor classes defined above , and to understand the relations between them .
the hodge class @xmath4 and the boundary classes @xmath153}$ ] generate @xmath131 , and for @xmath46 they are linearly independent .
it is easy to recognize the restrictions of @xmath108 and @xmath109 to a curve @xmath6 in @xmath5 as the previously defined @xmath88 and @xmath21 .
for example , the restriction of @xmath3 to the base curve @xmath6 counts , with appropriate multiplicities , the number of intersections of @xmath6 with the boundary components @xmath44 of @xmath5 . as a final remark , applying grothendieck riemann - roch theorem ( grr ) to the map @xmath135 and the sheaf @xmath136 , implies the basic relation : @xmath154 let @xmath89 be our family of stable curves with a smooth general member . by definition , @xmath155 .
further , all locally free quotients of the hodge bundle @xmath93 have non - negative degrees @xcite . if @xmath7 is a non - isotrivial family , then @xmath156 , and since the relative canonical divisor @xmath157 is nef , @xmath158 @xcite . in particular
, we can divide by @xmath19 . * definition 2.2 . * the _ slope _ of a non - isotrivial family @xmath89 of stable curves with a smooth general member is the ratio @xmath159 suppose we make a base change @xmath160 of degree @xmath87 , and set @xmath161 to be the pull - back of our family over the new base @xmath162 ( cf . fig .
[ basechange ] ) .
then the three invariants on @xmath6 will pull - back to the corresponding invariants on @xmath162 , and their degrees will be multipied by @xmath87 , e.g. @xmath163 , etc . in particular
, the slope of @xmath164 will be preserved .
( 1.8,1.8)(-0.2,3.9 ) ( 0,4)@xmath165 ( 0,5.1)@xmath166 ( 0.2,5)(1.25,0)2(0,-1)0.6 in view of ( [ grr ] ) , restricting to the base curve @xmath6 yields @xmath167 from the positivity conditions above , we deduce that @xmath168 it is natural to ask whether we can find a better estimate for the slope of @xmath7 .
the first fundamental result in this direction is the following let @xmath89 be a nonisotrivial family with smooth general member .
then the slope of the family satisfies : @xmath169 equality holds if and only if the general fiber of @xmath92 is hyperelliptic , and all singular fibers are irreducible .
[ chx ] note that the upper bound is achieved only for hyperelliptic families .
such families are of very special type since the hyperelliptic locus @xmath170 has codimension @xmath171 in @xmath12 .
on the other hand , if the base curve @xmath6 is general enough , a much better estimate can be shown ( cf .
@xcite ) : if @xmath6 passes through a general point @xmath172\in \overline{\mathfrak{m}}_g$ ] , then @xmath173 [ generalbound1 ] for example , when @xmath11 is odd , we can set @xmath174 and define the divisor @xmath29 in @xmath175 as the closure of the @xmath30-sheeted covers of @xmath31 : @xmath176 @xmath177 if our family is not entirely contained in @xmath29 , or equivalently , if @xmath6 passes through a point @xmath172\not\in\overline{{\mathcal}{d}}_k$ ] ( cf .
[ generalcurve ] ) , @xmath178 higher than the `` generic '' ratio can be obtained , therefore , only for a very special type of families : those entirely contained in @xmath29 , and hence possessing @xmath179 , @xmath180 , etc . in proving
the maximal bound @xmath27 , cornalba - harris also describe @xmath181 by exhibiting generators and relations ( cf .
@xcite ) . here
we briefly discuss their result .
recall the irreducible divisors @xmath44 on @xmath12 . for @xmath182 $ ]
, @xmath44 cuts out an irreducible divisor on @xmath170 , while the intersection @xmath183 breaks up into several components : @xmath184}.\ ] ] set @xmath185 for the class of @xmath186 in @xmath170 , and retain the symbols @xmath4 and @xmath43 for their corresponding restrictions to @xmath181 .
thus , @xmath187 for all @xmath61 .
finally note that the class @xmath56 is realised in @xmath181 as the sum @xmath188}.\ ] ] the coefficient @xmath189 roughly means that @xmath190 is _
double _ along @xmath186 , for @xmath149 .
the classes @xmath191}$ ] and @xmath192}$ ] freely generate @xmath193 .
the hodge class @xmath194 is expressed in terms of these generators as the following linear combination : @xmath195}2(i+1)(g - i)\xi_i + \sum_{j=1}^{[g/2]}4j(g - j)\delta_j . \label{chpic}\ ] ] [ theoremchpic ] for a specific family @xmath89 of hyperelliptic stable curves this relation reads : @xmath196}2(i+1)(g - i)\xi_{i}|_b + \sum_{j=1}^{[g/2]}4j(g - j)\delta_{j}|_b\ ] ] @xmath197 this yields the desired @xmath27 inequality for the slope of a hyperelliptic family , and shows that the maximum can be obtained exactly when all @xmath198},\delta_1,\cdots,\delta_{[\frac{g}{2}]}$ ] vanish on @xmath6 .
in other words , the singular fibers of @xmath7 belong only to the boundary divisor @xmath199 , and hence are irreducible . in appendix
we review the description of the divisors @xmath186 via admissible covers , and give an alternative proof of theorem [ theoremchpic ] .
we present here a typical example in which the upper bound @xmath27 is achieved , and show how to calculate explicitly the basic invariants @xmath19 and @xmath20 for this family .
* example 2.1 . * consider a pencil @xmath200 of hyperelliptic curves of genus @xmath11 on @xmath201 .
because of genus considerations
, its members must be of type @xmath202 . our family @xmath203 will be obtained by blowing - up @xmath204 at the @xmath205 base points of the pencil in order to separate its members ( cf .
[ ratio8 + 4/g ] ) . hence , @xmath206 for the corresponding topological euler characteristics .
riemann - hurwitz formula for the map @xmath92 gives a second relation : @xmath207 , where @xmath31 is the base @xmath6 and @xmath97 is the general fiber of @xmath7 .
putting together , @xmath208 ( 1,3)(0,3.1 ) ( -0.7,5.5)@xmath209 ( -0.5,5.4)(1,-1)1 ( 1.7,5.4)(-1,-1)1 ( 0,3.9)@xmath210 ( 0.4,2.7)@xmath31 ( 0.6,3.8)(0,-1)0.6 the total space of @xmath7 is a divisor on @xmath211 of type @xmath212 , and the map @xmath203 is the restriction to @xmath7 of the third projection @xmath213 . using standard methods , we compute @xmath214 . from the positivity of all free quotients of the hodge bundle on @xmath31 ,
@xmath93 splits as a direct sum @xmath215 for some @xmath216 .
then , for @xmath217 to have no sections , all @xmath218 s must be at most @xmath219 .
finally , @xmath220 in a similar vein as in the above example , we consider pencils of trigonal curves on ruled surfaces , and obtain the slope @xmath221 .
it is somewhat reasonable to expect that this is the maximal ratio .
recall that a bundle @xmath76 on a curve @xmath6 is _ semistable _ if for any proper subbundle @xmath222 , we have @xmath223 , where @xmath96 is the quotient of the degree and the rank of the corresponding bundle . following xiao s approach in the proof of theorem [ chx ] , konno shows that for non - hyperelliptic fibrations of genus @xmath11 with semistable hodge bundle @xmath224 ( cf .
@xcite ) : @xmath225 as for any trigonal families , he establishes the inequality ( cf .
@xcite ) : @xmath226 examples of trigonal families achieving this ratio were not found , which suggested that this bound might be too big . on
the other hand , in trying to disprove the smaller bound @xmath221 , we naturally arrived at counterexamples pointing to a third intermediate ratio ( cf .
theorem [ maximal bound2 ] ) : @xmath227 the difference between the last two estimates may seem negligible , but this would not be so when both @xmath19 and @xmath20 become large and we attempt to bound @xmath19 from below by @xmath20 . what is more important , the second ratio is in fact _ exact _ , and we give equivalent conditions for it to be achieved ( cf . sect .
[ whenmaximal ] , [ maroni - maximal ] ) .
this maximal bound confirms chen s result for genus @xmath228 in @xcite .
the reader may ask why the `` generic '' examples for the maximum in the hyperelliptic case fail to provide also the maximum in the trigonal case . as we noted in the introduction , the answer is closely related to the so - called _ maroni _ locus in @xmath17 . more precisely ,
if @xmath229 denotes the corresponding rational ruled surface , a general curve @xmath230 embeds in @xmath54 is @xmath11 is even , and in @xmath52 if @xmath11 is odd .
the maroni locus consists of those curves that embed in @xmath231 with @xmath232 .
the number @xmath233 is referred to as the _ maroni invariant _ of @xmath230 . in these terms , the examples of pencils of trigonal curves on @xmath54 and @xmath52 have the lowest possible constant maroni invariant , and we shall see that the maximum bound can be obtained only for families entirely contained in the maroni locus , and having very high maroni invariants .
the `` semistable '' bound @xmath221 appears in theorem [ 7 + 6/g bogomolov2 ] , where we give instead a sufficient _ bogomolov - semistability _ condition @xmath234 for a canonically associated to @xmath7 vector bundle @xmath8 of rank 2 .
the rational picard group of @xmath17 is described in terms of generators and relations in section [ generators ] , providing thus in the trigonal case an analog of theorem [ theoremchpic ] .
note the apparent similarity of the coefficients @xmath60 of the trigonal boundary classes and the coefficients of the hyperelliptic boundary classes .
this is not coincidental .
in fact , the @xmath60 s are in a sense the `` smallest '' coefficients that could have been associated to the corresponding classes @xmath58 ( cf . fig .
[ delta - k , i ] ) : they are symmetric with respect to the two genera of the components in the general member of @xmath58 . a crucial role in the proof of theorem [ pic trigonal ] is played by the interpretation of the above bogomolov semistability condition in terms of the maroni locus it @xmath17 ( cf .
[ interpretation ] ) .
let @xmath89 be a family of stable curves , whose general member @xmath97 is a smooth trigonal curve .
by definition , @xmath97 is a triple cover of @xmath31 .
we would like to study how this triple cover varies as @xmath97 moves in the family @xmath7 .
thus , it would be desirable to represent @xmath7 , by analogy with @xmath97 , as a triple cover of a ruled surface @xmath39 , comprised by the image lines @xmath31 . unfortunately , due to existence of hyperelliptic and other special singular fibers , this is not always possible .
( 3,4.9)(-1,2 ) ( 0,4)@xmath235 ( 0,5.1)@xmath236 ( 0.2,3.85)(-1.3,-0.7)2(1,-1)0.5 ( 1.4,3.85)(-1,-1)0.5 ( 0.5,3)(-0.5,0.85)2(-2,-1)0.8 ( 0.6,2.9)@xmath237 ( -1.3,3.3)@xmath7 ( -0.7,2.3)@xmath6 ( 0.1,5)(1.5,0)2(0,-1)0.6 ( 1.6,6.2)(0,-1)0.6 ( 1.35,6.3)@xmath238 ( 0.2,5.6)(2,1)1.2 the `` closest '' model of such a triple cover can be obtained after a finite number of birational transformations on @xmath7 , and a possible base change over the base @xmath6 .
this way we construct a _ quasi - admissible _ cover @xmath239 over a new base @xmath240 ( cf .
[ propquasi ] ) . here
@xmath241 is a _ birationally _ ruled surface over @xmath237 with reduced , but non necessarily irreducible , special fibers : @xmath242 allows for _ pointed stable _ rational fibers , i.e. trees of @xmath31 s with points marked in a certain ( stable ) way .
the map @xmath243 expresses any fiber @xmath244 as a triple quasi - admissible cover of the corresponding _ pointed stable _ rational curve @xmath245 . to calculate effectively our invariants @xmath246 and @xmath109
, we need that @xmath243 be _ flat _ , which could force a few additional blow - ups on @xmath247 and @xmath242 .
we end up with a flat proper triple cover @xmath248 , where certain fibers of @xmath249 and @xmath55 are allowed to be _ non - reduced _ : these are the scheme - theoretic preimages under the blow - ups on @xmath247 and @xmath242 .
we call such covers @xmath250 _ effective_. we observe next that any smooth trigonal curve @xmath230 can be naturally embedded in a ruled surface @xmath231 over @xmath6 . if @xmath251 is the corresponding triple cover , there is an exact sequence of locally free sheaves on @xmath252 : @xmath253 the projectivization @xmath254 of the rank 2 vector bundle @xmath8 is the ruled surface @xmath231 .
this construction can be extended as @xmath230 moves in the effective cover @xmath255 .
the flatness of @xmath250 forces the pushforward @xmath256 to be a locally free sheaf of rank 3 on @xmath55 , and the finiteness of @xmath250 ensures the existence of a _ trace map _ @xmath257 .
again , the kernel @xmath8 of @xmath258 is the desired rank 2 vector bundle on @xmath55 , in whose projectivization , @xmath238 , we embed @xmath249 ( cf . fig .
[ basic construction idea ] ) .
we can now use the relations in the chow rings of @xmath259 , @xmath260 and @xmath261 to calculate the invariants @xmath262 and @xmath263 , appropriately defined for the new family @xmath264 of semistable and occasionally non - reduced fibers . then , of course , we translate @xmath262 and @xmath263 into @xmath265 and @xmath266 with the necessary adjustments from the birational transformations on @xmath7 and the base change on @xmath6 .
we compare the resulting expressions to obtain a relation among @xmath267 and @xmath268 .
as we vary the base curve @xmath6 inside @xmath17 , we actually obtain a relation among the restrictions of @xmath4 and @xmath3 in @xmath69 , rather than just among @xmath269 and @xmath270 in @xmath271 .
have we thus represented and linked @xmath272 and @xmath273 ? to answer this question ,
we need first to understand the boundary divisors of the trigonal locus @xmath17 . as we shall see , there are seven types of such divisors , denoted by @xmath274 and @xmath275 for @xmath276 .
each type is determined by the specific geometry of its general member .
for example , @xmath277 is the closure of all irreducible trigonal curves with one node , while @xmath278 corresponds to joins in two points of a trigonal and a hyperelliptic curve with genera @xmath61 and @xmath279 , respectively ( cf . fig .
[ delta - k , i ] ) .
naturally , we derive an expression for the restriction of the divisor class @xmath280 to @xmath17 : @xmath281}}3\delta_{1,i } + \sum_{i=1}^{\scriptscriptstyle{g-2}}2\delta_{2,i } + \sum_{i=1}^{\scriptscriptstyle{[g/2]}}\delta_{3,i } + \sum_{i=1}^{\scriptscriptstyle{[(g-1)/2]}}3\delta_{4,i}+ \sum_{i=1}^{\scriptscriptstyle{g-1}}\delta_{5,i } + \sum_{i=1}^{\scriptscriptstyle{[g/2]}}\delta_{6,i}. \label{delta}\ ] ] here @xmath56 and @xmath58 are the divisor classes of @xmath277 and @xmath275 in @xmath69 . for a fixed family @xmath1 with a smooth trigonal general member , we establish a relation among the hodge class @xmath19 , the boundary classes @xmath282 , and the bogomolov quantity @xmath68 for the associated vector bundle @xmath8 : @xmath283 the polynomial coefficients @xmath60 are comparatively larger than the corresponding coefficients of the boundary divisors in the expression for @xmath284 . as a result ,
we rewrite ( [ tobelifted ] ) as @xmath285 where @xmath76 is an effective combination of the boundary classes on @xmath17 .
in particular , if @xmath8 is bogomolov semistable , the slope satisfies ( cf .
theorem [ 7 + 6/g bogomolov2 ] ) : @xmath286 further , we describe @xmath69 as generated freely by the restriction @xmath272 and the boundary classes of @xmath17 . in the even genus @xmath11 case , we can replace @xmath272 by a geometrically defined class @xmath67 , corresponding to the so - called maroni divisor in @xmath17 .
this , of course , means that the hodge class @xmath272 must be some linear combination of the boundary classes and @xmath67 .
the bogomolov quantity is interpreted as @xmath287 which in turn `` lifts '' ( [ tobelifted ] ) to the wanted relation in @xmath69 : @xmath288 we have not yet computed explicitly all coefficients @xmath289 . in the cases which we have completed ( @xmath290 and @xmath291 ) , these coefficients turn out again sufficiently large so that we can repeat the argument in ( [ e - argument ] ) .
thus , if @xmath7 has at least one non - maroni fiber , and its singular fibers belong to @xmath292 , then @xmath293 , and hence the stronger bound of ( [ idea7 + 6/g ] ) holds ( cf .
[ maroni inequality ] and conj .
[ maroni - conj ] ) .
since the bogomolov semistability condition @xmath294 is not always satisfied , the above discussion shows that @xmath221 is _ not _ the maximal bound for the slope of trigonal families , therefore , we need another , more subtle , estimate .
the expressions for @xmath19 and @xmath20 suggest that any maximal bound would be equivalent to an inequality involving @xmath295 , @xmath296 , and possibly some other invariants .
we construct a specific divisor class @xmath40 on @xmath247 , for which the _ hodge index _
theorem implies @xmath297 , and we translate this into @xmath298 ( cf . prop .
[ genindex ] ) .
we notice that the only reasonable way to replace bogomolov s condition @xmath294 by the newly found inequality is by subtracting the following quantities : @xmath299 so that the `` left - over '' linear combination of boundary divisors @xmath300 is again effective ( cf .
theorem [ maximal relation2 ] ) .
hence , we conclude that for _ all _ trigonal families : @xmath301 the presentation of the _ basic construction _ is done in several stages .
[ stages ] shows schematically the connection between the three types of covers , admissible , quasi - admissible and effective , in relation to the original family @xmath1 of stable curves .
@xmath302 we start in section [ hurwitz ] by introducing a compactification @xmath303 of the hurwitz scheme , parametrizing _ admissible _ @xmath87-uple covers of stable pointed rational curves . using its coarse moduli properties ,
we show in section [ admissible ] the existence of admissible covers of surfaces @xmath304 associated to the original family @xmath305 .
next we modify these covers to _ quasi - admissible _ covers @xmath306 ( cf .
[ propquasi ] ) , and further to _ effective _ covers @xmath307 in order to resolve the technical difficulties arising from the non - flatness of @xmath243 ( cf .
[ effectivecovers ] ) .
we devote section 4 to the study of the boundary components of the trigonal locus @xmath17 inside the moduli space @xmath5 , and express the restriction @xmath308 as a linear combination of the boundary divisors ( cf .
[ divisorrel ] ) . in section 6
we complete the basic construction by embedding the effective cover @xmath249 in a rank 1 projective bundle @xmath238 over @xmath55 . for convenience of the reader , the proofs of the maximal @xmath309 and the semistable @xmath221 bounds
are presented first in the special , but fundamental case when the original family @xmath305 is already an effective triple cover of a ruled surface @xmath39 ( cf .
sect . 7 ) .
the discussion results in finding the coefficients of @xmath56 in two different expressions of @xmath272 , but , as it turns out , the knowledge of these coefficients is enough to determine the desired two bounds .
we refer to this as the _ global _ calculation .
the hodge index theorem and nakai - moishezon criterion on @xmath7 complete the global calculation in sect .
[ indextheorem ] . a discussion of maximal bound examples can be found in section [ whenmaximal ] .
the _ local _ calculations in sections 8 - 10 compute the contributions of the other boundary classes @xmath58 , and express @xmath272 in terms of these contributions and the chern classes of the rank 2 vector bundle @xmath8 on @xmath55 . for clearer exposition ,
the proofs of the two bounds are shown first for a _ general _ base curve @xmath6 ( i.e. @xmath6 intersects transversally the boundary components in general points ) , and then in section 11 the results are extended to _ any _ base curve @xmath6 .
we develop the necessary notation and techniques for the local calculations in section [ conventions ] .
section 12 discusses the relation between the bogomolov semistability condition and the maroni locus , and describes the structure of @xmath69 . in section [ maroni - maximal ]
we give another interpretation of the conditions for the maximal bound .
we present further results and conjectures for @xmath87-gonal families in section 13 . in the appendix
, we give another proof of the @xmath27 bound in the hyperelliptic case and show an application of the maximal trigonal bound to the study of the discriminant locus of certain triple covers .
[ quasi - admissible ] we first review briefly the theory of admissible covers . for more details , we refer the reader to @xcite .
let @xmath311 be the _ small hurwitz scheme _ parametrizing the pairs @xmath312 , where @xmath230 is a smooth curve of genus @xmath11 and @xmath313 is a cover of degree @xmath87 , simply branched over @xmath314 distinct points .
since @xmath315 , there is a natural map @xmath316 , whose image contains an open dense subset of @xmath317 .
the theory of admissible covers provides the commutative diagram in fig .
[ hurwitz figure ] .
( 5,3.5)(-0.8,2.2 ) ( 0,4)@xmath318 ( 0.4,3.85)(1,-1)0.9 ( 1.9,3.85)(1,-1)0.9 ( 0.4,4.2)(1,1)0.9 ( 1.9,4.2)(1,1)0.9 ( 2.4,4.5)@xmath319 ( 2.4,3.5)@xmath320 ( 1.1,2.5)@xmath321 ( 1.1,5.2)@xmath322 there @xmath323 ( resp .
@xmath324 ) is the moduli space of @xmath325-pointed @xmath31 s ( resp . of stable @xmath325-pointed rational curves ) , and @xmath310 is a compactification of the hurwirz scheme .
the points of @xmath310 correspond to triples @xmath326 , where @xmath230 is a connected reduced nodal curve of genus @xmath11 , @xmath327 is a stable @xmath325-pointed rational curve , and @xmath328 is a so - called _
admissible cover_. * definition 3.1 . * given the curves @xmath230 and @xmath329 as above , an _ admissible cover _
@xmath328 is a regular map satisfying the following conditions : ( a1 ) @xmath330 and @xmath331 is simply branched over the distinct points @xmath332 ; ( a2 ) for every @xmath333 lying over a node @xmath334 , the two branches through @xmath96 map with the same ramification index to the two branches through @xmath335 .
@xmath336 note that @xmath230 is not necessarily a stable curve , but contracting its destabilizing rational chains yields the corresponding stable curve @xmath337 .
in such a case , we say that @xmath230 is the `` admissible model '' for @xmath338 ( cf .
[ nonstable ] ) .
harris - mumford have shown that the compactification @xmath310 is in fact a _ coarse moduli space _ for the admissible covers @xmath328 .
@xmath339 when we vary the admissible covers of curves in families , the local structure of the corresponding total spaces becomes apparent .
let @xmath340 be a proper flat family ( over a scheme @xmath341 ) of admissible covers of curves ( cf .
[ admfamilies ] ) .
assume that @xmath342 is tale everywhere except over the nodes of the fibers of @xmath343 , and except over some sections @xmath344 and their images @xmath345 : there @xmath342 is simply branched along @xmath346 over @xmath347 for all @xmath61 .
if @xmath348 is a point lying above a node @xmath349 for some @xmath350 , then @xmath351 has a node at @xmath96 , and locally analytically we can describe @xmath352 and @xmath342 near @xmath96 and @xmath335 by : @xmath353 one can see that @xmath119 is the index of ramification of @xmath342 at @xmath96 , and that fiberwise @xmath354 is an admissible cover ( of curves ) . from now on , by _
admissible covers _ we mean , more generally , families @xmath355 over @xmath341 with the above description . the local properties of the admissible cover @xmath340 over the nodes in @xmath356 forces singularities on the total spaces of @xmath357 and @xmath358 .
since we will be interested only in the cases when the base @xmath341 is a smooth projective curve @xmath6 and the general fiber of @xmath357 is smooth , we can always pick a generator @xmath102 for @xmath359 , and express @xmath360 for some @xmath361 .
@xmath362 * example 3.1 .
* let the triple admissible cover @xmath363 contain the fiber @xmath351 as in fig .
[ singular ] . at @xmath96 , @xmath364 is given by @xmath365 , and at @xmath335 , @xmath358 is given by @xmath366 , where @xmath367 .
this forces at @xmath368 the local equation @xmath369 ( @xmath370 ) .
even if @xmath357 is smooth at @xmath96 ( @xmath371 ) , @xmath357 and @xmath358 will be singular at @xmath368 and @xmath335 , respectively ( @xmath372 ) .
compare this with the non - flat cover of ramification index 1 in fig .
[ mult4 ] .
recall that a rational double point @xmath373 on a surface @xmath128 is of type @xmath374 if locally analytically @xmath128 is given at @xmath373 by the equation @xmath375 .
thus , @xmath368 and @xmath335 above are rational double points on @xmath357 and @xmath358 , respectively , of type @xmath374 .
* remark 3.1 . * in the sequel , we use the fact that the projection @xmath376 is a _ finite _ map . from the weak valuative criterion for properness , this means that given a family of admissible covers @xmath377 over the punctured disc @xmath378 , there is some @xmath379 for which @xmath342 extends to a family @xmath380 of admissible covers over @xmath381 $ ] . in particular ,
if the base for the admissible cover @xmath382 is an open set @xmath383 of a smooth projective curve @xmath6 , modulo a finite base change , we can extend this to a family of admissible covers @xmath304 over the whole curve @xmath6 .
consider a family @xmath89 of stable curves of genus @xmath11 , whose general member is smooth and @xmath87-gonal .
let @xmath384 be the canonical map , and let @xmath385 denote the fiber product @xmath386 .
( 5,4)(-0.7,1.9 ) ( -2,4)@xmath387 ( -0.8,3.85)(1.4,0)2(0,-1)0.9 ( -1,2.5)@xmath388 ( -2.5,3.3)@xmath7 ( 0.7,3.35)@xmath389 ( -3.5,4.8)@xmath390 ( -3.2,4.7)(1.5,-0.8)2(2,-3)0.7 ( -2.1,3.4)(-1,1.5)2(3,-2)1.1 ( 3,3.7)@xmath391 ( 4.2,5.05)(1.4,-1.5)2(0,-1)0.9 ( 3.2,5.05)(0,-1)0.9 ( 5.4,2.2)@xmath392 ( 5.7,3.1)@xmath389 ( 3,5.2)@xmath393 ( 4.3,3.6)(1,-1)1 ( 4.8,3.1)@xmath394 if the general member of @xmath7 has infinitely many @xmath395 s , the variety @xmath385 will have dimension @xmath396 .
we can resolve this by considering an intersection of the appropriate number of hyperplane sections of @xmath385 , and picking a one - dimensional component @xmath397 dominating @xmath6 .
the curve @xmath397 might be singular , but by normalizing it and pulling @xmath7 over it , we get another family of stable curves ( cf . fig .
[ map eta ] ) : @xmath398 since the two families have the same basic invariants , we can replace the original with the new one , and assume the existence of a map @xmath399 compatible with @xmath400 .
in other words , @xmath40 associates to every fiber @xmath230 of @xmath7 a specific @xmath401 on @xmath230 or , possibly , a @xmath401 on an admissible model @xmath402 of @xmath230 .
let @xmath383 be the open subset of @xmath6 over which _ all _ fibers are smooth and @xmath87-gonal
. for simplicity , assume for now that all the fibers over @xmath383 can be represented as admissible covers of @xmath31 via the chosen @xmath401 s , i.e. they are _ simply branched _ covers of @xmath31 over @xmath325 distinct points of @xmath31 .
denote by @xmath403 the restriction of @xmath7 over @xmath383 ( cf .
fig.12 ) .
the map @xmath404 induces a section @xmath405 where @xmath406 is the _ relative degree @xmath87 picard variety _ of @xmath403 over @xmath383 .
@xmath406 parametrizes the line bundles on @xmath403 of relative degree @xmath87 .
the image @xmath407 is a class of line bundles on @xmath403 whose fiberwise restrictions are the chosen @xmath401 s .
let @xmath408 be a representative of this class , and let @xmath409 be the ruled surface @xmath410 over @xmath383 .
the map @xmath382 induced by @xmath408 defines an admissible cover over @xmath383 , as shown in fig .
[ construction y * ] . (
5,2.2)(-0.3,2.6 ) ( 0,4)@xmath411 ( 0.2,3.85)(1,-1)0.5 ( 1.4,3.85)(-1,-1)0.5 ( 0.6,2.9)@xmath383 ( 0.05,3.4)@xmath92 ( 1.3,3.4)@xmath412 from remark 3.1 , @xmath342 extends to a family of admissible covers @xmath413 over the whole base @xmath6 .
since @xmath414 and @xmath7 are isomorphic over @xmath383 , they are birational to each other .
in other words , the fibers @xmath230 of @xmath7 , over which @xmath408 does not extend to the base - point free linear series @xmath415 , are modified by blow - ups and blow - downs so as to arrive at their admissible models in @xmath414 .
we have thus proved the following let @xmath89 be a family of stable curves , whose general member over an open subset @xmath416 is a smooth @xmath87-uple admissible cover of @xmath31 .
then , modulo a finite base change , there exists an admissible cover of surfaces @xmath417 over @xmath6 such that @xmath414 is obtained from @xmath7 by a finite number of birational transformations performed on the fibers over @xmath418 .
[ quasicov ] in case the general member of @xmath7 is _ not _ an admissible cover of @xmath31 , e.g. it is trigonal with a total point of ramification , we have to modify the above construction .
to start with , we can not expect to obtain an _ admissible _ cover @xmath419 , even modulo a finite base change .
this leads us to consider a different kind of covers , which we call _ quasi - admissible_. * definition 3.2 . * a _ quasi - admissible cover _ @xmath420 of a nodal curve @xmath230 over a semistable pointed rational curve @xmath329 is a regular map which behaves like an admissible cover over the singular locus of @xmath329 , i.e. for any @xmath421 lying over a node @xmath334 the two branches through @xmath96 map with the same ramification index to the two branches through @xmath335 .
@xmath422 quasi - admissible covers differ from admissible covers in allowing more diverse behavior of @xmath230 over @xmath423 , e.g. having singularities , higher ramification points and multiple simple ramification points .
[ quasicovers ] displays several degree 3 quasi - admissible covers over @xmath31 : however , any quasi - admissible cover can be obtained from an admissible cover @xmath424 by simultaneous contractions of components in @xmath425 and their ( rational ) inverses on @xmath402 . * definition 3.3 . * a _ minimal _ quasi - admissible cover @xmath426 is minimal with respect to the number of components of @xmath329 . in other words
, one can not apply more simultaneous contractions on @xmath427 and end with another quasi - admissible cover .
* example 3.2 . *
a smooth trigonal curve @xmath230 with a total point of ramification @xmath96 is a minimal quasi - admissible cover of @xmath428 . blowing up @xmath96 on @xmath230 and @xmath429 ,
gives an admissible cover @xmath430 , where @xmath431 maps three - to - one onto @xmath432 with a total point of ramification @xmath433 ( cf .
[ quasi / adm ] ) .
@xmath434 the motivation for using _ minimal _ quasi - admissible covers , instead of just admissible or quasi - admissible covers , is that the former are the closest covers to the original families @xmath1 of stable curves , and calculations on them will yield the best possible estimate for the ratio @xmath435 ( cf .
[ stages ] ) .
now let us consider the remaining case of a family @xmath1 , whose general member over @xmath383 is smooth and @xmath87-gonal , but _ not _ an admissible cover of @xmath31 .
after a possible base change , we still have the map ( cf .
12 ) @xmath436 it associates to every fiber @xmath230 a @xmath401 on its admissible model @xmath402 .
let @xmath437 be the corresponding admissible cover .
since @xmath230 itself is @xmath87-gonal , and by assumption it does not possess a @xmath438 with @xmath439 , @xmath230 must be a @xmath87-uple cover of some component of @xmath425 .
in particular , the @xmath401 on @xmath402 restricts to a @xmath401 on @xmath230 .
thus , in effect , @xmath40 gives again a section @xmath440 .
as before , we obtain a degree @xmath87 finite map @xmath382 to the ruled surface @xmath441 over @xmath383 .
note that this is a family of _ minimal quasi - admissible _ covers .
we extend @xmath342 over the curve @xmath6 as follows .
for simplicity , assume that @xmath442 .
let @xmath443 be the ramification divisor of @xmath342 in @xmath403 . by hypothesis
, there is a component @xmath444 of @xmath443 which passes through total ramification points and dominates @xmath383 .
letting @xmath445 be the closure of @xmath444 in @xmath7 , we can normalize it and pull the family @xmath7 over it .
so we may assume that @xmath445 is a section of @xmath446 .
if there are some other components @xmath447 of the ramification divisor @xmath443 passing through higher ramification points , we repeat the same procedure for them , until we `` straighten out '' all @xmath448 s into sections of @xmath1 .
let @xmath449 be the corresponding sections of @xmath409 over @xmath383 .
we can shrink @xmath383 in order to exclude any fibers with isolated higher ramification points . consider a fiber @xmath230 in @xmath403 .
let @xmath450 be its total ramification points , and let @xmath451 be their images on @xmath452 in @xmath409 .
it is clear that blowing - up all @xmath453 s and @xmath454 s will give an admissible triple cover @xmath455 .
the @xmath401 , giving this cover , is the original one assigned by @xmath456 .
we globalize this construction by blowing - up the sections @xmath457 on @xmath403 and @xmath458 on @xmath409 .
similarly as above , we obtain a triple admissible cover of surfaces @xmath459 over @xmath383 .
the properness of @xmath460 allows us to extend this to an admissible cover @xmath461 over @xmath6 ( cf .
[ blowing up ] ) .
( 3,5)(4.3,-0.5 ) ( 2,3.7)@xmath462 ( 2,2.2)@xmath463 ( 2,0.7)@xmath464 ( 2.2,3.5)(6.1,0)2(0,-1)0.9 ( 3.9,3.5)(2.8,0)2(0,-1)0.9 ( 2.2,2)(6.1,0)2(0,-1)0.9 ( 3.9,2)(2.8,0)2(0,-1)0.9 ( 3,0)(6.7,-2.3 ) ( 7.8,2.2)@xmath465 ( 8,2)(1.4,0)2(0,-1)0.9 ( 8.1,1.5)(1.4,0)2@xmath466 ( 7.8,0.7)@xmath467 ( 8,0.5)(1,-1)0.5 ( 9.4,0.5)(-1,-1)0.5 ( 8.5,-0.4)@xmath383 [ over b * ] denote by @xmath468 the component of @xmath469 , obtained by blowing up @xmath470 , and let @xmath471 be its closure in @xmath472 .
define similarly @xmath473 and @xmath474 .
the admissible cover @xmath475 maps @xmath476 to @xmath477 , so that after removing all the @xmath476 s and @xmath477 s we still have a triple cover @xmath478 note that @xmath479 and @xmath480 over the open set @xmath383 , and that @xmath481 is a birationally ruled surface over @xmath6 ( cf .
finally , note that from the quasi - admissible cover @xmath482 we obtain a family @xmath483 of _ minimal _ quasi - admissible covers : simply contract the unnecessary rational components in the fibers of @xmath484 and @xmath481 , and observe that the triple map @xmath485 restricts to the corresponding triple map @xmath243 .
this completes the construction of minimal quasi - admissible covers for any family @xmath1 with general smooth trigonal member .
the cases @xmath486 are only notationally more difficult .
one has to keep track of the possibly different higher multiplicities in @xmath230 and multiple double points in @xmath230 over the same @xmath334 .
the construction of an admissible cover @xmath417 goes through with minimal modifications .
we combine the results of this section in the following let @xmath89 be a family of stable curves , whose general member over an open subset @xmath416 is smooth and @xmath87-gonal .
then , modulo a finite base change , there exists a minimal quasi - admissible cover of surfaces @xmath487 over @xmath6 such that @xmath247 is obtained from @xmath7 by a finite number of birational transformations performed on the fibers over @xmath418 .
[ propquasi ]
[ boundarycomponents ] in this section we shall see that there are _ seven types _ of boundary divisors of @xmath17 , each denoted by @xmath275 for @xmath489 . the second index @xmath61 is determined in the following way .
let @xmath490 be the general member of @xmath275 , where @xmath491 and @xmath492 are smooth curves .
if @xmath491 and @xmath492 are both trigonal or both hyperelliptic , then we set @xmath61 to be the smaller of the two genera @xmath493 or @xmath494 . if , say , @xmath491 is a trigonal , but @xmath492 is hyperelliptic , then we set @xmath61 to be genus of the trigonal component @xmath491 .
the only exception to this rule occurs when @xmath230 is irreducible ( and hence of genus @xmath11 with exactly one node ) .
we denote this boundary component by @xmath277 . when we view a general member @xmath230 roughly as a triple cover of @xmath31 s in the hurwitz scheme ( consider the pull - back @xmath495\in\overline{{\mathcal}{h}}_{3,g}$ ] ) , then it may or may not be ramified .
if there is no ramification , then @xmath230 lies in one of the first four types of trigonal boundary divisors @xmath275 , @xmath496 .
ramification index 1 characterizes the general members of @xmath497 and @xmath498 , and in case of @xmath499 the ramification index is 2 ( cf .
[ delta - k , i ] ) .
there is an alternative description of the boundary components @xmath275 s of @xmath17 .
if one such @xmath275 lies in the restriction @xmath500 of the divisor @xmath190 in @xmath5 , then @xmath275 is one of @xmath501 or @xmath497 .
the partial normalization of their general members @xmath230 is still connected , i.e. @xmath230 is either irreducible , or the join of two smooth curves meeting in at least two points .
correspondingly , for the general member @xmath230 of the remaining three types of boundary components , @xmath502 and @xmath499 , the irreducible components of @xmath230 intersect transversely in exactly one point , so that the normalization of @xmath230 is disconnected .
@xmath503 ( 6,1)(2.7,-1.4 ) ( -0.7,2.45)@xmath504 ( 2.8,1.2)@xmath505\hspace{17 mm } \scriptstyle{i=1,2 , ... , g-2}\hspace{17mm}\scriptstyle{i=1,2 , ... , } \left[\frac{g}{2}\right]$ ] ( -0.2,0.5)@xmath506 ( 01.1,-0.6)@xmath507\hspace{24 mm } i=1,2, ... ,g-1 \hspace{22mm}i=1,2, ... ,\left[\frac{g}{2}\right]}$ ] the boundary divisors of @xmath17 can be grouped in seven types : @xmath274 and @xmath275 for @xmath276 .
their general members and range of index @xmath61 are shown in fig .
[ delta - k , i ] .
the boundary of @xmath17 consists of @xmath274 , @xmath275 , and the codimension 2 component @xmath16 of hyperelliptic curves .
[ boundary ] consider the projection map @xmath508 , whose image is the trigonal locus @xmath17 .
thus , the inverse image of each boundary divisor @xmath275 will be a boundary divisor @xmath509 in @xmath510 .
the converse , however , is not always true , i.e. certain boundary divisors of @xmath510 contract under @xmath511 to smaller subschemes of @xmath17 , e.g. the hyperelliptic locus @xmath16 .
with the description of the hurwitz scheme @xmath510 , given in section 3 , it is easier to determine first @xmath510 s boundary divisors .
thus , we postpone the proof of proposition [ boundary ] until the end of the next subsection .
the boundary divisors of @xmath510 can be grouped in six types : @xmath509 for @xmath276 .
their general members and range of index @xmath61 are shown in fig .
[ admissible - k , i ] . @xmath512 ( 5,1)(3,-2 ) ( -0.9,1.85)@xmath513 ( 0.3,0.7)@xmath505\hspace{29 mm } \scriptstyle{i=1 , ... , g-1}\hspace{24mm}\scriptstyle{i=0,1 , ... , } \left[\frac{g}{2}\right]$ ] ( -0.9,0.15)@xmath514 ( 0.3,-1.2)@xmath507\hspace{29 mm } i=1,2, ... ,g-1 \hspace{23mm}i=1,2, ... ,\left[\frac{g}{2}\right]}$ ] [ boundary2 ] a general member @xmath515 of the boundary @xmath516 is a triple admissible cover of a chain of _ two _ @xmath31 .
( from the dimension calculations that follow it will become clear that an admissible cover of a chain of three or more @xmath517 s will generate a subscheme in @xmath510 of codimension @xmath396 . )
note that _ three _ connected components of @xmath515 over one @xmath31 means that they are all smooth @xmath31 s themselves , and hence they can all be contracted simultaneously , leaving us with a smooth trigonal curve , or with a hyperelliptic curve with an attached @xmath518 , neither of which cases by dimension count corresponds to a _
general _ member of a boundary component @xmath509 . considering all combinations of one or two connected components of @xmath515 over each @xmath31
, we generate a list of the possible general members of the boundary divisors @xmath509 .
to see which of these are indeed of codimension 1 in @xmath510 , we do the following calculation .
first we note that , for a fixed set of @xmath519 ramification points in @xmath31 , there are finitely many covers of degree @xmath520 and genus @xmath61 , that is , @xmath521 substracting @xmath520 takes into account the projectively equivalent triples of points on @xmath31 .
in particular , @xmath522 . a similar agrument ( with @xmath523 ramification points ) shows that for the hyperelliptic locus : @xmath524 these computations are valid for @xmath149 , whereas @xmath525 .
thus , to compute the dimensions of the six types of subschemes of @xmath510 , one adds the corresponding dimensions of @xmath526 and @xmath527 , making the necessary adjustments for the choice of intersection points on the components of each curve @xmath515 .
for example , when @xmath149 the dimension of the subscheme with general member @xmath515 , shown in fig .
[ admissible - k , i ] , is @xmath528 the final 1 s account for the choice of triples of points in the @xmath35 s on each component .
we conclude that for @xmath529 $ ] the join at three points of two trigonal curves , one of genus @xmath61 and the other of genus @xmath64 , is the general member of a boundary component of @xmath510 .
we denote it by @xmath530 .
the range of @xmath61 stops at @xmath531 $ ] for symmetry considerations .
when @xmath147 , the corresponding subscheme has a smaller dimension of @xmath532 and hence no boundary divisor is generated by such curves .
as another example , consider the fifth sketch in fig .
[ admissible - k , i ] . it corresponds to the join at one point of a trigonal curve @xmath491 of genus @xmath61 , a hyperelliptic curve @xmath492 of genus @xmath148 , and an attached @xmath31 to @xmath492 to make the whole curve a triple cover .
note that @xmath491 and @xmath492 intersect transversally at a point @xmath96 , but when presented as covers of @xmath31 they both have ramifications at @xmath96 of index 1 . on all such curves
@xmath491 and @xmath492 the total number of ramification points over @xmath31 is finite , and hence their choice does not affect the dimension of our subscheme .
thus , @xmath533 therefore , this subscheme is in fact a divisor in @xmath510 , which we denote by @xmath534 .
the cases of @xmath147 or @xmath535 lead to contractions of unstable rational components ( @xmath491 or @xmath492 ) , and do not yield the necessary dimension of @xmath536 .
hence , @xmath537 . in the case of @xmath538 , the two components @xmath491 and @xmath492 meet transversally in one point @xmath96 , but both have ramification of index @xmath189 at @xmath96 as triple covers of @xmath31 .
smooth trigonal curves of genus @xmath61 with such high ramification form a codimension 1 subscheme of the trigonal locus @xmath539 , hence the dimension of @xmath538 is @xmath540 thus , @xmath538 is a boundary divisor in @xmath510 for @xmath541 $ ] .
the case of @xmath147 yields dimension @xmath542 , and hence we disregard it . the remaining cases are treated similarly .
we conclude that @xmath510 has six types of boundary divisors , @xmath509 , whose general members and range of indices are indicated in fig .
[ admissible - k , i ] .
having described the boundary of @xmath510 , it remains to check which of the divisors @xmath543 preserve their dimension under the map @xmath511 and hence map into divisors of @xmath57 .
the only `` surprises '' can be expected where @xmath511 contracts unstable @xmath31 , such as in @xmath544 , @xmath545 , and @xmath546 .
in fact , only @xmath547 and @xmath548 diverge from the common pattern ; in all other cases , we set @xmath549 to be the corresponding boundary divisor in @xmath17 . the map @xmath511 contracts the three rational components of the general member of @xmath548 , leaving only a smooth hyperelliptic curve of genus @xmath11 .
thus , the image @xmath550 is the hyperelliptic locus @xmath16 , which is of dimension @xmath542 .
hence @xmath548 does not yield a divisor in @xmath17 , but a boundary component of codimension 2 .
finally we consider @xmath547 .
after we contract its two rational components , we arrive at an _ irreducible nodal _ trigonal curve with exactly one node .
the dimension of the subscheme of such curves is @xmath551 where the final @xmath219 indicates the choice of a triple of points on a smooth trigonal curve ( belonging to the @xmath35 ) , two of which will be identified as a node .
correspondingly , we obtain another divisor in @xmath17 , which we denote by @xmath552 by abuse of notation , we will denote by @xmath56 and @xmath58 the classes in @xmath69 of @xmath277 and @xmath275 , respectively .
the divisor class @xmath554 restricts to @xmath17 as the following linear combination of the boundary classes in @xmath17 : @xmath281}}3\delta_{1,i } + \sum_{i=1}^{\scriptscriptstyle{g-2}}2\delta_{2,i } + \sum_{i=1}^{\scriptscriptstyle{[g/2]}}\delta_{3,i } + \sum_{i=1}^{\scriptscriptstyle{[(g-1)/2]}}3\delta_{4,i}+ \sum_{i=1}^{\scriptscriptstyle{g-1}}\delta_{5,i } + \sum_{i=1}^{\scriptscriptstyle{[g/2]}}\delta_{6,i}. \label{divisorrel}\ ] ] _ proof .
_ let us rewrite equation ( [ divisorrel ] ) in the form @xmath555 and call @xmath556 the _ multiplicity _ of @xmath58 in @xmath553 .
this linear relation simply counts the contribution of each singular curve of a specific boundary type in @xmath557 to the degree of @xmath3 .
recall that for any trigonal family @xmath89 : @xmath558 here @xmath559 denotes the local analytic multiplicity of the total space of @xmath7 nearby @xmath96 measured by the equation @xmath99 , where @xmath100 and @xmath101 are local parameters on the singular fiber @xmath97 , and @xmath102 is a local parameter on @xmath6 near @xmath560 .
for each boundary class @xmath275 of @xmath17 , we consider its general member @xmath561 , and a base curve @xmath6 in @xmath17 which intersects transversally @xmath275 in @xmath172 $ ] . in the corresponding one - parameter trigonal family @xmath89 , we must find the sum of the multiplicities @xmath559 of the singularities of @xmath230 .
thus , @xmath562 for most of the divisors classes , this sum is actually quite straight forward .
for example , the general member @xmath172\in \delta{\mathfrak{t}}_{3,i}$ ] is the join of two smooth hyperelliptic curves @xmath491 and @xmath492 , which intersect transversally in one point @xmath96 . the family @xmath7 , constructed as above , will be given locally analytically nearby @xmath96 by @xmath563 , and hence @xmath564 .
a similar situation occurs in the cases of @xmath565 and @xmath566 : there is one point of transversal intersection ( or one node ) forcing @xmath567 in the cases of @xmath568 and @xmath569 there are correspondingly two or three points of transversal intersection , forcing @xmath570 this can be also interpreted by the fact that @xmath568 and @xmath569 lie entirely in the divisor @xmath190 in @xmath5 with , @xmath190 being _
double _ along @xmath568 and _ triple _ along @xmath569 .
a slightly more complex situation occurs in the case of @xmath571 .
the general member @xmath230 consists of two curves @xmath491 and @xmath492 , meeting transversally in two points @xmath96 and @xmath368 ( see fig . [ mult4 ] ) .
but , as in an admissible triple cover of two @xmath31 s , the points @xmath96 and @xmath368 behave differently : at one of them , say @xmath368 , the triple cover is _ not _ ramified , while at @xmath96 there is ramification of index @xmath219 . in the local analytic rings of @xmath572 and
@xmath368 the generators of @xmath573 map into the squares of the generators of @xmath574 : @xmath575 , and of course , @xmath576 , so that the local equation of @xmath39 near @xmath335 is @xmath577 , and that of @xmath7 near @xmath96 is @xmath563 .
but since the triple cover is a local isomorphism of @xmath573 into @xmath578 , the total space of @xmath7 near @xmath368 is given locally analytically by @xmath579 ( @xmath580 ) .
therefore , @xmath581 , but @xmath582 , and @xmath583 @xmath584 although the relations proved in this paper will be valid on the picard group @xmath585 , it will be interesting to check what happens with the hyperelliptic curves inside the trigonal locus @xmath17 .
we noted that @xmath16 is the only boundary component of @xmath17 of codimension 2 .
it is obtained as the image @xmath586 . by blowing up a point on a smooth hyperelliptic curve @xmath230
, we add a @xmath517component to @xmath230 to make it a triple cover @xmath587 ( cf .
[ smoothhyper ] ) .
it terms of the quasi - admissible covers , such @xmath587 behaves exactly as an irreducible singular trigonal curve in @xmath277 .
however , @xmath230 does not contribute to the invariant @xmath20 , as defined in section [ definition ] .
in fact , in a certain sense , it even decreases @xmath20 . to simplify the exposition
, we shall postpone the discussion of families with hyperelliptic fibers until section 11 , where we will explain the behavior of trigonal families with finitely many hyperelliptic fibers in terms of the exceptional divisor @xmath588 of the projection @xmath511 .
a similar phenomenon occurs with the boundary component @xmath589 , but it does not make sense to exclude its members from our discussion , since they behave exactly as members of the boundary divisor @xmath569 for @xmath590 . in the transition from the original family @xmath1 to the minimal quasi - admissible family @xmath487 over @xmath237 , certain changes occur in the calculation of the basic invariants . to start with ,
it is easy to redefine @xmath592 and @xmath593 for @xmath594 : simply use the corresponding definitions from section [ definition ] .
since we are interested in the slope of the family , which is invariant under base change , we may assume that @xmath595 and that @xmath7 is the pull - back over the new base @xmath237 .
now the difference between @xmath7 and @xmath247 is reduced to several `` quasi - admissible '' blow - ups on @xmath7 .
blowing up smooth or rational double points on a surface does not affect its structure sheaf .
therefore , the degrees of the hodge bundles on the two surfaces @xmath7 and @xmath247 will be the same : @xmath596 .
on the other hand , blowing up a smooth point on a surface decreases the square of its dualizing sheaf by 1 , while there is no effect when blowing up a rational double point .
each type of singular fibers @xmath230 in @xmath7 requires apriori different quasi - admissible modifications ( or no modifications at all ) , and thus decreases @xmath597 by some nonnegative integer , denoted by @xmath591 : @xmath598 thus , @xmath591 counts the number of `` smooth blow - ups '' on @xmath230 , which are needed to obtain the minimal quasi - admissible cover @xmath599 within the surface quasi - admissible cover @xmath600 . in the following lemma
, we compute the invariants @xmath591 only for the general members of the boundary @xmath488 ( cf . fig .
[ delta - k , i ] ) .
the remaining , more special , singular curves in @xmath488 will be linear combinations of these @xmath591 s ( cf . sect .
11 ) . if @xmath601 denotes the invariant @xmath591 for a general curve @xmath602 , then @xmath603 [ mu(c ) ] _ proof .
_ the general members of the boundary @xmath604 are in fact the minimal quasi - admissible covers associated to the general members of the boundary @xmath605 , except for @xmath190 which has @xmath606 .
thus , we trace the blow - ups necessary to transform the curves in fig .
[ delta - k , i ] to the curves in fig .
[ admissible - k , i ] .
for example , no blow - ups are needed in the case of @xmath607 , so that @xmath608 , while we need 2 blow - ups in the case of @xmath609 , and hence @xmath610 .
the only interesting situation occurs for @xmath611 .
apparently , there is only _
one _ added component @xmath517 to the original @xmath612 , but the lemma states that @xmath610 .
the difference comes from the fact that near the intersection @xmath613 the surface @xmath247 has equation @xmath614 , i.e. @xmath368 is a rational double point on @xmath247 of type @xmath615 ( a similar situation occurred in fig .
[ mult4 ] ) . to obtain such a point @xmath368 in place of a smooth point @xmath616 on @xmath7
, we first blow up @xmath616 , and then on the obtained exceptional divisor we blow up another point @xmath617 , so as to end with a _ chain of two _ @xmath517 s ( cf . fig .
[ mu-5,i ] ) .
finally , we blow down the first @xmath517 , and develop the required rational double point @xmath368 . as a result
, we have two `` smooth '' and one `` singular '' blow - ups , which implies @xmath618 @xmath619
[ effectivecovers ] in this section we construct the final type of triple covers in the basic construction .
these will not be necessary for the global calculation in section 7 , so the reader may wish to skip this more technical part on a first reading , and assume in section 6 that all covers are flat . consider a quasi - admissible cover @xmath621 , as given in prop .
[ propquasi ] . in order to use the fact that the pushforward @xmath622 is locally free on @xmath242 , we need to assure that the map @xmath243 is _
flat_. unfortunately , there are certain points on @xmath247 where this fails to be true : exactly where the fibers of @xmath247 are ramified as triple covers of the corresponding fibers of @xmath242
. the situation can be resolved by several further blow - ups .
namely , we work locally analytically near the points in @xmath249 of ramifications index 1 or 2 , and consider correspondingly two cases .
this case involves members of the boundary divisors @xmath497 and @xmath498 .
let @xmath96 be the point of ramification in the fiber of @xmath247 over the point @xmath335 in the fiber of @xmath242 ( cf .
[ ram ] ) .
we use the pull - back of the map @xmath243 to study the embedding of the completion of the local ring of @xmath335 into that of @xmath96 : @xmath623 \big/_{\displaystyle{(uv - t^2 ) } } \stackrel{\widetilde{\phi}^{\ # } } { \hookrightarrow}\widehat{{\mathcal}{o}}_{\widetilde{x}\!,q}= \mathbb{c}[[x , y , t]]\big/_{\displaystyle{(xy - t)}}.\ ] ] @xmath624 therefore , as an @xmath625-module , @xmath626 however , this is not a locally - free @xmath627-module : for instance , one relation among the generators is @xmath628 .
alternatively , the fiber of @xmath342 over @xmath335 is supported at @xmath96 , but it is of degree 3 rather than 2 , which would have been necessary for the flatness of a degree @xmath189 map . indeed , as @xmath629vector spaces
: @xmath630\big/_{\displaystyle{(x^2,y^2,xy)}}=\mathbb{c}\oplus \mathbb{c}x\oplus\mathbb{c}y.\ ] ] in fig .
[ ram ] one can visually observe the two distinct tangent directions at @xmath96 making it a _ fat _ point of degree @xmath520 .
we conclude that @xmath243 is indeed non - flat at @xmath96 . to resolve this , we blow - up @xmath242 at @xmath335 and @xmath247 at @xmath96 , denoting the new surfaces by @xmath55 and @xmath249 .
it is easy to see that they fit into the following coming diagram : @xmath631 in order to keep the map @xmath255 of degree 3 , we need to blow - up one further point on @xmath247 : if the inverse image of @xmath335 is @xmath632 we blow - up @xmath368 , and thus we add the necessary component to @xmath249 to make it a triple cover of @xmath55 ( cf .
fig [ coef2.fig ] ) .
the only boundary component , where ramification index 2 occurs , is @xmath499 .
similarly as above , @xmath600 is non - flat at @xmath96 .
indeed , @xmath633 is generated as an @xmath625-module by @xmath634 , but not - freely due to the relation @xmath635 .
to resolve the apparent non - flatness of @xmath243 , we can blow - up once @xmath247 and @xmath242 at @xmath96 and @xmath335 , but this would not be sufficient .
in fact , we must make further blows - ups on each surface , as fig .
[ resolve2 ] suggests : two more on @xmath247 and one more on @xmath242 .
@xmath636 in both cases of ramification index 1 or 2 , the new map @xmath307 is obtained from @xmath243 by a base change , and hence @xmath250 is _ proper _ and _ finite _ , and by construction , also a _ flat
_ morphism .
we call such covers
_ effective_. the above considerations combined with prop .
[ quasicov ] imply the existence of effective covers for our families of trigonal curves : let @xmath25 be a family of trigonal curves with smooth general member .
after several blow - ups ( and possibly modulo a base change ) we can associate to it an effective cover @xmath255 .
[ effexist ] here @xmath55 is a birationally ruled surface over @xmath6 .
if the base curve @xmath6 is _ not _ tangent to the boundary divisors @xmath637 , then the resulting surfaces @xmath249 and @xmath55 will have smooth total spaces . if , moreover , @xmath6 intersects the @xmath638 s only in their general points ( as given in fig . [ delta - k , i ] )
, then the special fibers of @xmath55 and @xmath249 are easy to describe ( cf . fig .
[ coef1.fig]-[coef3.fig ] ) .
for example , @xmath55 s special fibers are either chains of two or three reduced projective lines , or chains of five smooth rational curves with non - reduced middle component of multiplicity two .
the special fibers of @xmath249 can also contain nonreduced components ( of multiplicity 2 or 3 ) , and this occurs only in the ramification cases discussed above ( @xmath275 for @xmath639 ) .
this is an analog to the discussion in section [ the invariants ] .
after the necessary base changes we again identify , without loss of generality , the new base curve @xmath237 with @xmath6 , and the pull - back of @xmath7 over @xmath237 with @xmath7 , and we redefine the basic invariants @xmath262 and @xmath641 for the effective family @xmath249 over @xmath237 .
( it does nt make sense to define directly @xmath263 , because of the nonreduced fiber components in @xmath249 .
we could , of course , set @xmath642 , but we will not need this in the sequel . ) now the original @xmath7 and the effective @xmath249 differ by `` quasi - admissible '' and `` effective '' blow - ups .
the connections between the invariants of @xmath7 , @xmath247 and @xmath249 are expresssed by the following with the above notation , @xmath643 [ changeinv ] in view of lemma [ mu(c ) ] , the first and the second statements are obvious . obtaining a flat cover @xmath644 requires blowing up on @xmath247 one smooth point for each ramification index 1 , and three smooth points for each ramification index 2 .
hence the relation between @xmath641 and @xmath645
[ embedding ] given the effective degree @xmath520 map @xmath307 , our next step is to embed @xmath249 into a projective bundle @xmath254 of rank @xmath219 over the birationally ruled surface @xmath55 .
we shall consider a degree @xmath520 morphism @xmath250 , but the same discussion is valid for any degree @xmath87 .
since @xmath250 is flat and finite , the pushforward @xmath646 is a locally free sheaf on @xmath55 of rank @xmath520 .
define the _ trace _ map @xmath647 as follows .
the finite field extension @xmath648 of @xmath649 induces the _ algebraic _ trace map @xmath650 , defined by @xmath651 . here
the @xmath218 s are the conjugates of @xmath80 over @xmath649 in an algebraic closure of @xmath648 .
the restriction @xmath652 . over an affine open @xmath653 and its affine inverse @xmath654
, @xmath6 is the integral closure of @xmath515 in its field of fractions @xmath648 .
therefore , the trace map restricts to the @xmath515-module homomorphism @xmath655 .
we have a commutative diagram : ( 4,2.5)(4,0 ) ( 7.8,2.2)@xmath656 ( 8,1.2)(1.6,0)2(0,1)0.9 ( 7.8,0.7)@xmath657 ( 7.8,1.65)(0.35,0.9)[l ] ( 7.8,1.2)(1,0)0.1 ( 7.8,2.1)(1,0)0.1 ( 9.8,1.65)(0.35,0.9)[r ] ( 9.8,1.2)(-1,0)0.1 ( 9.8,2.1)(-1,0)0.1 ( 7,1.5)(3.1,0)2@xmath658 ( 3.2,0.7)@xmath659 ( 2.3,2.2)@xmath660 ( 3.7,1.5)@xmath661 ( 3.7,1.4)@xmath661 ( 3.3,1.5)@xmath250 the so - defined local maps @xmath662 patch up to give a global trace map @xmath663 let @xmath664 be the kernel of @xmath258 : @xmath665 note that @xmath664 is locally free of rank @xmath189 .
the natural inclusion @xmath666 , composed with @xmath258 , is the identity on @xmath667 , hence the exact sequence splits : @xmath668 it is useful to interpret the trace map geometrically in terms of the corresponding vector bundles @xmath669 , @xmath670 and @xmath8 associated to the sheaves @xmath671 , @xmath672 and @xmath664
. we again localize over affine opens , and if necessary , we shrink @xmath659 so that @xmath673 becomes a _ free _ @xmath674-module
. let @xmath335 be a closed point in @xmath675 with maximal ideal @xmath676 , having three distinct preimages @xmath677 with maximal ideals @xmath678 . since @xmath6 is a free @xmath515-module ,
the quotient @xmath679 is a 3-diml algebra over the ground field @xmath680 , i.e. a 3-diml vector space over @xmath681 . on the other hand , from @xmath682 and the chinese remainder theorem ,
it is clear that @xmath683 the generators @xmath684 are chosen as usual : @xmath685 , for instance , is the residue in @xmath686 of a function @xmath687 such that @xmath688 . in the groethendieck style ,
the vector bundle over @xmath689 associated to @xmath690 is @xmath691 , where @xmath692 is the symmetric algebra of @xmath6 over @xmath515 .
its fiber over @xmath335 is the pull - back @xmath693 we prefer to work with the dual @xmath669 of this bundle , and the same goes for projectivizations : we projectivize the 1-diml subspaces of @xmath669 rather than its 1-diml quotients . in view of this convention ,
the fiber of the bundle @xmath669 is canonically identified as @xmath694 the generators @xmath684 span three lines in @xmath695 , which can be canonically described : the line @xmath696 , for example , corresponds to all functions regular at @xmath697 and @xmath373 , and vanishing at @xmath368 and @xmath373 . similarly , the vector bundle @xmath670 associated to @xmath674 has fiber @xmath698 , where @xmath699 is a function near @xmath335 having residue @xmath700 in @xmath701 .
the trace map @xmath702 translates fiberwise in terms of the vector bundles @xmath669 and @xmath670 as : @xmath703 finally , the locally free subsheaf @xmath704 is associated to a vector bundle @xmath8 with fiber @xmath705 equivalently , from the direct sum ( [ directsum ] ) , @xmath706 , where the line @xmath707 corresponds to pull - backs of functions regular at @xmath335 .
@xmath708 since the four lines @xmath709 and @xmath710 are in general position in the fiber @xmath711 , modding out by @xmath710 yields three distinct lines in the fiber @xmath712 ( cf .
[ geometric ] ) . therefore , projectivizing @xmath712 naturally induces three distinct points @xmath713 in the fiber @xmath31 of @xmath238 .
going the other way around the diagram , we first projectivize @xmath714 , and then we project from the point @xmath715 $ ] onto the fiber of @xmath238 . in other words ,
@xmath716 } : { { \mathbf p}}^2\dashrightarrow { { \mathbf p}}^1 $ ] is well - defined at @xmath717,[\lambda_r]$ ] and @xmath718 $ ] . this completes the interpretation of the trace map in the case of three distinct preimages @xmath719 in @xmath249 . in case of only two or one preimage of @xmath335 in @xmath249 , one modifies correspondingly the above interpretation .
we construct the map @xmath720 via the use of an invertible relative dualizing sheaf @xmath721 .
its existence imposes a mild technical condition on the schemes @xmath722 and @xmath689 : we assume that they are gorenstein , i.e. cohen - macaulay with invertible dualizing sheaves @xmath723 and @xmath724 .
in our situation this will be sufficient . as we noted in section [ constructioneffective ] , when the base curve @xmath6 is _
not _ tangent to the boundary divisors @xmath275 , then @xmath249 and @xmath55 are smooth surfaces .
the remaining cases are `` local '' base changes of these , and the construction carries over .
let @xmath725 be a flat and finite degree @xmath87 morphism of gorenstein schemes , with @xmath689 integral . then @xmath250 factors through a natural embedding of @xmath726 into the projective bundle @xmath254 , followed by the projection @xmath727 ( cf .
[ basicconstruction ] ) . [ propembedding ] for easier referencing in the sequel
, we have kept the notation @xmath722 and @xmath689 , but the statement is true for _ any _ schemes satisfying the gorenstein condition . for another proof of prop .
[ propembedding ] , see @xcite .
_ proof of prop .
[ propembedding]_. here we construct the map @xmath728 , give the proof of its regularity , and point out how to show its injectivity .
( 6,2.5)(-2.3,2.5 ) ( -1.7,4)@xmath729 ( 0.95,3.05)(0,1)0.8 ( 0.75,2.6)@xmath722 ( 0.6,3.35)@xmath730 ( 1.1,2.95)(2,1)1.85 ( 1.7,3.35)@xmath61 according to prop .
ii.7.12 @xcite , to give a morphism @xmath731 over @xmath689 is equivalent to give an invertible sheaf @xmath408 on @xmath689 and a surjective map of sheaves @xmath732 .
recall from _ relative serre - duality _ that @xmath733 , and let @xmath734 .
the natural morphism @xmath735 is _
surjective_. this is in fact true for any quasicoherent sheaf @xmath736 on @xmath722 . indeed , restricting to the affine open sets @xmath737 , we have @xmath738 for some finitely generated @xmath6-module @xmath739 , and @xmath740 the surjective @xmath6-module homomorphism @xmath741 , given by @xmath742 , induces @xmath743 .
thus , the above map @xmath744 naturally defines a morphism @xmath745 over @xmath689 .
projectivizing @xmath746 gives a sequence of projective bundles , as in fig .
[ construction of i ] .
the map @xmath747 is undefined exactly on the image of @xmath748 .
composing @xmath747 with the map @xmath730 yields the map @xmath749 , which we claim is a regular map . to see regularity ,
we show that the restriction of @xmath750 is also surjective .
indeed , we again work locally , and let @xmath751 be the decomposition of @xmath6 via the trace map as a free @xmath515-module , where @xmath752 with @xmath753 .
let @xmath754 for some finitely generated @xmath6-module @xmath739 . recall that @xmath755 , so that as @xmath515-modules : @xmath756 , and @xmath6 acts on @xmath739 by @xmath757 naturally , the sheaf @xmath758 , and @xmath759 , where we think of @xmath760 as an element of @xmath761 by extending it via @xmath762 .
our statement is equivalent to showing that the @xmath6-module homomorphism @xmath763 is surjective . in fact
, it suffices to show that the trace map is in the image of @xmath744 , i.e. to find @xmath764 such that @xmath765 here @xmath766 gives the @xmath767-coordinate of @xmath98 , @xmath768 .
set @xmath769 and @xmath770 .
evaluating both sides of ( [ trace equation ] ) at @xmath771 and @xmath772 yields the same result , hence the identity is established , and @xmath750 is surjective .
we have shown that the composition @xmath773 is a regular map on @xmath249 .
alternatively , one could employ the geometric interpretation of the trace map .
point @xmath774 has three preimages @xmath719 in @xmath722 , each of which defines canonically a distinct point @xmath717,[\lambda_r]$ ] or @xmath718 $ ] in the fiber of @xmath775 . as we pointed above , the projection @xmath716 } : { { \mathbf p}}^2\dashrightarrow { { \mathbf p}}^1 $ ] is well - defined at @xmath717,[\lambda_r]$ ] and @xmath718 $ ] . but
@xmath716}$ ] is precisely the fiberwise restriction of @xmath776 , which shows again that the composition @xmath777 is regular on an open set of @xmath722 .
one makes the necessary modifications in the cases of fewer preimages of @xmath335 in @xmath249 .
finally , one can show , using similar methods ( either algebraically or geometrically ) , that the map @xmath61 is also an embedding .
* remark 6.1 .
* since the general fiber @xmath230 of @xmath249 is a smooth trigonal curve , the restriction @xmath778 embeds @xmath230 in a ruled surface @xmath779 over the corresponding fiber @xmath780 of @xmath55 , where @xmath781 . in section [ maroniinvariant ]
we will describe how the ruled surface @xmath231 varies as the fiber @xmath230 varies in @xmath249 .
[ triplecover ] in this section we consider the simplest case of effective covers , namely , when the original family @xmath7 is itself a triple cover of a _ ruled surface _ @xmath39 over the base curve @xmath6 .
this happens exactly when all fibers of @xmath7 are irreducible , and the linear system of @xmath35 s on the smooth fibers extends over the singular fibers to base - point free line bundles of degree 3 with two linearly independent sections . as we saw in section [ quasi - admissible ] , we can patch together all these @xmath35 s in a line bundle @xmath783 on the total space of @xmath7 .
thus , @xmath7 will map to @xmath784 via @xmath785 , and this map will factor through our ruled surface @xmath39 over @xmath6 : @xmath786 equivalently , we can describe such a family @xmath1 via the classification of the boundary components of the trigonal locus in section 4 : in @xmath10 the base curve @xmath6 meets only the boundary component @xmath277 of irreducible curves ( @xmath787 ) , and there are no hyperelliptic fibers in @xmath7 ( @xmath788 ) . as it will turn out , the calculation of the slope @xmath789 in this basic case yields the actual maximal bound @xmath790 : any addition of singular fibers belonging to other boundary components of @xmath10 will only decrease the ratio . henceforth , we distinguish among two types of calculation : _ global _ and _ local_. the _ global _ calculation refers to finding the coefficients of @xmath56 and the hodge bundle @xmath791 in a relation in @xmath792 involving _ all _ boundary classes .
calculation provides the remaining coefficients by considering _
local invariants _ of each individual boundary class ( cf . sect .
8) . for the remainder of this section
, we consider a family @xmath1 such that , as above , @xmath7 is a triple cover of the corresponding ruled surface @xmath39 , and we carry out the proposed global calculation .
0.11 in ( 10,9)(25,21 ) ( 26.5,24)@xmath7 ( 28.4,24.2)(1,0)3.1 ( 32,24)@xmath39 ( 29,20)@xmath6 ( 27.3,23.5)(1,-1)2 ( 32.3,23.5)(-1,-1)2 ( 26.6,22)@xmath92 ( 31.75,22)@xmath412 ( 29.45,24.7)@xmath342 ( 27.3,25.4)(1,1)2 ( 30.3,27.4)(1,-1)2 ( 29,27.8)@xmath238 ( 32,26.4)@xmath137 ( 27.4,26.4)@xmath61 recall that the pushforward of the structure sheaf @xmath793 to @xmath39 is a locally free sheaf of rank @xmath520 . in the exact sequence : @xmath794 the kernel of the trace map @xmath258 is a vector bundle @xmath795 on @xmath39 of rank @xmath189 , and @xmath7 naturally embeds in the rank 1 projective bundle @xmath238 over @xmath39 , where @xmath796 . any rank 2 vector bundle @xmath795 has the same projectivizations as its dual bundle @xmath8 since @xmath797 , where @xmath798 is a line bundle . for easier notation , in the trigonal case we use the dual @xmath8 instead of @xmath795 from section [ embedding ] . a basis for @xmath799 can be chosen by letting @xmath800 be the fiber of @xmath39 , and @xmath801 be any section of @xmath802 .
hence @xmath803 .
we normalize by replacing @xmath801 with the @xmath804-linear combination @xmath805 , and provide a @xmath806-basis for @xmath807 : @xmath808 let @xmath809 denote the class of the hyperplane line bundle @xmath810 on @xmath238 , and let @xmath811 and @xmath296 be the chern classes of @xmath8 on @xmath39 .
the chow ring @xmath812 is generated as a @xmath813-module by @xmath809 with the only relation : @xmath814 in particular , for the picard groups : @xmath815 as a divisor on @xmath238 , the surface @xmath7 meets the fiber @xmath816 of @xmath137 generically in three points ( @xmath7 maps three - to - one onto @xmath39 ) .
thus in the chow ring @xmath812 we have @xmath817\cdot [ f_{\pi}]=3 $ ] , which simply means that @xmath7 can be expressed as @xmath818 for some divisor @xmath819 on @xmath39 ( see ( [ q - basis ] ) ) .
with respect to the chosen basis for @xmath807 : @xmath820 for some @xmath821 . note that @xmath822 and @xmath823 .
it is evident that the divisors @xmath819 and @xmath811 can not be independent on ruled surface @xmath39 since both are canonically determined by the surface @xmath7 .
the relation is in fact quite straightforward . with the above notation for the triple cover @xmath824
, we have @xmath825 in @xmath799 . [ d=2c_1(v ) ] we start with the standard exact sequence for the divisor @xmath7 on @xmath238 : @xmath826 pushing to @xmath39 yields : @xmath827 it is easy to show that @xmath828 and @xmath829 .
this follows from grauert s theorem @xcite : @xmath830 and @xmath831 furthermore , @xmath832 and @xmath833 , so that ( [ pushforward ] ) becomes @xmath834 from relative serre - duality , and using the first chern class of the relative dualizing sheaf , @xmath835 ( cf .
( [ omega - pi ] ) ) : @xmath836 since @xmath837 ( cf .
@xcite ) , @xmath838\widehat{\phantom{t}}. \label{tranformedsequence}\ ] ] finally , combining ( [ tranformedsequence ] ) with ( [ remainingnonzero ] ) and @xmath839 , we arrive at @xmath840 and hence @xmath825 in @xmath841 in the following proposition we express @xmath267 and @xmath597 in terms of @xmath823 and the chern class polynomial @xmath842 , both of which are independent of the choice of the vector bundle @xmath8 . indeed , if we twist @xmath8 by a line bundle @xmath739 on @xmath39 and set @xmath843 , then @xmath844 @xmath845 recall the notation of ( [ define d , c1(v ) ] ) . in order to make @xmath87 also invariant
, we use @xmath846 from lemma [ d=2c_1(v ) ] and write @xmath847 .
now , if we replace @xmath238 with its isomorphic @xmath848 ( cf .
[ invariance ] ) , and set @xmath849 to be the new hyperplane bundle , then in @xmath850 : @xmath851 with @xmath852 .
hence @xmath853 and equating their degrees on @xmath854 , we obtain @xmath855 .
0.11 in ( 10,7)(25,20.5 ) ( 26,24)@xmath848 ( 28.4,24.2)(1,0)3.1 ( 32,24)@xmath238 ( 29,20)@xmath39 ( 27.3,23.5)(1,-1)2 ( 32.3,23.5)(-1,-1)2 ( 26.5,22)@xmath856 ( 31.75,22)@xmath137 ( 29.45,24.7)@xmath61 in other words , the following formulas for @xmath267 and @xmath597 would be valid for any vector bundle @xmath857 in place of the canonically defined @xmath8 , as long as the diagram of the basic construction ( cf . fig .
[ basicconstruction ] ) is satisfied , and as long as we adjust the degree @xmath858 by its invariant form @xmath859 .
let @xmath824 be a degree 3 map from the original family @xmath7 of trigonal curves to the ruled surface @xmath39 over @xmath6 .
the invariants @xmath267 and @xmath597 are given by the formulas : @xmath860 [ lambda_x , kappa_x ] we defer the proof of prop .
[ lambda_x , kappa_x ] to subsections 7.4.2 - 3 . the proof of prop .
[ lambda_x , kappa_x ] consists of two calculations in the chow ring of @xmath238 ; one uses versions of riemann - roch theorem on @xmath7 and @xmath238 , and the other uses the adjunction formula on @xmath238 for the divisor @xmath7 . here
we discuss these statements and set up the necessary notation . in order to work in @xmath861
, we express the chern classes of @xmath238 in terms of the hyperplane class @xmath809 and the chern classes of @xmath39 .
we first recall that @xmath862 . in the euler sequence on @xmath863 : @xmath864 we compare the chern polynomials @xmath865 , and obtain : @xmath866 here @xmath867 and @xmath868 are correspondingly the relative tangent and the relative dualizing sheaves of @xmath137 , while @xmath869 is the class of the canonical sheaf on @xmath238 . on the ruled surface @xmath39 over the curve @xmath6 of genus @xmath870
we similarly have @xmath871 now let @xmath230 be the general fiber of @xmath7 , i.e. a smooth trigonal curve of genus @xmath11 .
assuming the basic construction for the triple cover @xmath782 ( cf .
[ basicconstruction ] ) , we have the following lemmas .
if @xmath872 denotes the holomorphic euler characteristic of any sheaf @xmath76 , then the invariant @xmath267 is expressible as @xmath873 [ euler ] _ proof . _ from grothendieck - riemann - roch theorem for the map @xmath89 , @xmath874 where @xmath875 and @xmath876 are the corresponding tangent sheaves . since the fibers of @xmath92 are one - dimensional , @xmath877 .
substituting : @xmath878 @xmath879 @xmath880 note the similarity between this formula and the formula for @xmath881 in example 2.1 .
both quantities are expressed as differences of the euler characteristic ( holomorphic or topological ) on the total space of @xmath7 and the product of the corresponding euler characteristics on the base @xmath6 and the general fiber @xmath230 .
lemma [ euler ] suggests that in order to calculate @xmath267 , we must have control over @xmath882 . in the chow ring of @xmath238 : @xmath883\ ] ] [ holomorphiceuler ] from the standard exact sequence ( [ x - divisorsequence ] ) for the divisor @xmath7 on @xmath238 we have @xmath884 . on the other hand , hirzebruch - riemann - roch claims that for any sheaf @xmath885 on @xmath238 : @xmath886 . applying this to the line bundles @xmath887 and @xmath888 , and subtracting the results completes the proof of the lemma .
the reader may have noticed that all quantities discussed in the above lemmas are elements of the third graded piece @xmath889 of the chow ring @xmath890 .
hence they are cubic polynomials in the class @xmath809 , whose coefficients are appropriate products of pull - backs from @xmath891 .
the higher degrees @xmath892 and @xmath893 can be decreased using the basic relation ( [ zeta - relation ] ) , while @xmath809 itself can be altogether eliminated by noticing that for any @xmath894 : @xmath895 it is also useful to remember the trivial fact that for any divisors @xmath896 on @xmath39 , @xmath897 implies @xmath898 .
@xmath899 the canonical bundle @xmath900 of a smooth divisor @xmath901 on the smooth variety @xmath902 can be expressed as @xmath903 .
consequently , @xmath904 [ adjunction ] _ proof .
_ for the general statement of the adjunction formula see @xcite .
the expression for @xmath905 is a straightforward application to the divisor @xmath7 on @xmath238 : @xmath906 is being squared in @xmath907 . as for the genus
@xmath11 of the general member @xmath230 of our family , we consider a general fiber @xmath800 of @xmath39 ( cf .
[ intersection ] ) .
its pullback @xmath908 is a rational ruled surface @xmath909 over @xmath800 , embedded in the 3-fold @xmath238 .
the intersection of @xmath909 with the surface @xmath7 is the trigonal fiber @xmath910 . from the adjunction formulas for @xmath911 and @xmath912 : @xmath913 we substitute in lemma [ holomorphiceuler ] the expressions ( [ k_pv][c_2(pv ) ] ) for @xmath914 and @xmath296 , as well as the identity @xmath825 : @xmath915 @xmath916.\ ] ] applying the necessary reductions , we arrive at :
@xmath917 we expect our formula for @xmath267 to be independent of the base curve @xmath6 .
the contribution of @xmath870 in @xmath882 can be written as : @xmath918 , but this is precisely the adjustment @xmath919 given in lemma [ euler ] .
thus , @xmath920 it remains to notice that @xmath921 and rewrite @xmath267 in the form @xmath922 since @xmath923 , @xmath924 from lemma [ adjunction ] we calculate @xmath925 we calculate the contribution of @xmath870 in @xmath905 : @xmath926 , which is exactly the necessary adjustment for @xmath597 in ( [ kappa ] ) .
therefore , @xmath927 now that we have completed the proof of prop .
[ lambda_x , kappa_x ] , we notice that any bound on the ratio @xmath928 would be equivalent to some inequality involving the genus @xmath11 and the two invariants discussed earlier : @xmath929 and the quantity @xmath930
. this inequality should be a fairly general one , since the only relevant information in our situation is that @xmath7 is a triple cover of a ruled surface @xmath39 .
one way of obtaining such general inequalities in @xmath931 is via let @xmath932 be an ample divisor on the smooth surface @xmath7 , and let @xmath40 be a divisor on @xmath7 , numerically not equivalent to 0 . if @xmath933 , then @xmath934 . [ index ]
the question here , of course , is how to find suitable divisors @xmath932 and @xmath40 that would yield our result for the maximal slope bound .
for that , we make use of the triple cover @xmath824 . if @xmath932 is any _ ample _ divisor on @xmath39 , then its pullback @xmath935 to @xmath7 is also ample .
this follows from a divisor @xmath515 on the smooth surface @xmath7 is ample if and only if @xmath936 and @xmath937 for all irreducible curves @xmath230 in @xmath7 .
[ nakai ] since @xmath932 is ample itself , @xmath938 and @xmath939 for any curve @xmath230 on @xmath7 , so that @xmath935 is also ample on @xmath7 .
now , if we find a divisor @xmath40 on @xmath7 such that @xmath940 , we will have assured that @xmath941 , and then the index theorem will assert @xmath297 . as @xmath7 is a divisor itself on @xmath238 ,
its picard group naturally contains the restriction of @xmath942 to @xmath7 .
we look for @xmath40 inside this subgroup , and for our purposes we may write it in the form @xmath943 for some @xmath944 .
let @xmath230 be any divisor class @xmath945 .
we compute @xmath946 we want this to be zero for all @xmath230 , so we naturally take @xmath947 .
we summarize the above discussion in the divisor class @xmath948 on @xmath7 satisfies @xmath949 .
in particular , for an ample divisor @xmath932 on @xmath39 , the pullback @xmath935 is also ample on @xmath7 and @xmath950 .
consequently , @xmath297 with equality if and only if @xmath40 is numerically equivalent to @xmath951 on @xmath7 .
[ eta ] we have shown that @xmath952 or equivalently , @xmath953 we are now ready to find a maximal bound for the slope of @xmath7 .
recall the formulas for @xmath267 and @xmath597 ( cf .
[ lambda_x , kappa_x ] ) , and write @xmath954 in view of the type of bound for the ratio @xmath928 , which we aim to achieve , we have to eliminate any extra terms and use inequality ( [ eta ] ) .
our only choice is to subtract @xmath955\\ & \!\!\!=\!\!\!&(g-3)(9c_2(v)-2c_1 ^ 2(v))\geq 0\end{aligned}\ ] ] as a result , we establish an exact maximal bound for the slopes of our triple covers : given a triple cover @xmath956 satisfying in the basic construction , the slope of @xmath7 satisfies @xmath957 equality is achieved if and only if @xmath958 , or @xmath959 and @xmath71 on @xmath7 .
[ maintheorem ] from grr , applied to @xmath824 and the sheaf @xmath960 , we obtain a description of @xmath811 : @xmath961 @xmath962 @xmath963 for the _ ramification _ divisor @xmath443 on @xmath7 we know @xmath964 , so that @xmath965 .
hence @xmath966 .
in other words , from lemma [ d=2c_1(v ) ] we conclude that @xmath811 is half of the _ branch _ divisor @xmath819 on @xmath39 . on the other hand , we can rewrite the condition @xmath71 in the following way : @xmath967 @xmath968 the self - intersection of @xmath7 on @xmath238 satisfies ( cf .
@xcite ) @xmath969 in particular , our condition @xmath71 can be expressed as @xmath970 .
constructing examples of families achieving the maximal bound is not so easy , considering that the condition @xmath71 is not useful in practice .
instead , we start from the basic construction and attempt to find a ruled surface @xmath39 and a rank 2 vector bundle @xmath8 on it satisfying the equality in ( [ indexinequality ] ) , as well as the `` genus condition '' given in lemma [ adjunction ] .
the former will ensure the maximal ratio @xmath971 , while the latter will imply that the fibers of our family are indeed of genus @xmath11 .
the remaining question is what linear series @xmath972 has an irreducible member with at most rational double points as singularities , which would serve as the total space of our family @xmath7 .
it is hard to work with the canonically defined bundle @xmath973 , since not every vector bundle @xmath974 of rank @xmath189 on @xmath39 is of this form for some surface @xmath7 .
but any @xmath974 is congruent to some @xmath8 after a twist by an appropriate line bundle @xmath739 : @xmath975 , and @xmath976 .
so , it seems reasonable to start with @xmath974 rather than @xmath8 , and use the invariant forms of our required equalities ( cf .
[ globalcalculation ] ) .
this means replacing the degrees of @xmath811 on @xmath854 and @xmath800 by the corresponding invariant degrees of @xmath977 .
thus , we need for some divisor @xmath978 on @xmath39 : @xmath979 @xmath980 for a general fiber @xmath800 of @xmath39 consider the rational ruled surface ( cf . fig .
[ intersection ] ) : @xmath981 let @xmath982 be the section in @xmath983 with self - intersection @xmath984 , and let @xmath816 be the fiber of @xmath983 ( in terms of the map @xmath985 , @xmath986 ) . since a general fiber @xmath230 of our family is embedded in @xmath983 , the linear system @xmath987 has an irreducible nonsingular member .
equivalently , @xmath988 , i.e. @xmath989 ( compare with lemma [ gentrig ] ) .
this forces three types of extremal examples according to the remainders @xmath990 .
* example 7.1 ( @xmath991 ) .
* let @xmath992 for some @xmath993 .
set the base curve @xmath994 , and the ruled surface @xmath995 let @xmath801 be the section in @xmath39 with smallest self - intersection : @xmath996 , thus @xmath997 with @xmath998 .
let @xmath999 , and we choose two divisors @xmath978 and @xmath795 on @xmath39 as follows : @xmath1000 for the vector bundle @xmath974 on @xmath39 we set @xmath1001 so that @xmath1002 and @xmath1003 .
we claim that the linear system @xmath1004 on the 3-fold @xmath1005 contains an irreducible smooth member , which we set to be our surface @xmath7 with maximal ratio @xmath36 . @xmath1006 indeed , it is trivial to check conditions ( [ condition1]20 ) .
further , for _ any _ fiber @xmath800 of @xmath39 : @xmath1007 so that @xmath1008 satisfies the required conditions ( [ e - conditions ] ) .
the only nontrivial fact is the existence of the wanted member @xmath7 in the linear system @xmath125 on @xmath1005 .
consider two sections @xmath1009 and @xmath1010 of @xmath1005 corresponding to the subbundles @xmath1011 and @xmath1012 of @xmath974 , respectively : @xmath1013 so that @xmath1014 ( cf . fig .
[ maximal1.fig ] ) .
note that @xmath1015 and @xmath1016 .
in other words , if @xmath1017 is the ruled surface over @xmath801 , then @xmath1009 intersects every irreducible member of @xmath125 in the curve @xmath1018 . on the other hand ,
if a member of @xmath125 meets @xmath1009 in a point outside @xmath443 , then this member contains entirely @xmath1009 .
thus , @xmath125 does not distinguish the points on @xmath1009 , and @xmath443 is in the base locus of @xmath125 .
similarly , the restriction @xmath1019 has exactly one section on @xmath1020 , namely , @xmath1021 .
again it follows that @xmath125 does not distinguish the points on @xmath1020 .
away from the closed subset @xmath1022 , the linear system @xmath125 is in fact very ample .
this can be checked by showing directly that @xmath125 separates points and tangent vectors on @xmath1023 .
therefore , @xmath125 defines a rational map @xmath1024 the map @xmath1025 is regular on @xmath1026 , embeds @xmath1027 , and contracts @xmath1028 and @xmath1029 to two distinct points @xmath335 and @xmath96 in @xmath1030 . by bertini s theorem ( cf .
@xcite ) , the _ general _ member of @xmath125 is _ smooth _ away from the base locus @xmath443 .
let @xmath932 be a general hyperplane in @xmath1030 not passing through @xmath335 and @xmath96 .
pulling @xmath932 back to @xmath1005 yields a member @xmath7 of @xmath125 not containing @xmath1009 or @xmath1020 , and hence irreducible .
it remains to show that the total space of @xmath7 has at most finitely many double point singularities along the curve @xmath443 .
since the member @xmath1031 is smooth along @xmath443 , then the general member of @xmath1032 must be smooth along @xmath443 .
hence our surface @xmath7 has , in fact , a smooth total space .
this concludes the construction of our maximal bound family of trigonal curves .
* example 7.2 ( @xmath1033 ) . * set @xmath1034 for @xmath1035 .
then @xmath1036 satisfies the requirements of our construction : @xmath1037 and @xmath1038 . for the ruled surface @xmath39 we choose @xmath1039 .
let @xmath795 and @xmath978 be the following divisors on @xmath39 : @xmath1040 where @xmath1041 .
the vector bundle @xmath974 on @xmath39 is then defined by @xmath1042 .
finally , we indentify the total space of the surface @xmath7 with an irreducible smooth member of the linear system @xmath1004 on the 3-fold @xmath1043 .
the verification of this construction is similar to the previous example . here
@xmath125 is very ample everywhere on @xmath1043 except on the section @xmath1009 , which is contracted to a point under the map @xmath1044 .
this example , in somewhat different context , is shown in @xcite .
* remark 7.1 * the case of @xmath1045 is complicated by the fact that we can not take @xmath1046 , for then @xmath1047 .
for example , if @xmath1048 , then the only possibility is @xmath1049 . in the notation of section 12 , all trigonal curves have lowest maroni invariant of @xmath219 , and
there is no maroni locus . for now , in this case we have not been able to construct a trigonal family with singular general member , whose ratio is @xmath309 .
[ generalcase ] in this section we consider the general case of a trigonal family @xmath1 . for convenience of notation
, we shall assume that the base curve @xmath6 intersects transversally and in general points the boundary divisors of @xmath17 ( cf .
[ delta - k , i ] ) .
we will call such a base curve _ general _ , and use this definition throughout section 8 - 10 .
since we work in the rational picard group of @xmath5 , all arguments and statements in the remaining cases are shown similarly in sect .
11 . from prop . 6.1
, we may assume that modulo a base change , our family @xmath1050 fits in the following commutative diagram : ( 3,4.3)(-1,2 ) ( 0,3.9)@xmath1051 ( 0,5.1)@xmath1052 ( 0.2,5)(0,-1.2)3(0,-1)0.6 ( 1.45,3.8)(-1,-2)1 ( 0,2.8)@xmath7 ( 0,1.6)@xmath6 ( 1.55,5)(0,-1)0.6 ( 1.55,6.2)(0,-1)0.6 ( 1.65,5.8)@xmath137 ( 1.3,6.3)@xmath238 ( 0.2,5.6)(2,1)1.2 the special fibers of of @xmath249 and of the birationally ruled surface @xmath55 over @xmath6 are described in fig .
[ coef1.fig][coef3.fig ] . since each such fiber in @xmath55 is a _ chain
_ @xmath902 of rational components , we can fix one of the end components to be the _ root _ @xmath443 .
we keep the notation @xmath1055 ( @xmath1056 , respectively ) for the ancestor ( descendants , respectively ) of a component @xmath795 in @xmath902 .
we also fix a general fiber @xmath1057 of @xmath55 , and a section @xmath1058 , which is the pullback of the corresponding section @xmath854 in @xmath242 ( cf .
( [ normalize ] ) ) .
the rational picard group of @xmath55 is generated by @xmath1059 , @xmath1058 and all non - root components @xmath795 of the special fibers of @xmath55 : @xmath1060 the intersection numbers of these generators are as follows : @xmath1061 ( 3,2.3)(12.6,4.5 ) ( 6,5.3)(1,0)2 ( 5,4.6)(0,-1.7)2(2,1)1.6 ( 6,3.6)(0,-0.1)2(1,0)2 ( 9.1,4.2)(-2,-1)1.6 ( 4.9,5)(0,-1.8)2@xmath1055 ( 6.8,5.45)(0,-1.7)2@xmath795 ( 8.7,4.3)@xmath1056 ( 7,4)(3,-2.4 ) ( 4.45,2.9)(2,1)1.6 ( 5.5,3.6)(1,0)3 ( 6.5,3.4)(-1,1)1.2 ( 7.5,3.4)(1,1)1.2 ( 4.5,3.2)@xmath1055 ( 6.8,2.9)@xmath795 ( 5.1,4.75)@xmath1056 ( 5.9,4.85)@xmath1056 ( 7.6,4.85)@xmath1056 ( 8.4,4.75)@xmath1056 ( 6.75,3.4)(-1,2)0.65 ( 7.25,3.4)(1,2)0.65 ( 3,4)(-1,-1.9 ) ( 6,5.3)(0,-0.1)2(1,0)2 ( 5,4.6)(0,-1.65)2(2,1)1.6 ( 6,3.6)(1,0)2 ( 9.1,4.2)(0,1.8)2(-2,-1)1.6 ( 9.1,4.3)(-2,-1)1.6 ( 4.8,5)(0,-1.7)2@xmath443 ( 6.8,5.45)(0,-1.7)2@xmath1055 ( 8.7,4.3)(0,1.8)2@xmath795 we also set @xmath1062 ( cf . fig . [ m e ] ) : @xmath1063 in this notation , due to the fact that @xmath1064 , the self - intersection of any @xmath795 is computed by ( cf . fig .
[ e^2 ] ) : @xmath1065 in order to express the dualizing sheaf @xmath1066 in terms of the above generators of @xmath1067 , for each component @xmath795 in @xmath55 we denote by @xmath1068 the length of the path @xmath1069 , omitting any nonreduced components except for @xmath795 itself .
for example , in the two cases in fig .
[ theta_e ] we have @xmath1070 and @xmath1071 .
note that @xmath1072 .
considering the `` effective '' blow - ups on @xmath242 , necessary to construct @xmath55 , we immediately obtain the following identities ( compare with ( [ k_pv ] ) and ( [ k_yglobal ] ) ) . in @xmath1067 and @xmath1073
: @xmath1074 [ kdivisors ] the hyperplane section @xmath809 of @xmath238 and the rank 2 vector bundle @xmath8 on @xmath55 are defined similarly as in section 7 .
thus , in @xmath1073 we have @xmath1075 for a certain divisor @xmath819 on @xmath55 .
by analogy with lemma [ d=2c_1(v ) ] , one shows that @xmath1076 in @xmath1077 , so that @xmath1078 using the above notation for @xmath1067 we can write for some half - integers @xmath1079 : @xmath1080 here we can assume that @xmath1081 by replacing @xmath443 with a linear combination of the remaining components @xmath795 in its chain @xmath902 ( compare with ( [ define d , c1(v ) ] ) ) .
finally , we need the top chern classes of @xmath55 and @xmath238 in terms of intersections of known divisors and other known invariants of the two surfaces ( compare with ( [ c_2(pv ) ] ) and ( [ c_2(y)global ] ) ) . in the chow rings @xmath1082 and @xmath1083 the following equalities are true : @xmath1084 [ conventions ] in the sequel , we will work with several functions defined on the set of components @xmath1085 in @xmath55 . for easier calculations , to any such function @xmath92 we associate the _ difference function _ @xmath1086 by setting @xmath1087 for all @xmath795 .
since @xmath1088 does not exist , we define @xmath1089 for all roots @xmath443 in @xmath55 . for any functions @xmath92 and @xmath412 defined on the set of components @xmath1085 in @xmath55 , the following identity holds true : @xmath1090 [ technical ] _ proof .
_ we rewrite the lefthand side as @xmath1091 @xmath1092 @xmath1093 we have noted that all three functions @xmath1094 and @xmath1095 are zero on the roots @xmath443 in @xmath55 .
since we shall be working specifically with these three functions , it makes sense to restrict from now on all sums @xmath1096 only to the non - roots @xmath795 in @xmath55 . with this in mind , in every application of lemma [ technical ]
one must check that the corresponding functions @xmath92 and @xmath412 have the same property : @xmath1097 , so that we can restrict the sums in lemma [ technical ] also to all _ non - roots _ @xmath795 in @xmath55 .
in fact , in all cases this verification will be obvious as @xmath92 and @xmath412 will be , for the most part , linear combinations of @xmath1098 and @xmath1095 .
* example 8.1 . * from expression ( [ genc_1(v ) ] ) for @xmath811 as a divisor on @xmath55 , and lemma [ technical ] : @xmath1099 the following proposition [ hatlambda ] is a generalization of the corresponding statement in section 7 ( cf .
[ lambda_x , kappa_x ] ) .
we set @xmath1101 and @xmath1102 to be the difference functions of @xmath1095 and @xmath1098 .
the degrees of the invariants @xmath262 and @xmath641 on @xmath249 are given by @xmath1103 @xmath1104 [ hatlambda ] one starts with the euler characteristic formula @xmath1105 or the adjunction formula @xmath1106 the rest of the proof is a straight forward calculation , which uses the equalities given in [ conventions ] , and is substantially simplified by lemma [ technical ] .
the degree @xmath3 on the original family @xmath7 is given by @xmath1107 [ hatdelta ] here @xmath1108 stands for the quasi - admissible contribution to @xmath641 of the preimage @xmath1109 in @xmath249 , as defined in lemma [ mu(c ) ] .
since @xmath1110 , @xmath1111 , and @xmath1112 , the statement immediately follows from prop .
[ hatlambda ] . for a component @xmath795 in a special fiber @xmath902 of @xmath55
, we define @xmath1116 to be the subtree of @xmath902 generated by the component @xmath795 . in other words , @xmath1116 is the union of all components @xmath1117 such that @xmath1118 ( cf . fig . [ subtree ] ) . for simplicity , we set @xmath1119 to be the arithmetic genus of the inverse image @xmath1120 in @xmath249 . it can be easily computed via the following analog of lemma [ adjunction ] , where @xmath902 consisted of a single component @xmath1121 .
@xmath1122 for a general base curve @xmath6 and for any non - root component @xmath1123 : @xmath1124 [ arithgenus ] _ proof .
_ from the adjunction formula for the divisor @xmath1120 in @xmath249 : @xmath1125 here @xmath1126 if @xmath1127 , and @xmath1128 otherwise .
thus , the sums above are effectively taken over all @xmath1118 . substituting the expressions for @xmath869 and @xmath249 as divisors in @xmath863 from lemma [ kdivisors ] and ( [ genx ] )
, we arrive at @xmath1129 set @xmath1130 , i.e. @xmath1131 only if @xmath1132 ; otherwise , @xmath1133 . by lemma
[ technical ] , @xmath1134 now we can easily compute the invariants @xmath1135 , @xmath1136 and @xmath1137 , appearing in the formulas for @xmath1138 and @xmath1139 .
there are three possibilities for the triple @xmath1140 , depending on whether the components @xmath795 and @xmath1055 of @xmath902 are reduced : @xmath1141 [ constants ] note that for the list all possible special fibers @xmath902 of @xmath55 , each component @xmath795 fits in exactly one of the three cases above ( cf .
[ coef1.fig][coef3.fig ] ) .
the proof of the statement is immediate from the definitions of @xmath1135 and @xmath1136 , and from lemma [ arithgenus ] .
[ bogomolov1 ] with the conventions of section 8 , we state the main proposition of the section .
there exists an effective @xmath145-linear combination @xmath76 of boundary divisors @xmath275 , not containing @xmath277 , such that for a general base curve @xmath6 in @xmath17 : @xmath1143 [ bogomolov1 ] for a shorthand notation , we denote by @xmath1144 the difference @xmath1145 using the results of the previous section , we can write : @xmath1146 we defer the proof of prop .
[ bogomolov1 ] until the end of this section , when all of the terms in this sum will be computed . substituting the results of corollary [ constants ] in the expression for @xmath1144 ,
we eliminate @xmath1135 , @xmath1136 and @xmath1137 : @xmath1147 for each chain @xmath902 in @xmath55 , the inverse image @xmath1148 in @xmath249 is a member ( or a blow - up of a member ) of exactly one boundary divisor @xmath275 . consequently , to find the contribution to @xmath1144 of a specific @xmath275
, we calculate the sum in @xmath1144 corresponding to all types of special fibers @xmath1148 . @xmath1149 fig .
[ coef1.fig ] presents the special fibers corresponding to the boundary divisors @xmath1152 . in each of these cases , there is only one component @xmath795 in @xmath902 besides the root @xmath1153 .
thus , the subchain @xmath1116 in @xmath902 is trivial it consists only of @xmath795 .
its inverse image @xmath1154 is connected for @xmath63 , and consists of two connected curves for @xmath278 and @xmath1151 . setting the genus of the inverse image of @xmath443 to be @xmath61 , it is easy to see that the genus @xmath1155 of @xmath1156 is @xmath64 in the first two cases , and @xmath1157 in the third case.(the total genus of the original fiber of @xmath7 , drawn in full lines , must be @xmath11 . ) finally , counting the number of `` quasi - admissible '' blow - ups ( drawn by dashed lines ) , we see that @xmath1158 for @xmath63 , @xmath1159 for @xmath278 , and @xmath1160 for @xmath1151 ( cf .
lemma [ mu(c ) ] ) .
note that there are no ramification modifications .
the contribution of each such fiber @xmath1161 to the sum @xmath1144 is only one summand of the first type ( @xmath1162reduced ) , plus the quasi - admissible adjustment @xmath1163 .
if @xmath1161 corresponds to the boundary divisor @xmath275 , we denote this contribution by @xmath1164 . in conclusion ,
@xmath1165 in each of these cases , the fiber @xmath902 of @xmath55 consists of two rational curves @xmath1166 and @xmath1167 , and the root @xmath1168 ( cf .
[ coef2.fig ] ) .
there are no nonreduced components in @xmath902 , so the contribution to @xmath1144 consists of two summands of the first type ( @xmath1162 nonreduced ) , plus a quasi - admissible adjustment of @xmath1159 for @xmath498 , and a ramification adjustment of @xmath11 in both cases : @xmath1169 @xmath1170 the arithmetic genus of the nonreduced component of @xmath249 is @xmath1171 , and its intersection number with each of the neighboring components is 2 . setting @xmath1172 forces @xmath1173 .
hence , @xmath1174 and @xmath1175 .
substituting : @xmath1176 @xmath1177 it remains to consider the case of ramification index 2 .
here there are four components @xmath795 besides the root @xmath443 in the special fiber @xmath1178 . consequently , there are four summands in @xmath1144 corresponding to the @xmath458 s : @xmath1166 and @xmath1179 yield summands of the first type ( @xmath1162 reduced ) , @xmath1167 yields a summand of the second type ( @xmath795 nonreduced ) , and @xmath1180 yields a summand of the third type ( @xmath1055 nonreduced ) .
@xmath1181 since @xmath1158 , and the ramification adjustment is @xmath1182 , we obtain for the contribution of @xmath499 to @xmath1144 the following expression : @xmath1183 the arithmetic genera of the components in @xmath249 are denoted in the fig .
[ coef3.fig ] .
it is easy to see that @xmath1184 , @xmath1185 , @xmath1186 , @xmath1187 .
finally , @xmath1188 in the above discussion we calculated the contributions of the boundary divisors @xmath275 to the sum @xmath1144 , so that @xmath1189 with @xmath276 , and the corresponding limits for the index @xmath61 ( cf .
[ boundary ] ) .
it is now clear what the divisor @xmath76 should be .
we set @xmath1190 , and thus , @xmath1191 , @xmath1192 using the restrictions on the index @xmath61 for each type of boundary divisor @xmath275 , one can easily deduce that all coefficients @xmath1193 .
for instance , when @xmath182 $ ] : @xmath1194 in other words , @xmath76 is an effective rational linear combination of boundary divisors in @xmath17 , which by construction does not contain @xmath1195 recall that a vector bundle @xmath8 of rank 2 is _ bogomolov semistable _ if @xmath1196 . for a general base curve @xmath6 ,
if the canonically associated vector bundle @xmath8 is bogomolov semistable , then the slope of @xmath25 is bounded by @xmath66 [ 7 + 6/g bogomolov ] the statement follows directly from prop .
[ bogomolov1 ] . indeed , since @xmath76 is effective , then @xmath1197 .
by hypothesis , @xmath1198 , and @xmath46 .
hence , @xmath1199 for a general base curve @xmath6 the following relation holds true : @xmath1200}\frac{3}{2}(i+2)(g - i)\delta_{1,i}|_b+ \sum_{i=1}^{g-2}\frac{3}{2}(i+2)(g - i)\delta_{2,i}|_b\\ & \!\!\!+&\sum_{i=1}^{[g/2]}\frac{3}{2}(i+1)(g - i+1)\delta _ { 3,i}|_b+\sum_{i=1}^{[(g-1)/2 ] } \big(3(i+1)(g - i)-\frac{g-3}{2}\big)\delta_{4,i}|_b\\ & \!\!\!+&\sum_{i=1}^{g-1}\big(3(i+1)(g - i)-\frac{g-3}{2}\big ) \delta_{5,i}|_b+\sum_{i=1}^{[g/2 ] } \big(\frac{9}{2}i(g - i)-{\frac{g-3}{2}}\big)\delta_{6,i}|_b.\end{aligned}\ ] ] [ analog1 ] _ proof .
_ this is an immediate consequence of the established relation in prop .
[ bogomolov1 ] .
we replace @xmath3 by the linear combination ( [ divisorrel ] ) of the boundary classes of @xmath17 , and write @xmath1201 for some new coefficients @xmath60 .
recall that @xmath1202 denotes the _ multiplicity _ of @xmath58 in @xmath3 , so that @xmath1203 . for example , the coefficient of @xmath1204 is @xmath1205 or the coefficient of @xmath1206 is @xmath1207
[ index1 ] for a general base curve @xmath6 and for the rank 2 vector bundle @xmath8 on @xmath55 , we have @xmath1208 [ genindex ] the proof is identical to that of theorem [ indextheorem ] .
one considers the divisor @xmath40 on @xmath249 defined by @xmath1209 and shows that @xmath40 kills the pullback of any divisor on @xmath55 . in particular , @xmath40 kills an ample divisor on @xmath249 . by the index theorem on @xmath249 , @xmath1210 . from expression ( [ genx ] ) , this can be also written as @xmath1208 as in section 7 , the index theorem on @xmath249 suggests to replace the bogomolov difference @xmath1211 by another linear combination of @xmath296 and @xmath1212 , which would behave in a more `` predictable '' way , namely , by @xmath1213 . in the process of doing so , the only way to eliminate the unnecessary global terms @xmath87 and @xmath1214 from a relation among @xmath19 and @xmath20 is to subtract : @xmath1215 for a general base curve @xmath6 and an effective rational combination @xmath300 of the boundary divisors @xmath275
, not containing @xmath277 , we have : @xmath1216 [ indexrelation ] note the apparent similarity between this relation and prop .
[ bogomolov1 ] .
one may use the latter to prove the former , but the calculations are not simpler than if one starts from scratch .
we will show a sketch of this proof , leaving the details to the reader , and referring to the proof of prop .
[ bogomolov1 ] for comparison .
we denote by @xmath1217 the difference @xmath1218 substituting for @xmath1219 and @xmath295 the corresponding identities from prop . [ hatlambda ] and example 8.1 , and recalling that @xmath1220 ( cf .
lemma [ adjunction ] ) , we write @xmath1217 as @xmath1221 as in lemma [ bogomolov1 ] , we group the above summands for every special fiber in @xmath249 , and correspondingly , for every chain @xmath902 in @xmath55 .
recall corollary [ constants ] , and the computations of the arithmetic genera @xmath1155 in the previous section : @xmath1222 with this at hand , it is not hard to calculate the contributions @xmath1223 of each boundary component @xmath275 to the sum @xmath1217 : @xmath1224 [ d_k , itable ] let @xmath1225 .
then @xmath1226 , and the desired relation would be established if @xmath300 is effective . given the restrictions on the indices @xmath61 of the coefficients
@xmath1223 in prop .
[ boundary ] , one easily shows that all @xmath1227 for a general base curve @xmath6 , the slope satisfies : @xmath1228 with equality if and only all fibers of @xmath7 are irreducible curves , and either @xmath958 or the divisor @xmath40 on the total space of @xmath1229 is numerically zero .
[ genmaximal ] from the index theorem on @xmath249 , it follows that @xmath298 . since @xmath300 is effective ,
@xmath1230 . then prop .
[ indexrelation ] implies @xmath1231 , with equality exactly when @xmath1232 and @xmath1233 .
the latter means that @xmath1234 because all coefficients @xmath1223 of @xmath300 are strictly positive .
in other words , the family @xmath249 has only _ irreducible _ fibers ( @xmath1235 ) .
this takes us back to section 7 , where we presented the global calculation on the triple cover @xmath782 .
there we concluded that the _ index condition _
@xmath1232 was equivalent to @xmath71 on @xmath7(@xmath1236 ) , or the genus @xmath958 . for a general base curve @xmath6 , @xmath1237}8(i+2)(g - i)\delta_{1,i}|_b+ \sum_{i=1}^{g-2}8(i+2)(g - i)\delta_{2,i}|_b\\ & \!\!\!\!+\!\!\!\!\!&\sum_{i=1}^{[g/2]}8(i+1)(g - i+1)\delta_{3,i}|_b+ \!\!\!\sum_{i=1}^{[(g-1)/2]}\!\!\!\!\ !
\big(16(i+1)(g - i)-2(g-3)\big)\delta_{4,i}|_b\\ & \!\!\!\!+\ !
\!\!\!&\sum_{i=1}^{g-1}\big(16(i+1)(g - i)-2(g-3)\big)\delta_{5,i}|_b+ \sum_{i=1}^{[g/2]}24i(g - i)\delta_{6,i}|_b.\end{aligned}\ ] ] [ analog2 ] [ page analog2 ] _ proof .
_ we only need to substitute the known expressions for the divisors @xmath300 and @xmath3 into prop .
[ indexrelation ] : @xmath1238 the rest is a simple calculation . for example , the total coefficient @xmath1239 of @xmath1240 equals @xmath1241
[ arbitrary ] we extend now the results of sect . 8 - 10 to arbitrary nonisotrivial families @xmath1242 with smooth trigonal general member .
the essential case is when @xmath6 is _ not _ tangent to the boundary @xmath605 , from which the remaining cases easily follows .
we now drop the hypothesis of the base curve @xmath6 intersecting the boundary divisors in general points .
instead , for now we only assume that the base curve @xmath6 is not tangent the boundary @xmath605 .
this means that all special fibers of @xmath7 locally look like the general ones ( cf .
[ coef1.fig][coef3.fig ] ) .
therefore , from the quasiadmissible cover @xmath487 we can construct an effective cover @xmath255 of _ smooth _ surfaces @xmath249 and @xmath55 . ( the _ smoothness _ indicates that @xmath6 is _ not _ tangent to any @xmath275 . otherwise , there would be a higher local multiplicity @xmath1244 near a node of a special fiber @xmath1245 , @xmath1246 .
hence @xmath249 would be obtained locally via a base change from a smooth surface , but @xmath249 would have a singular total space . )
now the special fibers of @xmath55 are _ trees _ @xmath902 ( rather than just chains ) of reduced smooth rational curves with occasional nonreduced rational components of multiplicity 2 .
the latter occur again exactly for each singular point in @xmath1247 of ramification index 2 under the quasiadmissible cover @xmath1248 ( cf .
[ coef3.fig ] ) .
the notation and conventions from sections [ conventions ] are also valid here .
in particular , for any tree @xmath902 , we fix one of its end ( nonreduced ) components to be its root @xmath443 , and we define as before the functions @xmath1249 on the components @xmath795 of @xmath902 .
moreover , since lemma [ technical ] can be applied also for any tree @xmath902 , the calculations of @xmath1250 and @xmath3 in prop .
[ hatlambda ] and cor .
[ hatdelta ] go through without any modifications .
finally , we wish to extend all results of sections 8 - 10 over the new base @xmath6 .
the only difference arises in the final calculation of the coefficients @xmath1164 and @xmath1223 .
the fiber @xmath1251 in @xmath7 , corresponding to a tree @xmath902 , may now lie in the intersection of _ several _ boundary divisors @xmath275 .
such a trigonal curve @xmath1251 is called a _ special boundary _ curve . accordingly
, its contribution @xmath1252 to @xmath1144 ( or @xmath1253 to @xmath1217 ) will be distributed among these divisors @xmath275 s , rather than just yielding a single coefficient @xmath1164 ( or @xmath1223 ) as before .
@xmath1254 this can be easily resolved .
the idea is to replace any special singular fiber in @xmath249 by a suitable combination of _ general _ fibers , without changing the sums @xmath1144 and @xmath1217 .
we can imagine this as `` moving '' the base curve @xmath6 in @xmath17 _ away from _ the special singular locus of @xmath17 , and replacing it with a _
general _ base curve @xmath801 , as defined in section 8 .
for example , in fig .
[ arbitrary1 ] the base @xmath6 passes through a point @xmath335 in the intersection of two boundary divisors @xmath1255 .
two new general points @xmath1256 and @xmath1257 , each lying on a single @xmath1255 , replace the special point @xmath335 , and thus @xmath6 moves to a _
general _ curve @xmath801 .
let @xmath1251 be a special boundary curve in @xmath17 . denote by @xmath1258 the degree of the point @xmath1259 $ ] in the intersection @xmath1260 .
then the contributions of @xmath1261 to @xmath1144 and to @xmath1217 are @xmath1262 and @xmath1263 , respectively .
[ contributions d_tc_t ] _ proof : _ rewrite @xmath1144 and @xmath1217 as sums over the non - root components @xmath795 of the special trees @xmath902 : @xmath1264 where @xmath1265 is the quasi - admissible and effective adjustment , and the functions @xmath1266 and @xmath1267 are quadratic polynomials in @xmath1155 with linear coefficients in @xmath11 .
recall that in these sums each non - root component @xmath795 appears exactly once , and @xmath1155 is the arithmetic genus of the inverse image @xmath1268 of the subtree @xmath1116 generated by @xmath795 .
there is a simple way to recognize the boundary divisors @xmath275 in which a special trigonal fiber @xmath1251 lies .
consider the corresponding `` effective '' fiber @xmath1269 in @xmath249 .
for any non - root component @xmath795 in @xmath902 there are two possibilities : either @xmath1270 and @xmath1271 are both reduced , or @xmath795 is part of a chain of length 3 or 5 , constructed to resolve ramifications in the quasi - admissible fiber @xmath1247 .
@xmath1272 consider the first situation , and denote by @xmath587 the preimage @xmath1274 in @xmath249 .
thus , @xmath587 corresponds to a general member of @xmath1275 , possibly of lower genus ( cf .
[ notchain ] ) . as part of the fiber @xmath1276
, the curve @xmath587 is represented for simplicity by the triple intersection of two _ smooth _ trigonal curves ( in @xmath63 ) , but it could have corresponded to any general member of @xmath278 or @xmath1151 . the solid box encompasses the preimage @xmath1277 , and the dashed box encompasses the preimage of the rest , @xmath1278 .
each of these boxes represents a limit of a quasi - admissible curve , @xmath491 or @xmath492 , which is naturally a triple cover of @xmath31 .
thus , we can _ `` smoothen '' _ each box to such a curve @xmath1279 . as a result
we obtain a quasiadmissible curve @xmath1280 of total genus @xmath11 , which corresponds to a general member of @xmath1281 or @xmath1151 .
depending on which divisor @xmath275 is evoked , there is a corresponding contribution of @xmath219 to the coefficient @xmath1258 : @xmath1259\in\delta{\mathfrak{t}}_{k , i}$ ] .
note that the arithmetic genus of @xmath492 is the previously defined @xmath1155 .
the contribution of @xmath795 to @xmath1144 is @xmath1282 plus the possible quasi - admissible adjustment in @xmath1108 needed to obtain @xmath1283 . in view of the above `` smoothening '' , this can be thought of as the contribution of @xmath492 in the effective curve @xmath1280 , and by prop .
[ bogomolov1 ] this is exactly the coefficient @xmath1164 .
the same argument works in the case of @xmath1217 from prop .
[ indexrelation ] .
we conclude that @xmath1258 ( for @xmath1284 ) equals the number of @xmath1164 s and @xmath1223 s in @xmath1144 and @xmath1217 , respectively .
we treat analogously the remaining case when the component @xmath795 is part of a chain of length 3 or 5 . here , however , one must consider _ simultaneously all _ the components @xmath795 of @xmath902 participating in such a chain , and take a quasi - admissible limit only _ over the reduced _ curves in @xmath1276 . in fig . [ chain ]
one can see all three ramification cases , or equivalently , the boundary divisors @xmath1286 and @xmath499 . for simplicity ,
we have again depicted the reduced components in @xmath249 by smooth trigonal curves , which may not always be true for every tree @xmath902 : they could , for instance , be singular or reducible , but they will keep the ramification index 1 or 2 at the appropriate points .
@xmath1287 to see how @xmath1164 and @xmath1223 are obtained , let us calculate , for example , the contributions of @xmath1288 and @xmath1179 in the case of @xmath499 .
the inverse images in @xmath249 of @xmath1289 and @xmath1290 are marked by dashed and solid boxes , respectively .
`` smoothen '' _ each box by a smooth trigonal curve , @xmath491 or @xmath492 , and keep the inverse images of @xmath1166,@xmath1167 and @xmath1180 .
thus , we obtain a general member @xmath1291 of @xmath499 .
the arithmetic genera , necessary to calculate the contribution of @xmath1291 to @xmath1144 , are given from right to left by : @xmath1292 as in the proof of prop .
[ bogomolov1 ] , we substitute these in the sum @xmath1144 , and for @xmath1293 we obtain @xmath1294 combining all of the above results , we conclude that the contributions of any tree @xmath902 to the sums @xmath1144 and @xmath1217 are @xmath1295 this allows us to extend all results of sect .
[ bogomolov1][index1 ] to the case of a base curve @xmath6 meeting transversally the boundary @xmath488 . if the base curve @xmath6 happens to be _ tangent _ to a boundary divisor @xmath1255 at a point @xmath1296 $ ] , then over some node @xmath335 of the corresponding tree @xmath1261 _ all _ local analytic multiplicities @xmath559 ( cf .
[ definition ] ) will be multiplied by the degree of tangency of @xmath6 and @xmath1255 .
[ local multiplicities ] presents a few examples of possible fibers in @xmath247 : @xmath1297 in the nonramification cases of @xmath1298 and @xmath1299 , this would force rational double points as singularities on the total spaces of @xmath249 and @xmath55 , whereas in the ramification cases of @xmath1300 and @xmath566 , one may arrive at surfaces @xmath249 and @xmath55 , nonnormal over some nonreduced fibers .
but in both cases , one can roughly view the corresponding fibers as being obtained by a base change from the general or special fibers of sect .
8 and sect . 11.1 .
alternatively , one can go through the arguments of the paper for the new surfaces @xmath249 and @xmath55 ( normalizing , if necessary ) , and notice that all formulas ( e.g. euler characteristic formula for @xmath4 , index theorem on @xmath249 , adjunction formula in @xmath254 , etc . )
hold for surfaces with double point singularities .
thus , in effect , one may replace a given singular fiber @xmath1251 by a bunch of general boundary curves @xmath230 , following the procedure described in section [ nontangentb ] .
furthermore , if some of these general curves @xmath230 are `` multiple '' ( i.e. @xmath6 is tangent to @xmath1255 at @xmath172 $ ] ) , one may in turn replace each @xmath230 by several `` transversal '' general boundary curves , and refer to the statements in sections 8.3 and 9.3 .
the only notational difference in this approach will appear in the definition of the invariants @xmath1094 and @xmath1095 from sect . 8 :
now we have to allow for them to be _ rational _ , rather than integral , due to possible rational intersections @xmath1301 .
this will be `` compensated '' in the final calculations , which will take into account the multiplicity of the corresponding fibers , and roughly speaking , will `` multiply back '' our invariants @xmath1302 and @xmath109 by what they were divided by in the beginning of the calculations .
this concludes the proof of our results for all families of stable curves @xmath1 with general smooth trigonal members . in the following list of results ,
theorems [ 7 + 6/g relation2 ] and [ maximal relation2 ] can be viewed as local trigonal analogs of cornalba - harris s relation in the picard group of the hyperelliptic locus @xmath16 ( cf .
theorem [ chpic ] ) .
similarly , theorem [ maximal bound2 ] is the analog of the @xmath27 maximal bound in the hyperelliptic case ( cf . theorem [ chx ] ) . for any family @xmath1 of stable curves with smooth trigonal general member ,
if @xmath8 is the canonically associated to @xmath7 vector bundle of rank 2 , then the following relation holds true @xmath1303 where @xmath76 is an effective rational linear combination of boundary components of @xmath17 , not containing @xmath277 .
in particular , @xmath1304 where @xmath60 is quadratic polynomial in @xmath61 with linear coefficients in @xmath11 , and it is determined by the geometry of @xmath58 ( cf .
[ 7 + 6/g relation2 ] for any nonisotrivial family @xmath1 of stable curves with smooth trigonal general member , if the canonically associated to @xmath7 vector bundle @xmath8 is bogomolov semistable , then the slope of @xmath25 is bounded from above by @xmath1305 [ 7 + 6/g bogomolov2 ] for any family @xmath1 of stable curves with smooth trigonal general member , if @xmath8 is the canonically associated to @xmath7 vector bundle of rank 2 , then the following relation holds true @xmath1306 where @xmath300 is an effective rational combination of the boundary divisors @xmath275 , not containing @xmath277 .
in particular , @xmath1307 where @xmath1308 is quadratic polynomial in @xmath61 with linear coefficients in @xmath11 , and it is determined by the geometry of @xmath58 ( cf .
[ maximal relation2 ] for any nonisotrivial family @xmath1 of stable curves with smooth trigonal general member , the slope of @xmath25 satisfies : @xmath1228 with equality if and only all fibers of @xmath7 are irreducible curves , and either @xmath958 or the divisor @xmath40 on the total space of @xmath1229 is numerically zero .
[ maximal bound2 ] [ list of theorems ] as we promised in section [ hyperelliptic locus ] , we consider the contribution of the hyperelliptic locus to the above theorems . for any hyperelliptic curve @xmath230
, we need to blow up a point on @xmath230 before it starts `` behaving '' like a trigonal curve in the quasi - admissible and effective covers .
below we have shown what happens to a smooth or general singular hyperelliptic curve ( cf .
[ hyperboundary ] for the admissible classification of the boundary locus @xmath1309 ) .
@xmath1310 we blow up @xmath230 at a point , and thus add a smooth rational component @xmath517 to make it a triple cover @xmath587 ( cf .
[ smoothhyper ] ) .
the quasi - admissible adjustment of @xmath230 is @xmath1311 . from here on
, @xmath230 will behave essentially like a smooth trigonal curve .
therefore , in all relations @xmath230 is going to contribute @xmath11 or @xmath1312 , depending on what @xmath3 is multiplied by .
the necessary effective and quasi - admissible modifications are shown in fig .
[ singularhyper]47 . in the first case ,
there are two hyperelliptic components intersecting transversally in two points . for the quasi - admissible cover ,
we need two `` smooth '' blow - ups , which makes @xmath1314 . from now on ,
this curve will behave like a element of @xmath568 , where @xmath1315 . thus , the coefficient in , say , the maximal bound relation will be : @xmath1316 , due to the extra blow - up in @xmath67 .
@xmath1317 in the second case , two hyperelliptic components meet transversally in one point , but have a ramification index 1 at this point when viewed as double covers .
fig.46 presents first the quasi - admissible modification : as in the case of @xmath1313 , the local analytic multiplicity between the two rational components is @xmath189 , which means that we must have made three `` smooth '' blow - ups and one `` singular '' blow - down . as a result , @xmath1318 . from here on , this curve behaves exactly as a general member of @xmath1313 .
recall that @xmath1319 , and the extra @xmath219 in the hyperelliptic case accounts for the one extra blow - up .
therefore , the coefficient of this fiber @xmath230 , say , in the maximal bound relation , will be @xmath1320 . we conclude that a base curve @xmath6 , passing through the hyperelliptic locus , will contribute in the results listed in section [ results ] roughly @xmath11 , or @xmath1312 , times the number of elements in @xmath1321 .
we can not write the latter in the form of a scheme - theoretic intersection , since @xmath16 is of a larger codimension in @xmath17 .
[ hypercalculations ] one can explain these extra summands in the expressions for @xmath4 in the following way .
recall the projection map @xmath1322 .
the exceptional locus of @xmath511 is the admissible boundary divisor @xmath548 , which is blown down to the codimension 2 hyperelliptic locus @xmath16 inside @xmath17 . for calculation purposes
, it will be easier to work instead with the space of minimal quasi - admissible covers @xmath1323 , which replaces @xmath510 .
the same situation of a blow - down occurs , where the exceptional divisor in @xmath1323 consists of reducible curves @xmath587 , as shown in fig .
[ smoothhyper ] .
@xmath1324 let @xmath819 be the linear combination of divisors in @xmath17 given by the restriction @xmath308 , and consider a curve @xmath1325 , intersecting the hyperelliptic locus in finitely many points . by abuse of notation , we denote by @xmath511 the projection from @xmath1323 to @xmath17 .
then for the intersection @xmath1326 we have : @xmath1327 where @xmath1328 is the proper transform of @xmath6 , and the @xmath1329 s are the corresponding exceptional curves above @xmath6 .
note that each @xmath1329 is in fact a line @xmath517 representing all possible quasi - admissible covers , arising from a hyperelliptic curve @xmath172\in b\cap \overline{\mathfrak{i}}_g$ ] . from fig .
[ smoothhyper ] , these are the blow - ups of @xmath230 at a point , one for each involution pair @xmath1330 , and that is why @xmath1331
. the extra summands on p. , induced by the base curve @xmath6 , are result of the extra intersections @xmath1332 from above . indeed , the relations , as they stand ,
compute only @xmath1333 , the component corresponding to families with general smooth members . from the calculations on p.
, we expect that each @xmath1334 , and this will account for the extra @xmath219 apprearing in all @xmath67 s . to verify this , we only needs to show @xmath1335 . since we can not pick out canonically one point @xmath454 from each hyperelliptic pair @xmath1336 on @xmath230 , and
thus construct a family of blow - ups at @xmath454 of @xmath230 over @xmath1337 , we make a base change of degree two @xmath1338 .
@xmath1339 we construct a family over @xmath230 , corresponding to _ all _ blow - ups of @xmath230 at point @xmath1340 .
this is simply the products @xmath1341 and @xmath1342 , identified at two sections @xmath1343 : @xmath1344 is the diagonal on @xmath1341 , and @xmath1345 is a trivial section of @xmath1342 over @xmath230 ( cf .
[ deltahyper.fig ] ) . from @xcite , for the base curve @xmath230 of this family , the degree @xmath1346 is computed as @xmath1347 taking into account the base change @xmath1338 , @xmath1335 . finally , if we allow for our families to have finitely many hyperelliptic fibers , we adjust the relation in [ 7 + 6/g relation2 ] by @xmath1348 , and the relation in [ maximal relation2 ] by @xmath1349 .
the two bounds in theorems [ 7 + 6/g bogomolov2]-[maximal bound2 ] are unaffected by the above discussion .
[ bog - maroni ] for any smooth trigonal curve @xmath230 , consider the triple cover @xmath1350 .
the pushforward @xmath1351 , as we noted before , is a locally free sheaf of rank 3 on @xmath252 , and hence decomposes into a direct sum of three invertible sheaves on @xmath518 : @xmath1352 the first summand is trivial due to the split exact sequence @xmath1353 where @xmath1354 . from grr ,
we have observed in section 6 that @xmath230 embeds in the rational ruled surface @xmath1356 , for @xmath1357 . * definition 12.1 . * the _ maroni invariant _ of an irreducible trigonal curve @xmath230 is the difference @xmath1358 .
the _ maroni locus _ in @xmath17 is the closure of the set of curves with maroni invariants @xmath396 ( cf .
[ ma ] ) . for a general trigonal curve @xmath230 the vector bundle @xmath8 is _ balanced _ , i.e. the integers @xmath80 and @xmath81 are equal or 1 apart according to @xmath1359 .
[ gentrig ] let @xmath1360 .
the statement follows from a dimension count of the linear system @xmath1361 on the ruled surface @xmath1362 .
indeed , all trigonal curves with maroni invariant @xmath1363 are elements of @xmath125 . if @xmath1364 is the projection map , the projective dimension of @xmath125 equals @xmath1365 denoting by @xmath1366 the section of @xmath231 with smallest self - intersection of @xmath1367 , we have @xmath1368 . the necessary pushforward from above is : @xmath1369 since an irreducible trigonal curve @xmath230 lies in @xmath125 , we have @xmath1370 , hence @xmath1371 and @xmath1372 . evaluating the sections of this sum of sheaves , we obtain @xmath1373 the ruled surface @xmath231 has automorphisms , inducing automorphisms of the linear system @xmath125 .
we need to mod out these in order to obtain the dimension of the space of trigonal curves embedded in @xmath231 .
the group @xmath1374 is a product ( not necessarily direct ) of the base automorphisms @xmath1375 , and the projective automorphisms of the vector bundle @xmath8 .
the latter is an open set ( up to projectivity ) of the homomorphisms of @xmath8 into @xmath8 , and hence has the same dimension as : @xmath1376 for @xmath1377 , @xmath1378 , while for @xmath1379 , @xmath1380 .
we conclude that the dimension of the set of trigonal curves with maroni invariant @xmath233 is @xmath1381 when @xmath1379 or @xmath1382 , this space corresponds to an open dense set of @xmath17 . for
an even @xmath11 a general trigonal curve has maroni invariant @xmath951 and therefore embeds in @xmath1383 , while for an odd @xmath11 a general trigonal curve has maroni invariant @xmath219 and embeds in @xmath1384 . in both cases ,
the vector bundle @xmath8 is balanced .
for @xmath11 even , the maroni locus is a divisor in @xmath17 whose general member embeds in @xmath1385 . for @xmath11 odd , the maroni locus has codimension 2 in @xmath17 and its general member embeds in @xmath1386 .
[ maronilocus ] * remark 12.1 . * it will be useful to identify precisely the group of authomorphisms of the linear system @xmath125 for @xmath1387 .
we have @xmath1388 .
the last factor comes from the exchange of the fiber and the base of @xmath1389 and it is relevant only for @xmath228 : then @xmath1390 .
otherwise , for any even @xmath1391 : @xmath1392 when @xmath11 is odd , the ruled surface @xmath52 can be thought of as the blow - up of @xmath1393 at the point @xmath1394 $ ] .
any automorphism of @xmath1395 carries the exceptional divisor @xmath1396 of @xmath1397 to itself , and hence is induced by an automorphism of the plane preserving the point @xmath96 .
the group of such automorphisms of @xmath1393 is the subgroup of @xmath1398 corresponding to matrices : @xmath1399 taking into account the discriminant of these matrices , we easily identify for odd @xmath11 : @xmath1400 note that all of the above groups @xmath1401 have dimension @xmath1402 , which was claimed already in lemma [ gentrig ] .
the rational picard group of @xmath17 , @xmath69 , is freely generated by the boundary classes @xmath56 , @xmath58 , and one additional class , which for even genus @xmath11 coincides with the maroni class @xmath67 . [ genpic ] since a general trigonal curve @xmath230 embeds in the ruled surface @xmath231 ( @xmath1387 ) , @xmath230 is a member of the linear system @xmath1404 on @xmath231 .
let @xmath1405 be the open set inside @xmath1406 corresponding to the _ smooth trigonal _ members of @xmath125 .
the surjection @xmath1407 has a nontrivial kernel , because the set of singular trigonal curves in @xmath231 is a divisor in @xmath1408 .
hence @xmath1409 for some integer @xmath1410 , and @xmath1411 . the image of the natural projection map @xmath1412 is the open dense set @xmath974 of smooth trigonal curves with lowest marone invariant of @xmath951 or @xmath219 .
let @xmath1086 denote the fiber of @xmath335 . from remark 12.1 , @xmath1413 leray spectral sequence or other methods ( cf .
@xcite ) yield : @xmath1414 pushing the exponential sequence on @xmath1405 to @xmath974 : @xmath1415 combining with the exponential sequence on @xmath974 : @xmath1416 with @xmath1417 . for even @xmath11 , @xmath1418 is torsion ( a direct sum of copies of @xmath1419 ) , but for odd @xmath11
it is isomorphic to @xmath1420 .
hence , for even @xmath11 we have the natural embedding @xmath1421 , and in view of @xmath1422 , it follows that @xmath1423 .
the complement of @xmath974 in @xmath17 is the union of the boundary of @xmath17 and the maroni divisor .
therefore , @xmath56 , @xmath58 and @xmath67 generate @xmath1424 .
since the class of the hodge bundle @xmath4 is _ not _ a linear combination of the boundary classes ( cf .
p. ) , the boundary divisors are _ not _ sufficient to generate the rational picard group of @xmath17 , and @xmath67 must be linearly independent of them .
we conclude that @xmath56 , @xmath58 , and @xmath67 generate freely @xmath1424 for even genus @xmath11 . for @xmath11-odd ,
@xmath1425 is either an inclusion , or has a kernel with one generator .
since the maroni locus for @xmath11-odd is not a divisor , an inclusion would imply as above that @xmath4 is a linear combination of the boundary classes , which is not true .
hence , @xmath1426 and @xmath1427 is generated freely by the boundary classes @xmath56 and @xmath58 , and one additional class .
for even genus @xmath11 and a base curve @xmath6 , not contained in @xmath488 : @xmath1428 where @xmath289 are certain polynomial coefficients computed similarly as @xmath60 .
p. ) [ maroni2 ]
we set @xmath1429 .
let us consider for now only families with irreducible trigonal fibers , i.e. the base curve @xmath6 intersects only the boundary component @xmath277 .
_ if @xmath6 does not intersect the maroni divisor @xmath67 , then the maroni invariant of the fibers in @xmath7 is constant , and equal to @xmath951 .
the fibers @xmath230 of @xmath7 embed in the projectivization @xmath1430 .
since @xmath1431 and @xmath8 is balanced , the restriction of @xmath8 to the fiber @xmath800 on the ruled surface @xmath39 is @xmath1432 moreover , @xmath1433 does not jump as @xmath800 moves , so that @xmath8 can be written as : @xmath1434 for some vector bundle @xmath739 of rank 2 on @xmath6 .
but the bogomolov index @xmath68 is independent of twisting @xmath8 by line bundles , in particular , by @xmath1435 , so that @xmath1436 the last equality follows from @xmath1437 for any bundle on the curve @xmath6 .
we conclude that @xmath1438 .
@xmath1439 _ case 2 .
_ now let @xmath6 intersect the maroni divisor @xmath67 in _ finitely _ many points .
assume also that these points are _ general _ in @xmath67 , i.e. they correspond to trigonal curves @xmath230 embeddable in the ruled surface @xmath1385 .
we twist @xmath8 by a line bundle @xmath1440 , and set @xmath1441 , so that @xmath1442 and @xmath1443 then @xmath1444 is the middle term of a short exact sequence on @xmath39 @xmath1445 where @xmath1446 is a vector bundle of rank 2 on @xmath39 . in the notation of @xcite ,
let @xmath974 be the sum of the special fibers of @xmath39 , and let @xmath901 be the union of certain isolated points on each member of @xmath974 , so that @xmath1447 is the ideal sheaf of @xmath901 inside @xmath974 .
note that the number of the special fibers , which comprise @xmath974 , equals @xmath1448 .
we can now compute the chern classes of @xmath1444 : @xmath1449 the last equality follows from the fact that @xmath1450 is the pull - back of a bundle on the curve @xmath6 , hence of zero higher chern classes .
we conclude that @xmath1451 , and @xmath1452 putting the above two cases together , we have for any family with irreducible trigonal members , not entirely contained in the maroni locus : @xmath1453 prop .
[ genpic ] then implies that @xmath4 is a linear combination of the boundary and the maroni class : @xmath1454 where the coefficients @xmath289 are computed in a similar way , or by direct computation with families of singular trigonal curves ( cf .
@xcite ) . we can combine the above results in the following for even @xmath11 , @xmath69 is freely generated by all boundary classes @xmath56 and @xmath58 , and the maroni class @xmath67 .
the class of the hodge bundle on @xmath17 is expressed in terms of these generators as the following linear combination : @xmath288 [ pic trigonal ] * remark 12.2 . *
note that the coefficients @xmath289 depend on the specific decriptions of the maroni curves that appear in the boundary divisors @xmath275 , and they are _ not _ always equal to the corresponding coefficients @xmath60 in theorem [ 7 + 6/g relation2 ] .
indeed , in the above proposition , we have shown that @xmath1455 for some @xmath1258 , which may be non - zero .
hence , @xmath1456 .
for example , consider the case of @xmath63 , and let @xmath490 be a general member of it .
if @xmath230 is also maroni , then there exists a family @xmath1 , whose general fiber is an irreducible maroni curve , and one of whose special fibers is our @xmath230 .
we can assume , modulo a base change and certain blow - ups not affecting @xmath230 , that this family fits in the basic construction diagram ( cf .
[ general b ] ) .
let @xmath1457 and @xmath1458 be the two ruled surfaces in which @xmath491 and @xmath492 are embedded , and let @xmath1166 and @xmath1167 be the projections of @xmath491 and @xmath492 in the birationally ruled surface @xmath55 .
then @xmath1459 is a special fiber of @xmath55 , with self - intersections @xmath1460 .
now , the general member of @xmath7 , being maroni , is embedded in a ruled surface @xmath1461 with a section @xmath125 of self - intersection @xmath1171 .
the union of such @xmath125 s forms a surface in the 3-fold @xmath1462 , whose closure we denote by @xmath128 .
evidently , @xmath1463 , at least outside their special fibers .
let @xmath128 intersect @xmath1457 and @xmath1458 in curves @xmath1464 and @xmath1465 ( over @xmath1166 and @xmath1167 ) .
we claim that at least one of @xmath1457 and @xmath1458 is _ not _ isomorphic to @xmath1466
. it will suffice to show that @xmath1464 or @xmath1465 has negative self - intersection .
indeed , suppose to the contrary that @xmath1467 in @xmath1468 ( @xmath1469 ) .
note that @xmath1470 in @xmath1462 , so that @xmath1471 on the other hand , @xmath1472 is the fiber of the projection @xmath1473 , and as such it is linearly equivalent to the general fiber @xmath1461 .
hence @xmath1474 a contradiction .
we conclude that if @xmath490 is a maroni curve of boundary type @xmath63 , then either @xmath491 or @xmath492 ( or both ) is embedded in a ruled surface @xmath1475 with @xmath1476 .
this already distinguishes the cases of odd and even genus @xmath61 . when @xmath1477 is even ( and hence @xmath1478 is also even ) , the general member of @xmath63 is embedded in a join of two @xmath1479 s ( each @xmath1480 ) , and hence it is _ not _ maroni .
based on this observation , one can easily find the coefficient @xmath1481 for @xmath61-even . to do this ,
consider the birationally ruled surface @xmath39 which is the blow - up of @xmath1479 at one point .
let again the two components of the special fiber of @xmath39 be @xmath1166 and @xmath1167 , and projectivize the trivial vector bundle @xmath1482 : @xmath1483 . by taking an appropriate linear system in @xmath254 ,
one obtains a family of trigonal curves @xmath7 , whose fibers are all irreducible and embedded in @xmath1479 , except for a special reducible curve @xmath230 over @xmath1484 of the specified above type .
hence none of @xmath7 s members are maroni , and so @xmath1485 .
further , @xmath1486 , and @xmath1487 , so that equation ( [ alpha - coef ] ) implies @xmath1488 , and hence @xmath1489 for @xmath61-even .
the situation is quite different when the genus @xmath61 is odd .
then both components of the general member @xmath230 of @xmath63 are embedded in @xmath1397 s , and hence @xmath230 is potentially maroni .
one can take further the above general argument of intersection theory on @xmath254 , and show that the curves @xmath1464 and @xmath1465 are in fact both sections of negative self - intersection @xmath1490 in these @xmath1397 s : consider the product @xmath1491 and its variation over the special fiber of @xmath55 .
but we know that @xmath1464 and @xmath1465 intersect , as the fiber of @xmath128 over @xmath55 is connected
. thus , the curve @xmath230 would be maroni if and only if the two corresponding ruled surfaces @xmath1397 are glued along one of their fibers so that their negative sections intersect on that fiber .
( this decsription can be alternatively derived by considering the degenerations of the @xmath35 s on the irreducible maroni curves . ) to find @xmath1481 in this case , we construct a similar example as above , only changing @xmath8 to @xmath1492 .
this , while keeping the general fiber embedded in @xmath1479 , has the effect of embedding the special one in a `` maroni '' gluing of two @xmath1397 s .
we have @xmath1493 , @xmath1494 , and @xmath1487 , so that equation ( [ alpha - coef ] ) implies @xmath1495 for @xmath61-odd , and hence @xmath1496 .
@xmath1497 one can similarly compute the remaining coefficients @xmath1258 , by first figuring out which boundary curves in @xmath1498 are maroni , then constructing an appropriate vector bundle @xmath8 , and finally using equation ( [ alpha - coef ] ) to compute @xmath1258 , and hence @xmath289 . for @xmath11-even ,
if the base curve @xmath6 is not entirely contained in the maroni divisor , and the singular members of @xmath7 belong only to @xmath1499 , then the slope of the family @xmath1500 satisfies : @xmath1501 [ maroni inequality ] for @xmath11-even , if the base curve @xmath6 is not entirely contained in the maroni divisor , then the slope of the family @xmath1500 satisfies : @xmath1501 [ maroni - conj ] even though for odd genus @xmath11 the maroni locus is not large enough to be a divisor in @xmath17 , we can define a _ generalized maroni
_ divisor class by extending the relation from the @xmath11-even case .
* definition 12.2 . * for any genus , we define the _ generalized maroni _ class @xmath67 in @xmath69 by @xmath1502 the maximal bound @xmath309 is attained for a trigonal family of curves @xmath1 if and only all fibers of @xmath7 are irreducible and @xmath1503 [ maximalmaroni ] the fact that @xmath7 must have only irreducible fibers in order to attain the maximum bound is already known from theorem [ genmaximal ] .
this means @xmath1504 for all @xmath1505 .
then , theorem [ bogomolov1 ] implies : @xmath1506 assume that the maximal bound is attained , i.e. @xmath1507 . substituting for @xmath19 in the above equation , yields the desired equality .
the converse follows similarly .
* remark 12.3 . * in the @xmath11-even case , this equality has a specific meaning .
since the maroni class @xmath67 corresponds to an effective divisor on @xmath17 , the equality ( and hence the maximal bound ) is achieved only for base curves @xmath6 entirely contained in the maroni divisor , so that the restriction @xmath1508 can be negative .
in fact , in all found examples , the base @xmath6 is contained in a very small subloci of the maroni loci , defined by the highest possible maroni invariant .
* remark 12.4 .
* theorem [ pic trigonal ] and prop . [ maximalmaroni ] do not have analogs in the hyperelliptic case : there is no additional maroni divisor to generate @xmath1509 together with the boundary @xmath1309 .
* remark 12.5 .
* when @xmath958 , there is no maroni locus in @xmath1510 either . indeed ,
since an irreducible trigonal curve of genus @xmath520 embeds only in ruled surfaces @xmath231 with @xmath30-odd and @xmath1511 , then _ all _ irreducible trigonal curves embed in @xmath52 , and correspondingly they all have the lowest possible maroni invariant @xmath1382 .
however , @xmath1512 is not generated by the boundary classes of @xmath1510 : as prop .
[ genpic ] asserts , in the odd genus case there is always one additional generating class . on the other hand , the results on p. yield apriori _ two _ relations among @xmath4 and the @xmath58 s
this would have been a contradiction to the _ freeness _ of the generators above , unless these two relations are the same .
this is in fact what happens : @xmath1513 as restricted to any base curve @xmath1514 .
note the convenient disappearance of the `` extra '' @xmath1515summands in the coefficients of @xmath1516 ) .
then the maximal and the semistable ratios both equal @xmath1517 , and are attained for families with irreducible trigonal members .
[ furtherresults ] i have carried out some preliminary research in the @xmath87-gonal case , and while the methods and ideas for the trigonal case are in principle extendable , this appears to be a substantially more subtle and complex problem .
more precisely , let @xmath1519 be the closure in @xmath5 of the stable curves expressible as @xmath87-sheeted covers of @xmath31 .
one possible goal is to complete the program of describing generators and relations for the rational picard groups @xmath1520 , and to find the exact maximal bounds for the slopes of @xmath87-gonal families .
for example , i have obtained the following bound for the slope of a general tetragonal family with smooth general member ( for odd genus @xmath11 ) : @xmath1521 i have also conjectured formulas for the maximal and general bounds for any @xmath87-gonal and other families of stable curves . entering these formulas are the _ clifford index _ of curves , _ bogomolov semistability _
conditions for higher rank bundles , and some new geometrically described loci in @xmath1519 .
generalizing the idea of the maroni locus in the trigonal case , these loci are characterized , for example , in the tetragonal case by the dimensions of the multiples of the @xmath1522-series . in particular
, there will be another generator of @xmath1523 besides the boundary and maroni divisors . in the following i present some of these conjectures on the upper bounds for @xmath1519 .
we start by comparing all known maximal and general bounds functions of the genus @xmath11 : @xmath1524 the pattern appearing in this table is clear : the general bound @xmath1525 coincides with each of the other bounds exactly twice for some special values of the genus @xmath11 .
evidently , @xmath1526 is one of these special values , yielding 12 everywhere .
( i owe this observation to benedict gross . )
let @xmath1527 be the other genus @xmath11 for which the general formula in @xmath5 and the maximal formula for @xmath1519 coincide , i.e. @xmath1528 , @xmath1529 , @xmath1530 .
we notice that for these genera @xmath1527 the moduli spaces @xmath1531 and @xmath1532 consist only of hyperelliptic , trigonal or tetragonal curves , respectively . in general , _
brill - noether _ theory ( cf .
@xcite ) asserts that for complete linear series @xmath1533 the expected dimension of the variety of @xmath401 s on a smooth curve of genus @xmath11 is @xmath1534 and hence the smallest genus @xmath11 for which @xmath1535 is @xmath1536 .
thus we set @xmath1537 for @xmath1538 and @xmath1528 .
note that this coincides with the previously found @xmath1529 and @xmath1530 .
if @xmath1539 is an exact upper bound for the slopes of families of stable curves with smooth @xmath87-gonal general member ( locus @xmath1519 ) , then @xmath1540 [ conj2 ] it is reasonable to expect that the upper bounds for @xmath1519 will be ratios of linear functions of the genus @xmath11 : @xmath1541 .
conjecture [ conj2 ] then estimates the difference between @xmath1539 and the general bound for @xmath5 up to a factor @xmath1542 .
the exact upper bounds @xmath1539 are given by @xmath1543 or equivalently , @xmath1544 i have a conjecture on how to determine the remaining factor @xmath1545 , which seems to be closely related to the coefficients of the linear expression in [ emh ] for the divisor @xmath1546 in terms of the hodge bundle @xmath4 and the boundary classes @xmath43 on @xmath5 .
these conjectures are supported by the work of cornalba - harris on the hyperelliptic locus @xmath1547 , by the results of this paper on the trigonal locus @xmath1548 , and by partial results on the tetragonal locus @xmath1549 . in view of remark 12.5
, the equality between the maximal and semistable trigonal bounds for @xmath958 suggests that a similar situation might occur for other @xmath87-gonal families .
it is reasonable to expect two or more `` semistable '' bounds , depending on the number of extra generators in @xmath1550 .
one of these `` semistable '' bounds relates to families obtained as blow - ups of pencils of @xmath87-gonal curves on a ruled surface @xmath231 .
example 2.1 yields the maximal bound @xmath27 for hyperelliptic families ( no extra generator besides the boundary classes ) , and a similar example in the trigonal case yields the @xmath221 semistable bound ( one extra generator , the maroni locus ) .
we generalize this to any @xmath87-gonal family of curves embedded in an arbitrary ruled surface @xmath231 .
invariably , the slope of @xmath25 is : @xmath1551 let @xmath7 be a family of @xmath87-gonal curves of genus @xmath11 whose base @xmath6 is not contained in a certain codimension 1 closed subset of @xmath1552 .
then the slope of @xmath25 satisfies : @xmath1553 [ clifford ] conjectures [ clifford]4 are modifications of earlier conjectures of joe harris . the discussion so far was primarily concerned with the loci @xmath1556 corresponding to linear series @xmath401 . but all of our problems are well - defined and quite interesting to solve for curves with series @xmath1554 of dimension @xmath1557 .
equivalently , we consider the loci @xmath1558 of curves mapping with degree @xmath87 to @xmath1559 , @xmath1560 . * definition 13.1 . *
the _ clifford index _
@xmath1561 of a smooth curve @xmath230 is defined as @xmath1562 where @xmath125 runs over all effective special linear series @xmath1563 on @xmath230 .
clifford s theorem implies @xmath1564 , with equality if and only if @xmath230 is hyperelliptic , i.e. @xmath1565 ( cf .
@xcite ) . on the other hand
, @xmath1566 means that there exists a @xmath1554 on @xmath230 with @xmath1567 . from marten s theorem , @xmath1568 , where @xmath1569 is the variety parametrizing complete linear series on @xmath230 of degree @xmath87 and dimension at least @xmath368 .
therefore , we must have @xmath1570 . but then mumford s theorem asserts that @xmath230 is either trigonal , or bi - elliptic , or a smooth plane quintic .
the bi - elliptic case would mean that @xmath1569 consists of @xmath1571 s , which contradicts the dimension of @xmath1572 .
in short , @xmath1573 if and only if @xmath230 is not hyperelliptic and possesses a @xmath35 or a @xmath1574 .
thus , according to the clifford index , the first case with @xmath1555 is the space of plane quintics .
consider a general pencil of such , and blow up the plane at its 25 base points .
the resulting family @xmath1575 is easily seen to have slope @xmath1576 , which corresponds to the bound in conjecture [ clifford ] with @xmath1577 replaced by the clifford index @xmath1573 . finally , note that for a @xmath87-gonal curve @xmath230 of genus @xmath11 , by definition @xmath1578 , so that when @xmath1579 we may generalize to : for a general family @xmath1 of genus @xmath11 stable curves whose general member has clifford index @xmath1580 and whose base @xmath6 is a general curve in @xmath1581 , the slope of @xmath25 satisfies : @xmath1582 * remark 13.1 .
* it is worth noting that the stratification of @xmath5 , for which we asked in the introduction , is not obtained via the clifford index @xmath1580 .
for example , xiao constructs families of bi elliptic curves @xmath230 with slope @xmath1583 ( cf .
@xcite ) , which is between the hyperelliptic and the trigonal maximal bounds . since @xmath230 has a @xmath1522 as bi
elliptic , this already exceeds the conjectured maximal bounds for the tetragonal case .
this shows that in some of the above conjectures we have to exclude the subset of bi elliptic curves from the tetragonal locus @xmath1584 , and that similar modifications might be necessary for the other loci @xmath1552 .
more precisely , it seems plausible that the stratification of @xmath5 according to successively lower slope bounds is related not just to the existence of a specific linear series @xmath1554 , but also to the number , dimension and description of the irreducible components of corresponding varieties @xmath1569 .
the approach in the @xmath401-cases is based on a modification of the harris - mumford s [ ehm ] _ hurwitz scheme of admissible covers _ , which parametrized the @xmath87-uple covers of stable pointed rational curves .
however , in the more general situation for linear series with larger dimensions @xmath1557 , such a compactification via admissible covers does not exist , so we have to look for a different solution .
consider moduli spaces of stable maps @xmath1585 .
they parametrize _ stable _ maps @xmath1586 , where @xmath230 is a projective , connected nodal curve of arithmetic genus @xmath11 , the @xmath454 s are marked distinct nonsingular points on @xmath230 , and the map @xmath1587 has image @xmath1588)=d[{\operatorname}{line}]$ ] and satisfies certain stability conditions ( cf .
the space @xmath1585 seems to be the right compactification which we need in order to extend our results to families with @xmath1554-series on the fibers : the moduli space of stable maps is somewhat more `` sensitive '' in describing our loci @xmath1558 in terms of their geometry .
going back to the @xmath401-problems , one can also see the combinatorial flavor that stands in the background of these questions .
it is probably not coincidental that the spaces @xmath1585 are also combinatorially defined and give rise to many enumerative problems
. it will be useful to understand better the loci @xmath1558 via their connection with the kontsevich spaces @xmath1585 , and ultimately to solve the remaining questions on @xmath1589 for any @xmath1590 , as well as related interesting enumerative problems that will inevitably arise from such considerations .
[ hyperelliptic ] in this section we give a proof of theorems [ theoremchpic ] and [ chx ] , following the same ideas and methods as in the trigonal case .
we refer the reader to previous sections for a detailed proof of certain statements .
cornalba - harris describe the boundary of @xmath16 as consisting of several boundary components , whose general members and indexing are shown in fig .
[ hyperboundary ] ( cf .
the restriction of the divisor class @xmath3 to @xmath16 is the following linear combination : @xmath1591}\xi_i+ \sum_{j=1}^{\left[g/2\right]}\delta_j , \label{boundaryrel}\ ] ] where @xmath1592 and @xmath43 are the classes in @xmath1593 of the boundary divisors @xmath186 and @xmath1594 . in the case of a hyperelliptic family @xmath89 ,
a minimal quasi - admissible cover coincides with the original family @xmath7 , because no blow - ups are necessary to perform on the fibers of @xmath7 : these are already quasi - admissible double covers .
thus , we have a degree 2 map @xmath1597 for some birationally ruled surface @xmath39 over @xmath6 . as for an effective cover @xmath255
, only the boundary divisors @xmath44 require blow - ups ( cf .
[ hyperboundary ] ) .
this is analogous to the `` ramification index 1 '' discussion in fig .
[ ram][resolve1 ] . thus , while in @xmath249 the special fibers may have occasional nonreduced rational components of multiplicity 2 , the fibers of @xmath55 are always trees of reduced smooth @xmath31 s . in the case of a smooth hyperelliptic curve @xmath230
, we consider the natural double sheeted map @xmath1598 .
the pushforward @xmath1599 is a rank 2 vector bundle on @xmath31 , which fits into the short exact sequence @xmath1600 we can embed @xmath230 in the rational ruled surface @xmath1601 .
we generalize this construction to the effective cover @xmath255 by setting @xmath1602 .
for some line bundle @xmath795 on @xmath55 : @xmath1603 then @xmath249 naturally embeds in the threefold @xmath238 .
let @xmath1604 be the corresponding projection map . as a divisor in @xmath238 , @xmath1605 , for some divisor @xmath819 on @xmath55 . from the adjunction formula , @xmath1606 , where @xmath1607
the arithmetic genus of the inverse image @xmath1277 is given by @xmath1608 it turns out that these are the only differences between the set - up of the hyperelliptic and the trigonal case .
the definitions of the functions @xmath1094 and @xmath1095 , as well as the formulas for @xmath1609 and the congruence @xmath1076 are valid without any modifications . as in the trigonal case
, it will be sufficient to consider only the cases when the base curve @xmath6 intersects _ transversally _ the boundary divisors of @xmath16 .
but then for all non - root components @xmath795 in @xmath55 : @xmath1610 we can now easily calculate the invariants on @xmath7 . with this , we are ready to show the linear relations among @xmath19 and the boundary restrictions @xmath1612 and @xmath1613 .
it is evident that in order to cancel the `` global '' term @xmath87 , one must subtract @xmath1614 , which is the main idea of the next theorem .
there exists an effective linear combination @xmath1615 of the boundary divisors of @xmath16 , not containing @xmath199 , such that for any family @xmath89 of hyperelliptic curves with smooth general member : @xmath1616 [ hyperrelation ] we consider the difference @xmath1617 in the hyperelliptic case , as opposed to the trigonal case , there is only _ one type _ of non - root components @xmath795 , namely , such that both @xmath795 and @xmath1055 are reduced .
that is why there is just one type of summands in @xmath1618 .
compare this with the contribution of @xmath1624 ( subsection [ contribution2 ] ) .
there are two non - root components @xmath1166 and @xmath1167 in the special fiber of @xmath55 ( @xmath1625 ) , whose invariants are @xmath1626 and @xmath1627 . with the ramification adjustment of @xmath11 , the contribution of @xmath1594 to the sum @xmath1618 is @xmath1628 where @xmath1629 $ ]
finally , for the appropriate indices @xmath61 and @xmath1630 we set @xmath1631 this is an effective combination of boundary divisors in @xmath16 , not containing @xmath190 by construction , and satisfying @xmath1632 .
let @xmath89 be a nonisotrivial family with smooth general member .
then the slope of the family satisfies : @xmath1633 equality holds if and only if the general fiber of @xmath92 is hyperelliptic , and all singular fibers are irreducible .
it is now straightforward to prove the fundamental relation in @xmath1593 , shown first in @xcite . in theorem
[ hyperrelation ] , we add to the coefficients @xmath1634 and @xmath1635 the corresponding multiplicities @xmath1636 and @xmath1637 : @xmath1638 using the fact that @xmath1593 is generated freely by the boundary classes @xmath1592 and @xmath1639 ( see @xcite ) , we obtain @xmath1640 in the picard group of the hyperelliptic locus , @xmath1593 , the class of the hodge bundle @xmath4 is expressible in terms of the boundary divisor classes of @xmath16 as : @xmath195}2(i+1)(g - i)\xi_i + \sum_{j=1}^{[g/2]}4j(g - j)\delta_j.\ ] ] [ chpic2 ] j. harris and d. mumford , _ on the kodaira dimension of the moduli space of curves , _ invent . math . * 67 * ( 1982 ) 23 - 86 . j. harris , _ on the kodaira dimension of the moduli space of curves , ii . the even genus case , _ invent
* 75 * ( 1984 ) 437 - 466 .
d. eisenbud and j. harris , _ the kodaira dimension of the moduli space of curves of genus @xmath1643,_invent . math .
* 90 * ( 1987 ) 359 - 387 . | we study the moduli of trigonal curves .
we establish the exact upper bound of @xmath0 for the slope of trigonal fibrations . here , the slope of any fibration @xmath1 of stable curves with smooth general member is the ratio @xmath2 of the restrictions of the boundary class @xmath3 and the hodge class @xmath4 on the moduli space @xmath5 to the base @xmath6 .
we associate to a trigonal family @xmath7 a canonical rank two vector bundle @xmath8 , and show that for bogomolov - semistable @xmath8 the slope satisfies the stronger inequality @xmath9 .
we further describe the rational picard group of the trigonal locus @xmath10 in the moduli space @xmath5 of genus @xmath11 curves . in the even genus case ,
we interpret the above bogomolov semistability condition in terms of the so - called maroni divisor in @xmath10 . |
the observational evidence for the acceleration of the universe demonstrates that canonical theories of gravitation and particle physics are incomplete , if not incorrect
. the next generation of astronomical facilities must therefore be able to carry out precision consistency tests of the standard cosmological model and search for definitive evidence of new physics beyond it .
codex @xcite is a spectrograph planned for the european extremely large telescope ( e - elt ) .
it should provide the first measurement of the cosmological redshift drift ( known as the sandage - loeb test @xcite ) ; a detailed feasibility study has been carried out by liske _
@xcite , and other aspects relevant for our work have been explored in @xcite .
another of its goals is an improved test of the stability of nature s fundamental couplings such as the fine - structure constant @xmath1 and the proton - to - electron mass ratio @xmath2 .
apart from the intrinsic importance of these measurements , they can be used ( under certain assumptions ) for detailed characterization of dark energy properties all the way up to redshift 4 .
this was suggested in @xcite ( see also @xcite for a related approach ) , and an assessment in the context of codex ( and its predecessor espresso ) can be found in @xcite .
we illustrate how codex can probe dark energy beyond the regime where it is dominating the universe s dynamics
deep in the matter era .
we introduce these two observational tools in sect .
ii , and discuss them in the context of two representative classes of models in sects .
iii - iv , highlighting their potential synergies .
our conclusions are in sect . v.
in realistic dynamical dark energy scenarios the ( presumed ) scalar field should be coupled to the rest of the model , unless one postulates a ( yet unknown ) symmetry to suppress these couplings .
the relevant coupling here is the one between the scalar field and electromagnetism , which we assume to be @xmath3 where the gauge kinetic function @xmath4 is linear , @xmath5 @xmath6 , and the coupling @xmath7 is related to equivalence principle violations .
local constraints are ( conservatively ) @xmath8 @xcite .
independent constraints can be obtained from the cosmic microwave background @xcite , and are currently about one order of magnitude weaker .
this form of @xmath4 can be seen as the first term of a taylor expansion , and given the tight low - redshift constraints on varying couplings and on equivalence principle violations it is a good approximation for the redshift range being considered .
the assumption here is that the dark energy and the varying @xmath1 are due to the same dynamical field , as in the case of nonminimally coupled quintessence models .
we will also assume a flat frw universe with @xmath9 , neglecting the radiation contribution since we are concerned with the low - redshift behavior .
the evolution of @xmath1 is given by @xmath10 and since the evolution of the scalar field can be expressed in terms of the dark energy properties @xmath11 and @xmath12 as @xcite @xmath13 ( where the prime denotes the derivative with respect to @xmath14 , @xmath15 being the scale factor ) we finally obtain the evolution of @xmath1 in this class of models @xmath16 as expected the magnitude of the variation is controlled by the strength of the coupling @xmath7 . the sandage - loeb test @xcite is a measurement of the evolution of the redshift drift of extragalactic objects , obtained by comparing quasar absorption spectra taken at different epochs . in any metric theory of gravity the redshift drift @xmath17 in a time interval @xmath18 , or equivalently
the spectroscopic velocity shift @xmath19 ( which is the directly measured quantity ) is @xmath20\,.\ ] ] this provides a direct measurement of the expansion history of the universe , with no model - dependent assumptions beyond those of homogeneity and isotropy .
a positive drift is a smoking gun for a dark energy component accelerating the universe ; a deccelerating universe produces a negative drift .
the lyman-@xmath1 forest ( and possibly other absorption lines , including metal ones ) is ideal for this measurement , but it can only be done at redshifts @xmath21 ( in what follows , we will assume measurements between @xmath22 and @xmath23 ) .
this applies to ground - based facilities ; measurements at lower redshift would be highly desirable ( since they would probe the dark energy dominated epoch ) , but they would need to be done from space , and there is currently no envisaged space - based spectrograph with the required resolution and stability .
_ @xcite have studied in detail the performance of the envisaged codex spectrograph , finding that the uncertainty in the spectroscopic velocity shift is expected to behave as @xmath24 where @xmath25 is the signal - to - noise of the spectra , and @xmath26 and @xmath27 and the number of the absorption systems and their respective redshifts .
this assumes photon - noise - limited observations and holds for @xmath28 ; beyond that the last exponent becomes @xmath29 . in our analysis
we will assume @xmath30 , 40 systems uniformly divided into 4 bins at @xmath31 and a time between observations of @xmath32 years .
suppose that the above assumption regarding varying @xmath1 does not hold : the dark energy is due to a cosmological constant ( with @xmath33 ) , and the variation of @xmath1 is due to some other field with a negligible contribution to the universe s energy density .
the bekenstein - sandvik - barrow - magueijo ( bsbm ) model @xcite is precisely of this type ( it has a varying @xmath1 field with an energy density that is no larger than that of radiation ) .
if one neglects the recent dark energy domination one can find an analytic solution for the behavior of @xmath1 @xmath34 where @xmath35 gives the magnitude of the variation .
this is sufficient for our purposes since we are mainly be interested in the matter - era behavior , but regardless of the bsbm motivation we can take this as a phenomenological parametrization . a logarithmic redshift dependence is typical for a dilaton - type scalar field in the matter - dominated era .
this must satisfy the atomic clock bounds at @xmath36 .
now @xmath37 ( @xmath38 being the hubble parameter in units of 100 km / s / mpc ) which according to @xcite is constrained to be @xmath39 one therefore finds @xmath40 .
this is smaller than the value inferred from the published evidence for a time variation of @xmath1 @xcite ( but see also @xcite ) ; conservatively we will assume a variation of 2 parts per million at @xmath41 , in which case @xmath42 .
although the difference is not big given the approximations being made , it does indicate that for the two measurements to be compatible the scalar field must freeze abruptly close to the present time @xcite ; the fact that we have neglected the effect of the onset of dark energy domination will not remove this difference .
given the @xmath1 variation of eq .
( [ alphabsbm ] ) , one can show @xcite that ( erroneously ) assuming eq .
( [ coupling ] ) to hold would lead to the following reconstructed equation of state @xmath43}\right]^{-1}\,,\ ] ] where @xmath44 the above assumptions and conservative assumptions on @xmath45 and @xmath46 imply @xmath47 -@xmath46 parameter space leading to allowed values of @xmath48 ( marked in the contour lines ) compatible with a webb - like variation of @xmath1 and the local bounds on @xmath7 . ]
( bottom band ) and @xmath49 ( middle ) , compared to the standard @xmath0cdm case ( top band ) .
the bands correspond the range @xmath50 , and the vertical error bars show the uncertainty in the sandage - loeb signal for a codex dataset with an observation time of 20 years . ] in fig .
[ fig1 ] we show the region of the @xmath51 parameter space compatible with this range of @xmath48 , and in fig .
[ fig2 ] we plot the sandage - loeb signal for some representative models .
notice that for a given amount of @xmath1 variation ( that is , a value of @xmath35 ) , a larger coupling @xmath7 implies a smaller @xmath48 ( a slower - moving scalar field ) and vice - versa .
large values of @xmath48 produce a sandage - loeb signal that codex can easily distinguish from @xmath0cdm , which would highlight the presence of an inconsistency in the assumptions .
conversely small values of @xmath48 yield a sandage - loeb signal indistinguishable from @xmath0cdm , but such a @xmath48 implies a large @xmath7 which could be checked with forthcoming ( improved ) local constraints . in either case , on the assumption that the current evidence for variations is correct , codex in combination with local experiments can support or rule out this class of models .
if the dark energy is a cosmological constant its contribution to the universe s energy budget is subdominant by redshift @xmath52 and negligible for @xmath53 . we now consider the opposite case , where the dark energy remains a significant fraction of the universe s energy density .
this is realized by the early dark energy models of doran and robbers @xcite , in which the dark energy density parameter and equation of state are @xmath54 @xmath55 } \frac{d\ln\omega_{\phi}(a)}{d\ln a } + \frac{a_{eq}}{3(a + a_{eq } ) } \label{eq : edew}\ ] ] @xmath56 being the scale factor at matter - radiation equality ; we still assume a flat universe , so @xmath57 .
( solid ) and @xmath58 ( dashed ) ; the dotted line shows the standard @xmath0cdm . in the sandage - loeb plot
the vertical error bars correspond to the uncertainty in the spectroscopic velocity shift for a codex dataset with an observation time of 20 years.,title="fig : " ] ( solid ) and @xmath58 ( dashed ) ; the dotted line shows the standard @xmath0cdm . in the sandage - loeb plot the vertical error bars correspond to the uncertainty in the spectroscopic velocity shift for a codex dataset with an observation time of 20 years.,title="fig : " ] , at redshift @xmath59 , as a function of @xmath7 and @xmath46 and assuming an early dark energy model with @xmath60 .
the shaded region is ruled out by the bound of eq .
( [ clocks ] ) . ]
the dark energy has a scaling behavior , approaching a finite constant @xmath61 in the past , while its equation of state @xmath62 tracks the dominant energy component .
we assume that the early dark energy field is also coupling to electromagnetism and thus yielding a varying @xmath1 @xcite , and our discussion in sect .
ii , and in particular eqs .
( [ darkside1]-[darkside2 ] ) also apply in this case . here
our analysis is the opposite of that in the previous section : there we assumed a given amount of @xmath1 variation ; here we will assume a given amount of early dark energy , namely @xmath60 , consistent with current bounds allowing for possible @xmath1 variations @xcite .
the local atomic clock bound is now @xmath63 regardless of @xmath61 . for @xmath46 significantly different from @xmath64
this places a model - dependent constraint on @xmath7 that is stronger than the model - independent @xmath65 , but large values of @xmath7 are possible by having @xmath46 sufficiently close to @xmath64 : with @xmath66 and @xmath67 the largest allowed value is @xmath68 ; this highlights the importance of atomic clock constraints , and shows that this model is almost indistinguishable from @xmath0cdm if one is limited to low - redshift observations .
despite the change in the dark energy equation of state at redshift @xmath69few , the sandage - loeb test is unable to distinguish this model from @xmath0cdm , as can be seen in fig .
nevertheless , this model can yield significant variations of @xmath1 , as shown in [ fig4 ] .
codex s baseline sensitivity for @xmath1 measurements is around the @xmath70 level , and as good as @xmath71 for a few ideal systems ; this is enough to detect such variations for @xmath72 and use the measurements to reconstruct @xmath73 , as discussed in @xcite .
we discussed two examples of codex s ability to probe the nature of dark energy beyond the regime where it is dynamically important and highlighted the importance of carrying out both the sandage - loeb test and accurate measurements of nature s fundamental couplings .
all three theoretical pillars of the @xmath0cdm paradigm ( inflation , dark matter and dark energy ) rely on the presence of new , presently unknown physics . in the absence of strong indications for what this new physics is and where it can be found ,
it is important to search for it in multiple places , and codex will have a unique role to play in the @xmath74 redshift range .
our analysis is simplified , but the goal is to illustrate the point at proof - of - concept level .
a detailed study , with precise codex specifications , can be done when these are finalized . finally , we emphasize that these tests do not exist in isolation : synergies can be found with other cosmological experiments , including esa s euclid mission , which will probe lower redshifts .
an analysis of these possibilities , in the context of a broader observational strategy , is left for future work .
this work was done in the context of the joint master in astronomy of the universities of porto and toulouse , supported by project ai / f-11 under the crup / portugal cup / france cooperation agreement ( f - fp02/11 ) .
we acknowledge the support of fundao para a cincia e a tecnologia ( fct ) , portugal , through grant ptdc / fis/111725/2009 .
the work of cm is funded by a cincia2007 research contract , supported by fct / mctes ( portugal ) and poph / fse ( ec ) .
99 s. cristiani _
et al . _ , nuovo cim . * b122 * ( 2007 ) , 1159 .
a. sandage , astrophys .
j. * 136 * ( 1962 ) 319 .
a. loeb , astrophys . j. * 499 * ( 1998 ) , 111 .
_ , mon . not .
* 386 * ( 2008 ) 1192 .
corasaniti , d. huterer and a. melchiorri , phys .
* d75 * ( 2007 ) 062001 .
a. balbi and c. quercellini , mon . not .
soc . * 382 * ( 2007 ) 1623 .
n. j. nunes and j. e. lidsey , phys .
* d69 * ( 2004 ) 123511 .
d. parkinson , b. a. bassett , and j. d. barrow , phys .
lett . * b578 * ( 2004 ) 235
. l. amendola , a. c. o. leite , c. j. a. p. martins _ et al .
_ , arxiv:1109.6793 . k. a. olive and m. pospelov , phys
. rev .
* d65 * ( 2002 ) 085044 .
g. r. dvali and m. zaldarriaga , phys .
* 88 * ( 2002 ) 091303 .
e. calabrese , e. menegoni , c. j. a. p.
martins _ et al .
_ , phys .
* d84 * ( 2011 ) 023518 .
h. b. sandvik , j. d. barrow , and j. magueijo , phys .
* 88 * ( 2002 ) 031302 .
t. rosenband _
et al . _ , science * 319 * ( 2008 ) , 1808 .
m. t. murphy _
et al . _ ,
notes phys .
* 648 * ( 2004 ) 131 .
m. t. murphy , j. k. webb and v. v. flambaum , mon . not .
* 384 * ( 2008 ) 1053 - 1062
. j. k. webb _
_ , phys .
* 107 * ( 2011 ) 191101 .
n. j. nunes , t. dent , c. j. a. p. martins , and g. robbers , mem .
s. a. it . * 80 * ( 2010 ) 785 . m. doran and g. robbers , j. c. a. p. * 0606 * ( 2006 ) 026 . | precision measurements of nature s fundamental couplings and a first measurement of the cosmological redshift drift are two of the key targets for future high - resolution ultra - stable spectrographs such as codex .
being able to do both gives codex a unique advantage , allowing it to probe dynamical dark energy models ( by measuring the behavior of their equation of state ) deep in the matter era and thereby testing classes of models that would otherwise be difficult to distinguish from the standard @xmath0cdm paradigm .
we illustrate this point with two simple case studies . |
braneworld models with large extra dimensions @xcite bring us an interesting possibility to address the hierarchy problem by lowering the fundamental scale of gravity down to order of tev .
it has been argued in the context of tev scale gravity that mini black holes might be created through high - energy particle collision at future colliders @xcite .
much effort has been directed towards a theoretical understanding of the black hole formation at tev energies ( e.g. , @xcite ) .
after their production , the black holes will decay via hawking radiation @xcite .
this process provides a window to probe high - energy physics , gravity at small distances , and properties of extra dimensions , which motivates recent extensive studies on this topic .
a nonexhaustive sampling of the literature can be found in refs .
@xcite . for
a review see ref .
most of the related work to date has ignored the effect of brane tension , treating black holes as `` isolated '' ones ( see , however , refs .
@xcite for the effects of self - gravity of branes ) .
it is in general very difficult to obtain a black hole solution localized on a brane with finite tension because tension curves the brane as well as the bulk ( cf
however , codimension-2 branes exceptionally allow for a simple construction of localized black holes thanks to their special property ; starting from the myers - perry solution @xcite one rescales the polar angle around a symmetry axis as @xmath0 and then the brane tension is proportional to the deficit angle @xmath1 . in this way both nonrotating @xcite and rotating @xcite black holes on codimension-2 branes have been constructed . following the work of @xcite , hawking evaporation @xcite and
the quasi - normal modes for bulk scalars @xcite and fermions @xcite have been investigated in the nonrotating background , showing that the finite brane tension modifies the standard result derived assuming negligible tension . in this paper
, we shall consider a six - dimensional _ rotating _ black hole pierced by a tense 3-brane and discuss the emission of massless scalar fields into the bulk .
we intend to shed light on the spin - down phase in the life of a black hole , which is often neglected in the literature but could be of some significance .
( in fact , a rotating black hole does not necessarily spin - down to zero , but evolves toward a nonzero angular momentum @xcite . ) ignoring the brane tension , very recently creek _ et al .
_ studied the emission of scalars in the bulk in a higher - dimensional rotating black hole background @xcite .
they employed matching techniques to obtain an analytic solution to the scalar field equation , which is a good approximation in the low - energy ( @xmath2 ) and slow - rotation ( @xmath3 ) regime , where @xmath4 is the energy of the emitted particle , @xmath5 is the black hole horizon radius , and @xmath6 is the rotation parameter . in the present paper , with the help of numerical computations we are able to handle the intermediate regime ( @xmath7 and @xmath8 ) , and thus we not only include the effect of the finite tension but also extend the range of validity of @xcite .
this paper is organized as follows . in the next section
we give a quick review of the rotating black hole solution on a codimension-2 brane . in sec .
iii we present separated equations of motion for a massless scalar field and determine angular eigenvalues in the presence of the deficit angle . then in sec .
iv the radial equation is solved numerically to give the power and angular momentum emission spectra .
finally we summarize our conclusions in sec .
v. appendix contains the analytic calculation of the absorption probability , which complements the numerical results presented in the main text .
we begin with a brief review of the rotating black hole solution on a codimension-2 brane .
( for further detail see refs .
the solution shares some properties with the myers - perry black hole @xcite .
we are considering the models with five spatial dimensions , and so the rotation group is @xmath9 .
the number of casimirs ( i.e. , the number of mutually commuting elements of the group ) is equal to rank@xmath10=2 $ ] .
hence , we have two axes of rotation associated with two angular momenta . however , in the present article we will be focusing on the special but simple case of a single rotation parameter with the angular momentum pointing along the brane .
this is indeed an interesting case from the phenomenological point of view , because the black hole formed by the collision of two particles confined to the brane will have a single rotation parameter .
the exact metric that describes such a rotating black hole is given by @xcite @xmath11 where @xmath12 the coordinate ranges are @xmath13 , and @xmath14 .
the parameter @xmath15 is related to the brane tension @xmath16 as @xmath17 where @xmath18 is the six - dimensional fundamental scale .
we assume that @xmath19 .
when @xmath20 , the above metric reduces to the usual myers - perry solution with a single rotation parameter in six dimensions @xcite . when @xmath21 the solution is asymptotically conical
the parameters @xmath22 and @xmath6 denote the specific mass and angular momentum , respectively , related to the adm mass and angular momentum of the black hole as @xmath23 where @xmath24 $ ] is the area of a unit @xmath25-sphere .
note here that the effect of the deficit angle @xmath15 is separated out from the definition of the area .
the black hole horizon radius @xmath5 follows from @xmath26 . for later purpose
it is convenient to define the dimensionless measure of the angular momentum @xmath27 . since the sign flip @xmath28 simply changes the direction of rotation , in what follows we will assume @xmath29 without any loss of generality .
note that @xmath30 has a root for arbitrary @xmath6 .
this should be contrasted with the four - dimensional kerr black hole , which has an upper bound on the rotation parameter .
this point will become clearer by considering the black hole temperature .
the generator of the horizon is parallel to the killing vector @xmath31 , and the surface gravity of the black hole is given by @xmath32 . here
@xmath33 is the angular velocity of the horizon : @xmath34 then , the hawking temperature is given by @xcite @xmath35 and one sees that the temperature never attains zero .
this means the black hole is manifestly nondegenerate for any value of @xmath6 .
an interesting effect of the brane tension perhaps one is immediately aware of is the rescaling of the gravitational scale @xmath36 , as is observed in eq .
( [ mandj ] ) .
this modifies the scenario of the black hole production and evaporation as follows @xcite .
suppose that a black hole is formed through a particle collision and its mass @xmath37 is fixed by the accelerator .
the upper limit on the impact parameter @xmath38 is given by @xmath39 .
since @xmath40 , for fixed @xmath37 the horizon radius is _ larger _ than the naive estimate , @xmath41 , which is called the `` lightning rod '' effect @xcite . in the present situation producing black holes will become easier due to this effect . since the black hole formed in this way has the angular momentum @xmath42 , we have @xmath43 , and so the angular momentum can also be _ larger _ than naively expected , @xmath44 . now the time scale @xmath45 of mass loss is estimated from @xmath46 ( horizon area)@xmath47 , yielding @xmath48 .
thus the lifetime of the black hole will become longer .
similarly , the time scale @xmath49 of angular momentum loss is estimated roughly from @xmath50 .
this gives @xmath51 , and hence the time scale @xmath49 will also become larger .
the above mentioned modifications are caused by the rescaling of the gravitational scale . in the rest of the paper we separate this effect and discuss the impact of finite brane tension on hawking emission _ with the horizon radius @xmath5 fixed_.
let us finally remark that in models with extra dimensions the bulk will be compact . throughout the paper
we assume that the horizon size of the black hole is smaller than the typical compactification scale . with this
, the above solution provides a nice description of a six - dimensional black hole localized on a 3-brane .
we now turn to discuss hawking emission of massless scalar fields . to this end
we solve the equation of motion for a massless scalar field @xmath52 in the rotating black hole background ( [ bh ] ) .
the governing equation is given by @xmath53 where @xmath54 .
we can separate the above equation by assuming the ansatz @xmath55 where @xmath56 the radial equation is @xmath57r=0 , \label{radial_eq}\end{aligned}\ ] ] where @xmath58 and the angular equations are given by @xmath59s = -\eta s,&&\label{ang_s } \\
\frac{1}{\sin\chi}\frac{{{\rm d}}}{{{\rm d}}\chi}\left(\sin\chi\frac{{{\rm d}}}{{{\rm d}}\chi}t\right ) -\frac{\mu^2}{\sin^2\chi}t=-\nu(\nu+1)t , & \\
\frac{1}{b^2}\frac{{{\rm d}}^2}{{{\rm d}}\psi^2}\xi=-\mu^2\xi.\label{eqxi}&\end{aligned}\ ] ] here @xmath60 and @xmath61 are separation constants .
the above equations of motion are the same as those derived in ref .
@xcite except that in the last equation ( [ eqxi ] ) there appears `` @xmath15 . ''
we immediately see that @xmath62 as is emphasized in ref .
@xcite , the eigenvalues are coupled and therefore all of the eigenvalues will be modified when @xmath21 . since the eigenvalues @xmath63 and @xmath64 will be dependent on @xmath65 , we can restrict to @xmath66 without any loss of generality .
negative values of @xmath67 can be treated similarly . for selected values of @xmath68 and @xmath15 .
solid and dashed lines refer to the series expansion ( [ eingen_expansion ] ) truncated at 7th order , while points and crosses show the numerical results without relying on the small-@xmath69 expansion.,height=188 ] we proceed to determine the eigenvalue @xmath63 . performing the change of the variable and function as @xmath70 , @xmath71
, we have @xmath72 the boundary conditions are given by @xmath73 as in @xcite , we first consider the special case in which @xmath74 with @xmath75 and hence @xmath76 is an integer . in this case
it is easy to find @xmath77 thus the eigenvalue @xmath63 can be written in terms of an integer @xmath78 as @xmath79 where @xmath80 and the absolute value signs are inserted for clarity .
although eq .
( [ eigen_nu ] ) was derived for @xmath81 , this result can be generalized to arbitrary @xmath82 $ ] @xcite .
we have confirmed numerically that eq .
( [ eigen_nu ] ) indeed holds for noninteger values of @xmath83 . when @xmath20 , @xmath63 is independent of @xmath67 ( and as a result @xmath64 is also independent of @xmath67 ) . for the tensional case , however , @xmath63 depends on @xmath67 ( and hence does @xmath64 ) . from eq .
( [ eigen_nu ] ) we see that the brane tension _ increases _ the eigenvalue @xmath63 relative to the tensionless case . to determine the eigenvalue @xmath64
, we exploit the continued fraction method developed originally by leaver @xcite .
following @xcite , we write @xmath84 with @xmath85
. then , @xmath86 obeys @xmath87\frac{{{\rm d}}}{{{\rm d}}u}{{\cal s}}\nonumber\\ + \left [ a^2\omega^2u^2+\eta-\nu(\nu+3)-|m|(|m|+2\nu+3 ) \right]{{\cal s}}=0 , \nonumber\\\label{calseq}\end{aligned}\ ] ] subject to the regularity boundary conditions @xmath88 we seek for a series solution in the form of @xmath89 substituting this to eq .
( [ calseq ] ) we obtain the three - term recursion relation @xmath90 where @xmath91 the continued fraction equation for the eigenvalue is then given by @xmath92 the boundary conditions are automatically satisfied by the series solution ( [ ssol ] ) .
we expand the eigenvalue @xmath64 in powers of @xmath69 around @xmath93 : @xmath94 in order for the series to converge in the limit @xmath95 , we require that it has a finite number of terms . imposing @xmath96 for some integer @xmath97 and identifying @xmath98 , we obtain @xmath99 for @xmath93 . here
we have a restriction for the integer @xmath100 : @xmath101 to determine the coefficients in ( [ eingen_expansion ] ) it is convenient to use the @xmath102th inversion of ( [ cfe ] ) : @xmath103 with some manipulation we find @xmath104 and @xmath105[2(\ell+b_n)+5 ] } , \label{et2 } \\ \tilde\eta_4&=&\frac{(\ell - j-|m|)(\ell+j-|m|+1 + 2b_n)}{4[2(\ell+b_n)+1]^2 } \left[\frac{(\ell - j-|m|-2)(\ell+j-|m|-1 + 2b_n)}{4[2(\ell+b_n)-1]}-\tilde\eta_2\right ] \nonumber\\ & & -\frac{(\ell - j-|m|+2)(\ell+j-|m|+3 + 2b_n)}{4[2(\ell+b_n)+5]^2 } \left [ \frac{(\ell - j-|m|+4)(j+\ell-|m|+5 + 2b_n)}{4[2(\ell+b_n)+7 ] } + \tilde\eta_2 \right].\label{et4}\end{aligned}\ ] ] one can confirm that eqs .
( [ et0])([et4 ] ) correctly reproduce the known result in the case of @xmath106 @xcite .
although the expressions are too lengthy , higher order coefficients can be obtained easily .
one can instead integrate eq .
( [ ang_s ] ) numerically to determine @xmath64 .
the numerical computation has an advantage that it does not rely on the small-@xmath69 expansion . in fig .
[ fig : eigenvalue.eps ] we compare the series expansion truncated at 7th order with the numerical result that is free from any approximation .
it can be seen from this that the analytic result can reproduce the numerical computation remarkably well even for @xmath107 .
therefore , in the next section we will safely use the analytic approximation for @xmath64 .
we thus avoid numerical determination of the eigenvalues and so the rest of the problem simply reduces to solving the ordinary differential equation for the radial mode function .
we remark here that while the brane tension does not affect the @xmath108 mode , the eigenvalue @xmath64 ( for @xmath109 ) increases as @xmath15 decreases ( and so the tension increases ) , as can be seen from fig .
[ fig : eigenvalue.eps ] .
this behavior has been observed for a nonrotating black hole in @xcite , and now it turns out that the similar thing generally holds for a rotating one .
, @xmath110 , and various @xmath111.,height=170 ] we are going to solve the radial equation ( [ radial_eq ] ) to compute the emission spectra . to do
so we first specify the asymptotic form of the solution close to the horizon and far away from it . in terms of the new coordinate defined by @xmath118{{\rm d}}r$ ] and the function @xmath119 , eq .
( [ radial_eq ] ) can be rewritten as @xmath120 ^ 2-v(r)\right\}\phi=0.\end{aligned}\ ] ] where @xmath121 and @xmath122 \nonumber\\ & + \frac{1}{r(r^2+a^2)^{1/2}}\frac{{{\rm d}}^2}{{{\rm d}}r_*^2 } \left[r(r^2+a^2)^{1/2}\right ] .
\label{pot_v}\end{aligned}\ ] ] it is easy to see that @xmath123 as @xmath124 and @xmath125 . keeping this in mind
the asymptotic solutions are found to be @xmath126 where @xmath127 and @xmath128 is the tortoise - like coordinate defined by @xmath129 .
\label{tortoise}\end{aligned}\ ] ] we choose the boundary condition @xmath130 , i.e. , we impose that no outgoing modes exist near the horizon .
the absorption probability is then given by latexmath:[\ ] ] in the static limit we have @xmath222 ( horizon area ) , which reproduces the general result for the spherically symmetric case @xcite .
n. arkani - hamed , s. dimopoulos and g. r. dvali , phys .
b * 429 * , 263 ( 1998 ) [ arxiv : hep - ph/9803315 ] ; n. arkani - hamed , s. dimopoulos and g. r. dvali , phys .
d * 59 * , 086004 ( 1999 ) [ arxiv : hep - ph/9807344 ] ; i. antoniadis , n. arkani - hamed , s. dimopoulos and g. r. dvali , phys .
b * 436 * , 257 ( 1998 ) [ arxiv : hep - ph/9804398 ] . c. kokorelis , nucl .
b * 677 * , 115 ( 2004 ) [ arxiv : hep - th/0207234 ] .
l. randall and r. sundrum , phys .
lett . * 83 * , 3370 ( 1999 ) [ arxiv : hep - ph/9905221 ] ; l. randall and r. sundrum , phys .
lett . * 83 * , 4690 ( 1999 ) [ arxiv : hep - th/9906064 ] .
s. b. giddings and s. d. thomas , phys . rev .
d * 65 * , 056010 ( 2002 ) [ arxiv : hep - ph/0106219 ] ; s. dimopoulos and g. l. landsberg , phys .
lett . * 87 * , 161602 ( 2001 ) [ arxiv : hep - ph/0106295 ] ; s. hossenfelder , s. hofmann , m. bleicher and h. stoecker , phys .
d * 66 * , 101502 ( 2002 ) [ arxiv : hep - ph/0109085 ] .
d. m. eardley and s. b. giddings , phys . rev .
d * 66 * , 044011 ( 2002 ) [ arxiv : gr - qc/0201034 ] ; h. yoshino and y. nambu , phys .
d * 66 * , 065004 ( 2002 ) [ arxiv : gr - qc/0204060 ] ; h. yoshino and y. nambu , phys .
d * 67 * , 024009 ( 2003 ) [ arxiv : gr - qc/0209003 ] ; e. kohlprath and g. veneziano , jhep * 0206 * , 057 ( 2002 ) [ arxiv : gr - qc/0203093 ] ; e. berti , m. cavaglia and l. gualtieri , phys .
d * 69 * , 124011 ( 2004 ) [ arxiv : hep - th/0309203 ] ; n. kaloper and j. terning , gen .
. grav .
* 39 * , 1525 ( 2007 ) [ arxiv:0705.0408 [ hep - th ] ] . c. m. harris and p. kanti , jhep * 0310 * , 014 ( 2003 ) [ arxiv : hep - ph/0309054 ]
; p. kanti , j. grain and a. barrau , phys . rev .
d * 71 * , 104002 ( 2005 ) [ arxiv : hep - th/0501148 ] ; s. creek , o. efthimiou , p. kanti and k. tamvakis , phys .
b * 635 * , 39 ( 2006 ) [ arxiv : hep - th/0601126 ] .
y. morisawa and d. ida , phys .
d * 71 * , 044022 ( 2005 ) [ arxiv : gr - qc/0412070 ] . v. cardoso , g. siopsis and s. yoshida , phys .
d * 71 * , 024019 ( 2005 ) [ arxiv : hep - th/0412138 ]
. v. cardoso , m. cavaglia and l. gualtieri , phys .
* 96 * , 071301 ( 2006 ) [ erratum - ibid . *
96 * , 219902 ( 2006 ) ] [ arxiv : hep - th/0512002 ] ; v. cardoso , m. cavaglia and l. gualtieri , jhep * 0602 * , 021 ( 2006 ) [ arxiv : hep - th/0512116 ] .
p. kanti and r. a. konoplya , phys .
d * 73 * , 044002 ( 2006 ) [ arxiv : hep - th/0512257 ] ; p. kanti , r. a. konoplya and a. zhidenko , phys .
d * 74 * , 064008 ( 2006 ) [ arxiv : gr - qc/0607048 ] .
d. ida , k. y. oda and s. c. park , phys .
d * 67 * , 064025 ( 2003 ) [ erratum - ibid .
d * 69 * , 049901 ( 2004 ) ] [ arxiv : hep - th/0212108 ] .
v. p. frolov and d. stojkovic , phys .
d * 66 * , 084002 ( 2002 ) [ arxiv : hep - th/0206046 ] ; v. p. frolov , m. snajdr and d. stojkovic , phys .
d * 68 * , 044002 ( 2003 ) [ arxiv : gr - qc/0304083 ] ; v. p. frolov , d. v. fursaev and d. stojkovic , jhep * 0406 * , 057 ( 2004 ) [ arxiv : gr - qc/0403002 ] ; v. p. frolov , d. v. fursaev and d. stojkovic , class .
* 21 * , 3483 ( 2004 ) [ arxiv : gr - qc/0403054 ] .
d. m. gingrich , jhep * 0711 * , 064 ( 2007 ) [ arxiv:0706.0623 [ hep - ph ] ] .
d. c. dai , g. starkman , d. stojkovic , c. issever , e. rizvi and j. tseng , arxiv:0711.3012 [ hep - ph ] .
a. flachi , o. pujolas , m. sasaki and t. tanaka , phys .
d * 74 * , 045013 ( 2006 ) [ arxiv : hep - th/0604139 ] ; a. flachi and t. tanaka , phys .
lett . * 95 * , 161302 ( 2005 ) [ arxiv : hep - th/0506145 ] ; a. flachi and t. tanaka , phys .
d * 76 * , 025007 ( 2007 ) [ arxiv : hep - th/0703019 ] .
r. emparan , g. t. horowitz and r. c. myers , jhep * 0001 * , 007 ( 2000 ) [ arxiv : hep - th/9911043 ] .
h. kudoh , t. tanaka and t. nakamura , phys .
d * 68 * , 024035 ( 2003 ) [ arxiv : gr - qc/0301089 ] .
s. creek , r. gregory , p. kanti and b. mistry , class .
* 23 * , 6633 ( 2006 ) [ arxiv : hep - th/0606006 ] .
f. r. tangherlini , nuovo cim . * 27 * , 636 ( 1963 ) ; r. c. myers and m. j. perry , annals phys .
* 172 * , 304 ( 1986 ) .
n. kaloper and d. kiley , jhep * 0603 * , 077 ( 2006 ) [ arxiv : hep - th/0601110 ] .
d. kiley , arxiv:0708.1016 [ hep - th ] .
d. c. dai , n. kaloper , g. d. starkman and d. stojkovic , phys .
d * 75 * , 024043 ( 2007 ) [ arxiv : hep - th/0611184 ] . s. chen , b. wang and r. k. su , phys .
b * 647 * , 282 ( 2007 ) [ arxiv : hep - th/0701209 ] .
u. a. al - binni and g. siopsis , arxiv:0708.3363 [ hep - th ] . h. t. cho , a. s. cornell , j. doukas and w. naylor , arxiv:0710.5267 [ hep - th ] .
c. m. chambers , w. a. hiscock and b. taylor , phys .
lett . * 78 * , 3249 ( 1997 ) [ arxiv : gr - qc/9703018 ] ; b. e. taylor , c. m. chambers and w. a. hiscock , phys .
d * 58 * , 044012 ( 1998 ) [ arxiv : gr - qc/9801044 ] .
e. berti , v. cardoso and m. casals , phys .
d * 73 * , 024013 ( 2006 ) [ erratum - ibid .
d * 73 * , 109902 ( 2006 ) ] [ arxiv : gr - qc/0511111 ] .
a. a. starobinsky , jetp 37 , 28 ( 1973 ) ; a. a. starobinsky and s.m .
churilov , jetp 38 , 1 ( 1973 ) . | we study the emission of scalar fields into the bulk from a six - dimensional rotating black hole pierced by a 3-brane .
we determine the angular eigenvalues in the presence of finite brane tension by using the continued fraction method .
the radial equation is integrated numerically , giving the absorption probability ( graybody factor ) in a wider frequency range than in the preexisting literature .
we then compute the power and angular momentum emission spectra for different values of the rotation parameter and brane tension , and compare their relative behavior in detail . as is expected from the earlier result for a nonrotating black hole
, the finite brane tension suppresses the emission rates .
as the rotation parameter increases , the power spectra are reduced at low frequencies due to the smaller hawking temperature and are enhanced at high frequencies due to superradiance .
the angular momentum spectra are enhanced over the whole frequency range as the rotation parameter increases .
the spectra and the amounts of energy and angular momentum radiated away into the bulk are thus determined by the interplay of these effects . |
[ introduction ] the reflection and transmission of light at air - medium interface has been a topic of ongoing investigation and contention .
initially , the amplitudes and directions of reflected and transmitted light were determined by fresnel formulas and snell s law .
deviations were later found in total internal reflection ( tir ) , however .
they are goos - hnchen ( gh ) shift @xcite and imbert - fedorov ( if ) shift @xcite , which are parallel and perpendicular to the incident plane , respectively .
artmann formula @xcite for gh shift and schilling formula @xcite for if shift in tir are widely accepted . the exact formulas for if shifts in partial reflection and transmission , however , have divided many physicists @xcite .
owing to hosten s precise measurement , the debate was settled down @xcite .
apart from the constant spatial gh and if shifts , the light beam also experiences angular shifts @xcite , which increase proportionally with propagation distance @xmath0 . in general , the spatial gh and if shifts stem from spin - orbit interaction at the air - medium interface @xcite .
to satisfy the @xmath0-component angular momentum conservation law , the reflected and transmitted beams must possess extrinsic transverse orbital angular momenta , which results in spatial gh and if shifts .
the angular shift is essentially a diffractive correction on light beams and governed by the linear momentum conservation law @xcite .
the emergence of left - handed materials ( lhm ) has brought about great opportunities and sophisticated pathways to manipulate light @xcite . as a branch of metamaterials
, lhm usually refers to those materials whose permittivity and permeability are negative simultaneously .
it has shown very unique properties , like negative refraction @xcite , inverse doppler effect @xcite , unreversed rotational doppler effect @xcite and inverse cherenkov radiation @xcite .
apart from these properties , beam shifts in lhm have been demonstrated to be significantly different from right - handed materials ( rhm ) @xcite .
theoretical papers @xcite indicate that light beam experiences negative gh shift at air - lhm interface and this shift can be amplified in layered structure .
the if shift in partial reflection , however , remains unreversed owing to unreversed spin angular momentum in lhm @xcite . as for vortex beam
, it carries intrinsic orbital angular momentum .
therefore , orbit - orbit conversion is inevitable in reflection and transmission .
will the spatial gh and if shifts remain unreversed in the presence of orbital angular momentum ?
how about the angular shifts ?
what does the physical picture of spin - orbit and orbit - orbit conversions look like ? clarifying these problems is not only conducive to understanding the detailed behaviors of vortex beams at air - lhm interface but also sheds a new light on the essence of angular momentum and linear momentum of vortex beam . in this paper
, we endeavor to offer concrete expressions of spatial and angular shifts of vortex beam at air - lhm interface and elaborate on their relations with the linear and angular momenta .
we also contrast these results with air - rhm interface .
although some of the results are similar to our previous paper @xcite , we conduct considerably rigorous analyses on the topics and thorough explanations on the results .
the rest of the paper is arranged as follows . in sec .
[ electric fields ] , we adopt angular spectrum method to derive the electric fields of reflected and transmitted vortex beams at air - lhm interface . except for some special cases , for instance , in the vicinity of critical angle of tir and brewster angle , the expressions of electric fields generally hold true . the longitudinal fields are included as well , which take on polarization - sensitive vortex structure @xcite , providing a new perspective on the mechanism of if shifts other than spin - orbit conversion . in sec .
[ shifts ] , we adopt the operator method to calculate the spatial and angular shifts , which is significantly different from conventional calculation method @xcite and save us considerable time and efforts . the gh and if shifts in tir and partial transmission are demonstrated and contrasted with air - rhm interface .
the impact of incident angle on beam shifts is analyzed as well . in sec .
[ momenta ] , we calculate the linear and angular momenta of incident , reflected , and transmitted beams , respectively
. the origins of spatial and angular shifts are clarified . the quantitative analysis on spin - orbit conversion and orbit - orbit conversion
is also demonstrated .
in this study , we adopt the angular spectrum method to establish a model for the reflected and transmitted vortex beams at air - lhm interface . the main procedures of this method are as follows .
we first decompose the vortex beams into plane waves with finite spectral width .
then , we analyze the incident angle and fresnel coefficients of each plane wave component separately . in this case , the incident angle and fresnel coefficients of each plane wave slightly differ from that of the main fourier component ( also known as central wave component ) .
therefore , we expand the fresnel coefficients in taylor series around the central incident angle and make approximation to the first order .
afterwards , we transform the electric fields from momentum space to position space . the geometry of reflection and transmission is demonstrated in fig .
[ geometry ] .
the incident vortex beam propagates along @xmath1 axis and impinges on the air - lhm interface .
the incident plane of the main fourier component is @xmath2 .
the reflected wave and transmitted wave of the main fourier component travel along the @xmath3 axis and @xmath4 axis , respectively .
the incident angle , reflection angle , and transmission angle of the main fourier component are @xmath5 , @xmath6 , and @xmath7 , respectively .
the relations among the coordinates @xmath8 , @xmath9 , @xmath10 , and @xmath11 are determined by the snell s law : @xmath12=\left[\matrix{\cos\vartheta_\tau&0&-\sin\vartheta_\tau\cr 0&1&0\cr\sin\vartheta_\tau&0&\cos\vartheta_\tau}\right]\left[\matrix{\hat{\mathbf{x } } \cr\hat{\mathbf{y}}\cr\hat{\mathbf{z}}}\right],\label{snell 's law}\ ] ] where @xmath13 , or @xmath14 , @xmath15 $ ] and @xmath16 $ ] are the unit basis vectors of the coordinates @xmath8 and @xmath17 , respectively , @xmath18=\left[\matrix{\theta_i&\pi-\theta_i&-\theta_t}\right]$ ] , @xmath19 and @xmath20 .
constant @xmath21 is the refractive index of the lhm .
( marked with green lines ) .
for an arbitrary plane wave , the incident plane is @xmath22 ( marked with blue lines).,width=302 ] for an arbitrary plane wave , we assume that the incident plane is @xmath22 . the incident , reflected , and transmitted plane wave travel along the @xmath23 , @xmath24 , and @xmath25 axis , respectively .
the incident angle , reflection angle , and transmission angle of an arbitrary fourier component are @xmath26 , @xmath27 , and @xmath28 , respectively .
the relations among coordinates @xmath29 , @xmath30 , @xmath31 , and @xmath32 are easily acquired by revising @xmath18 $ ] in eq .
( [ snell s law ] ) as @xmath33 $ ] . from fig .
[ geometry ] , we derive the incident angle of an arbitrary plane wave , @xmath34 , where @xmath35 is the wave number in vacuum , @xmath36 , @xmath37 , @xmath38 are the components of wave vector along the @xmath39 , @xmath40 , @xmath1 axis , respectively .
we can expand the incident angle @xmath26 around the central incident angle @xmath5 in series of @xmath41 and @xmath42 .
therefore , @xmath43 , @xmath44 , where @xmath45 .
after introducing the geometry of reflection and transmission , we start analyzing the angular spectrum of reflected and transmitted beams by using transformation matrix .
note that the incident , reflected , and transmitted vortex beams are presented in local coordinate systems @xmath46 .
the angular spectrum of incident beam is : @xmath47\tilde{u}_i,\nonumber\\ \tilde{u}_i&=&\frac{c_l w_0}{2}\left[w_0 \left(-i k_{x_i}+\text{sgn}[l]k_{y_i}\right)/\sqrt{2}\right]^{|l|}\nonumber\\ & & \times\exp\left[-w_0 ^ 2\left(k_{x_i}^2+k_{y_i}^2\right)/4\right],\label{incidentbeam}\end{aligned}\ ] ] where @xmath48 is the angular spectrum of vortex beams , @xmath49 is the vortex charge , @xmath50 is the normalization constant , @xmath51 is the width of beam waist , @xmath36 and @xmath37 are wave vector components along the @xmath52 and @xmath53 axis , @xmath35 is wave number in vacuum , @xmath54 $ ] is the sign function , @xmath55 and @xmath56 are jones vectors , @xmath57 .
there are two parameters that characterize the polarization state of paraxial beams : @xmath58 , \chi=2 ~\text{re}\left[\alpha^{*}\beta\right]$ ] .
@xmath59 is the degree of circular polarization .
a value of @xmath60 corresponds to left circularly polarized light beam , whereas a parameter of @xmath61 stands for right circularly polarized light beam .
@xmath62 represents linear polarization and values between 0 and 1 should correspond to elliptically polarized states .
@xmath63 is the degree of linear polarization .
it is generally recognized that the circularly polarized vortex beam has intrinsic angular momentum @xmath64 per photon @xcite .
we first write the electric field in coordinate @xmath30 .
this target would be achieved in three steps .
first , we write the electric field in @xmath8 by using the matrix in eq .
( [ snell s law ] ) .
second , we transform the electric field from @xmath8 to @xmath29 .
this step is accomplished by using the relations : @xmath65 , @xmath66 , @xmath67 , where @xmath68 .
therefore , this transformation matrix is given below : @xmath69=\left[\matrix{\frac{k_{x_i}\cos\theta_i+k_{z_i}\sin\theta_i}{k\sin\theta_i^{'}}&\frac{k_{y_i}}{k\sin\theta_i^{'}}&0\cr -\frac{k_{y_i}}{k\sin\theta_i^{'}}&\frac{k_{x_i}\cos\theta_i+k_{z_i}\sin\theta_i}{k\sin\theta_i^{'}}&0\cr 0&0&1}\right]\left[\matrix{\hat{\mathbf{x } } \cr\hat{\mathbf{y}}\cr\hat{\mathbf{z}}}\right].\label{matrix1}\ ] ] third , we transform the electric field from @xmath29 to @xmath30 using revised version of eq .
( [ snell s law ] ) .
the transformation matrix of these three steps is written as @xmath70=\left[\matrix{1&\frac{k_{y_i}\cot\theta_i}{k}&-\frac{k_{x_i}}{k}\cr -\frac{k_{y_i}\cot\theta_i}{k}&1&-\frac{k_{y_i}}{k}\cr \frac{k_{x_i}}{k}&\frac{k_{y_i}}{k}&1}\right]\left[\matrix{\hat{\mathbf{x}}_i \cr\hat{\mathbf{y}}_i\cr\hat{\mathbf{z}}_i}\right].\label{matrix2}\ ] ] in eq .
( [ matrix2 ] ) , we assume that terms equivalent to or higher than @xmath71 , @xmath72 are null . after we get the expression of electric field in @xmath30 coordinate , the reflected and transmitted fields can be easily obtained by multiplying the fresnel coefficients .
we define the amplitude reflection coefficients of the main fourier component as @xmath73 , the amplitude transmission coefficients of the main fourier component as @xmath74 , where @xmath75 and @xmath76 denote @xmath75-polarized and @xmath76-polarized state .
we also define the amplitude reflection coefficients of an arbitrary plane wave as @xmath77 , the amplitude transmission coefficients of an arbitrary plane wave as @xmath78 . to simplify the fresnel coefficients of an arbitrary wave component , we expand them around the central incident angle @xmath5 in series of @xmath41 and @xmath42 , and retain the first order term .
thus , @xmath79 , @xmath80 . note that , the reflected and transmitted electric fields are still presented in @xmath31 and @xmath32 coordinates .
hence , we need to transform them back to the @xmath10 and @xmath11 coordinates .
the transformation matrixes are @xmath81&=&\left[\matrix{1&\frac{k_{y_i}\cot\theta_i}{k}&-\frac{k_{x_i}}{k}\cr -\frac{k_{y_i}\cot\theta_i}{k}&1&\frac{k_{y_i}}{k}\cr \frac{k_{x_i}}{k}&-\frac{k_{y_i}}{k}&1}\right]\left[\matrix{\hat{\mathbf{x}}_i \cr\hat{\mathbf{y}}_i\cr\hat{\mathbf{z}}_i}\right],\nonumber\\
\left[\matrix{\hat{\mathbf{x}}_t\cr\hat{\mathbf{y}}_t\cr \hat{\mathbf{z}}_t}\right]&=&\left[\matrix{1&\frac{-\eta k_{y_i}\cot\theta_i}{k}&\frac{k_{x_i}}{n\eta k}\cr \frac{\eta k_{y_i}\cot\theta_i}{k}&1&\frac{k_{y_i}}{n k}\cr -\frac{k_{x_i}}{n\eta k}&-\frac{k_{y_i}}{n k}&1}\right]\left[\matrix{\hat{\mathbf{x}}_i \cr\hat{\mathbf{y}}_i\cr\hat{\mathbf{z}}_i}\right].\label{matrix3}\end{aligned}\ ] ] the detailed procedures are similar to the eq .
( [ matrix2 ] ) . as mentioned previously , we eliminate terms equal to or higher than @xmath71 and @xmath72 .
the fresnel coefficients only reveal the amplitudes of reflected and transmitted vortex beams . to get the full expressions of reflected and transmitted vortex beams
, we still need the phase matching conditions . for an arbitrary plane wave component
, the boundary conditions @xcite require that in the plane @xmath82 , @xmath83=\exp[i\mathbf{k}_r\cdot\mathbf{r}]=\exp[i\mathbf{k}_t\cdot\mathbf{r}]$ ] , where @xmath84 , @xmath85 , and @xmath86 are the wave vectors of incident , reflected , and transmitted beams , respectively .
therefore , the phase matching conditions are presented like this : @xmath87 combining the fresnel coefficients and eqs .
( [ incidentbeam ] ) and ( [ matrix2])-([boundaryconditions ] ) , we get the electric fields of reflected and transmitted beams : @xmath88\tilde{u}_r\hat{\mathbf{x}}_r\nonumber\\ & + & \left[\beta\left(r_s-\frac{\partial r_s}{\partial \theta_i}\frac{k_{x_r}}{k}\right)-\alpha\left(r_s+r_p\right)\cot\theta_i\frac{k_{y_r}}{k}\right]\tilde{u}_r\hat{\mathbf{y}}_r\nonumber\\ & -&\frac{1}{k}\left(\alpha r_p k_{x_r}+\beta r_s k_{y_r}\right)\tilde{u}_r\hat{\mathbf{z}}_r,\label{reflectedbeam}\end{aligned}\ ] ] @xmath89\tilde{u}_t\hat{\mathbf{x}}_t\nonumber\\ & + & \left[\beta\left(t_s+\eta\frac{\partial t_s}{\partial \theta_i}\frac{k_{x_t}}{k}\right)+\alpha\left(\eta t_p - t_s\right)\cot\theta_i\frac{k_{y_t}}{k}\right]\tilde{u}_t\hat{\mathbf{y}}_t\nonumber\\ & -&\frac{1}{nk}\left(\alpha t_p k_{x_t}+\beta t_s k_{y_t}\right)\tilde{u}_t\hat{\mathbf{z}}_t,\label{transmittedbeam}\end{aligned}\ ] ] where @xmath90 , @xmath91 .
note that the reflected and transmitted vortex beams experience extra phase shifts @xcite associated with diffraction while propagating , which are @xmath92~\text{and}~\exp\left[inkz_t\left(1-\frac{k_{x_t}^2+k_{y_t}^2}{2n^2k^2}\right)\right]$ ] , respectively .
axis , respectively .
we choose @xmath93 for incident vortex beam .
the incident , reflected , and transmitted vortex beams have anti - clockwise , clockwise , and clockwise helical wavefronts , respectively.,width=302 ] at this stage , we feel obliged to deliver several comments .
first , the phase matching conditions , as demonstrated in eq .
( [ boundaryconditions ] ) , play a crucial role in this paper . for instance , @xmath94 show that the wavefront of reflected vortex beam is reversed ( shown in fig .
[ wavefront ] ) , resulting in a reversed orbital angular momentum @xmath95 per photon .
@xmath96 shows that the transmitted beam in lhm undergoes negative phase velocity , resulting in a reversed helical wavefront ( shown in fig .
[ wavefront ] ) .
the reversed wavefront in lhm does not lead to reversed orbital angular momentum , which is explained in sec .
[ momenta ] .
the equation @xmath97 means the transmitted beam experiences a beam deformation in the cross section .
second , in the vicinity of critical angle @xmath98 of tir and brewster angle @xmath99 ( @xmath100 , where @xmath101 is the beam divergence angle ) , we should be cautious .
for example , in the vicinity of critical angle @xmath102 ) , the first derivatives of @xmath103 are extremely large @xcite , meaning that eqs .
( [ reflectedbeam ] ) and ( [ transmittedbeam ] ) should be revised to include terms higher than @xmath71 and @xmath72 .
if a p - polarized vortex beam impinges on the air - lhm interface with incident angle being in the vicinity of brewster angle ( @xmath104 ) , the power of reflected beam would be in the scale of @xmath105 ( approximately @xmath106 for paraxial beams ) comparing with the incident beam .
this constitutes the reason why the experiments @xcite fail to collect reliable data concerning the shift of p - polarized beams in the vicinity of brewster angle .
third , although the beam shape and shift of reflected vortex beam are already clear @xcite .
the profile of transmitted vortex beam is far from clear - cut .
we formulate a rigorous transmitted field in the appendix of our paper . by referencing this appendix , the beam shape of the transmitted vortex beam
is easily acquired .
the paraxial wave equation is identical with two - dimensional schrdinger equation with @xmath0 replaced by @xmath14 .
hence , the operator formalism proposed by stoler @xcite has become a powerful tool in physical optics @xcite . in this section
, we will apply the operator formalism to calculate the centroid of the reflected and transmitted beams . in momentum space ,
the transverse position operator is @xmath107 .
therefore , the centroid of beams in a given plane @xmath108 is readily given by @xmath109 where @xmath110 , @xmath111 , the propagation operator @xmath112 $ ] .
the above equation can be easily formulated into the following one : @xmath113 where @xmath35 is the wave number in vacuum and @xmath21 is the refractive index .
the first and second terms of eq .
( [ beamcentroid_2 ] ) are the spatial and angular shifts , which are independent and dependent on @xmath0 , respectively . from the second term of eq .
( [ beamcentroid_2 ] ) , we can easily find that the negative refractive index @xmath21 results in reversed angular shifts in lhm @xcite . for the reflected and transmitted beams , the gh and if shifts can be expressed as : @xmath114 and @xmath115 , where @xmath116 .
we first calculate the spatial and angular shifts of reflected beams . substituting eq .
( [ reflectedbeam ] ) into eq .
( [ beamcentroid_2 ] ) , we get the spatial and angular shifts of reflected beams @xmath117\bigg\},\label{y_r}\\ \delta \theta_{x_r}&=&\frac{-(|l|+1)\left(\left|\alpha \right|^2\left|r_p\right|\frac{\partial\left|r_p\right|}{\partial \theta_i}+\left|\beta\right|^2\left|r_s\right|\frac{\partial \left|r_s\right|}{\partial \theta_i}\right)}{kz_r\left(\left|\alpha r_p\right|^2+\left|\beta r_s\right|^2\right)},\label{x_theta_r}\\ \delta \theta_{y_r}&=&\frac{\chi(|l|+1)\cot\theta_i\left(\left|r_p\right|^2-\left|r_s\right|^2\right)}{2kz_r\left(\left|\alpha r_p\right|^2+\left|\beta r_s\right|^2\right)},\label{y_theta_r}\end{aligned}\ ] ] where @xmath118 $ ] , @xmath119 $ ] , @xmath120 $ ] , @xmath121 is the rayleigh length . when evaluating the energy term @xmath122 , we discard the cross - polarization terms ( terms proportional to @xmath123 or @xmath124 ) and @xmath3-component electric field @xmath125 , since their average energy density is proportional to @xmath105 .
the first terms of eqs .
( [ x_r ] ) and ( [ y_r ] ) are vortex - induced spatial gh and if shifts .
they were initially proposed by bliokh @xcite and fedoseyev @xcite , respectively .
experimental demonstrations were accomplished by merano @xcite and dasgupta @xcite at air - glass interface .
note that the vortex - induced spatial gh shift only exists in mixed linearly polarized beams ( polarization vector is oblique to the incident plane ) , while the vortex - induced spatial if shift occurs in arbitrary polarized state .
the second term of eq .
( [ x_r ] ) coincides with the well - known artmann formula @xcite . the second term of eq .
( [ y_r ] ) is the spin - dependent if shift @xcite . in partial reflection and tir
, it turns into the bliokh formula @xcite and schillings formula @xcite , respectively .
equations ( [ x_theta_r ] ) and ( [ y_theta_r ] ) are the angular shifts . in general
, they are proportional to @xmath105 . for p - polarized beams , these angular shifts could be remarkably magnified through brewster resonance @xcite ( finally in the scale of @xmath126 near the brewster angle ) . in these cases ,
the cross - polarization terms are not negligible when we calculate the energy intensity . , refractive index of lhm @xmath127 , refractive index of rhm @xmath128 , vortex charge @xmath93 , propagation distance @xmath129 , beam waist @xmath130 , where @xmath131 is the wavelength in vacuum . to make the shifts more noticeable , we amplify them by 20 times . in the first row ,
the incident beam is left - circularly polarized .
the actual beam shifts of fig .
[ totalreflection](a ) are @xmath132 and @xmath133 .
the actual beam shifts of fig .
[ totalreflection](b ) are @xmath134 and @xmath135 . in the second row , the incident beam is mixed linearly polarized with @xmath136 .
the actual beam shifts of fig .
[ totalreflection](c ) are @xmath132 and @xmath137
. the actual beam shifts of fig .
[ totalreflection](d ) are @xmath134 and @xmath138 . , width=302 ] a careful assessment of eqs .
( [ x_r])-([y_theta_r ] ) indicates that , in partial reflection region , the spatial and angular gh and if shifts at loss - free air - lhm and air - rhm interfaces are identical . in tir , shifts are different , however . in this case , the vortex - induced shifts and angular shifts are null . the phase @xmath139 has the same magnitude but the opposite sign for rhm and lhm , resulting in a negative gh shift @xcite in lhm
the if shifts for circularly polarized beams and mixed linearly polarized beams are @xmath140/k$ ] and @xmath141 , respectively . here
, we theoretically predict that owing to the phase reversion , the if shift in total reflection region would also be reversed at air - lhm interface when the incident beam is in mixed linearly polarized state .
this prediction is illustrated in fig .
[ totalreflection ] . with the refractive index @xmath142 , we choose incident @xmath143 to avoid the deformation of reflected beams @xcite . figure [ totalreflection](a ) and fig .
[ totalreflection](b ) show that the if shift in tir remains unreversed when the incident beam is in circularly polarized state .
figure [ totalreflection](c ) and fig .
[ totalreflection](d ) suggest that the if shift in tir is reversed when the incident beam is in mixed linearly polarized state . ,
refractive index of lhm @xmath127 , refractive index of rhm @xmath128 , polarization state @xmath144 , @xmath145 , propagation distance @xmath129 , beam waist @xmath130 . to make the shifts more noticeable ,
we amplify them by 200 times .
in the first row , the vortex charge is @xmath93 .
the actual beam shifts of fig .
[ partialtransmission](a ) are @xmath146 and @xmath147 .
figure [ partialtransmission](b ) has the same shifts with fig .
[ partialtransmission](a ) . in the second row ,
the vortex charge is @xmath148 .
the actual beam shifts of fig .
[ partialtransmission](c ) are @xmath149 and @xmath150 .
figure [ partialtransmission](d ) has the same shifts with fig .
[ partialtransmission](c ) .
, width=302 ] , the refractive index of rhm @xmath128 , polarization state @xmath144 , @xmath145 .
the spatial shifts are presented in the scale of @xmath131 .
, width=302 ] now , we begin to study the shifts of transmitted vortex in lhm .
we substitute eq .
( [ transmittedbeam ] ) into eq .
( [ beamcentroid_2 ] ) and get the spatial and angular shifts of transmitted beams @xmath151\bigg\},\label{y_t}\\ \delta \theta_{x_t}&=&\frac{\eta(|l|+1)\left(\left|\alpha \right|^2\left|t_p\right|\frac{\partial\left|t_p\right|}{\partial \theta_i}+\left|\beta\right|^2\left|t_s\right|\frac{\partial \left|t_s\right|}{\partial \theta_i}\right)}{kz_{r_x}\left(\left|\alpha
t_p\right|^2+\left|\beta t_s\right|^2\right)},\label{x_theta_t}\\ \delta \theta_{y_t}&=&\frac{\chi(|l|+1)\cot\theta_i\left(\left|t_p\right|^2-\left|t_s\right|^2\right)}{2kz_{r_y}\left(\left|\alpha t_p\right|^2+\left|\beta t_s\right|^2\right)},\label{y_theta_t}\end{aligned}\ ] ] where @xmath152 $ ] , @xmath153 and @xmath154 is the rayleigh length along @xmath155 and @xmath156 axis .
these rayleigh lengths of transmitted vortex beam in lhm are both negative owing to negative phase velocity @xcite .
the first terms of eqs .
( [ x_t ] ) and ( [ y_t ] ) are vortex - induced spatial gh and if shifts . until now
, no experiments have been reported on these vortex - induced gh and if shifts .
the vortex - induced gh shift of transmitted beam occurs when the incident beam is in mixed linearly polarized state . the vortex - induced if shift , however , exists in any polarization state . the second term of eq .
( [ x_t ] ) is the phase - dependent gh shift . in lossy media ,
this part is not negligible . the second term of eq .
( [ y_t ] ) is the spin - dependent if shift .
if the lhm is loss - free , this term degenerates into bliokh formula @xcite .
a special case is total transmission " , where @xmath157 . in this case , @xmath158 , we get @xmath159 . , the refractive index of rhm @xmath128 , polarization state @xmath144 , @xmath145 .
the angular shifts are presented in the scale of @xmath105 .
, width=302 ] we closely examine eqs .
( [ x_t])-([y_theta_t ] ) and find that spatial gh and if shifts of partial transmitted beams remain unreversed at air - lhm interface compared with air - rhm interface .
the angular shifts , however , are reversed .
figure [ partialtransmission ] shows the spatial gh and if shifts of transmitted beams at air - lhm and air - rhm interfaces .
the beam profiles are drawn according to eqs .
( [ transmittedbeam_position_space_1 ] ) , ( [ vortex_1 ] ) and ( [ vortex_-1 ] ) .
we can easily find that the transmitted beam is stretched along @xmath155 axis , which is a pure geometrical phenomenon .
we assume that the lhm is low - loss or loss - free .
this assumption is reasonable considering state - of - the - art micro / nano manufacturing technology .
for example , by incorporating gain media into fishnet structure @xcite or exploiting second - order magnetic resonance of the fishnet structure @xcite , we can obtain low - loss , 3-d , polarization - independent lhm in the visible spectral range . based on this assumption ,
the phase - dependent term of spatial gh shift vanishes and the spatial gh shift only depends on vortex .
this argument is confirmed by fig .
[ partialtransmission](a ) and fig .
[ partialtransmission](c ) , where the vortex charge is @xmath93 and @xmath148 , respectively .
they indicate that when the vortex charge is reversed , the spatial gh shift is also reversed .
the spatial if shift is both vortex and spin - dependent .
therefore , when the vortex charge is flipped , the absolute value of spatial if shift is altered [ but not flipped , see fig .
[ partialtransmission](a ) and fig .
[ partialtransmission](c ) ] . in fig .
[ spatialshifts ] , we demonstrates how incident angle impact the spatial gh and if shifts .
when the incident angle increases , both gh and if shifts increase .
the if shift fails to converge when @xmath5 approaches to @xmath160 .
we note that in this case , the energy transmission coefficient is zero .
hence , the spatial shifts of transmitted beams are meaningless . figure [ spatialshifts ] also suggests that both spatial gh and if can be enhanced by raising the vortex charge .
we proceed to analyze the angular gh and if shifts .
figure [ angularshifts ] demonstrates how the incident angle affects the angular shifts .
figure [ angularshifts](a ) and fig .
[ angularshifts](b ) show the angular gh shift at air - lhm and air - rhm interfaces , respectively .
they indicate that the angular gh shift is reversed in air - lhm .
figure [ angularshifts](c ) and fig .
[ angularshifts](d ) show the angular if shift at air - lhm and air - rhm interfaces , respectively .
they indicate that the angular if shift is reversed in air - lhm . for angular
if shift in air - lhm , there exists a minimum point as incident increases from @xmath161 to @xmath160 .
these angular shifts increase proportionally as we raise the absolute value of vortex charge .
in this section , we start to analyze the linear and angular momenta of reflected and transmitted vortex beams . in general , the linear momentum density can be divided into orbital part and spin part @xcite , which reads @xmath162+\frac{1}{2}\text{im}[\nabla\times(\mathbf{e}^*\times\mathbf{e})],\label{momentumdensity}\ ] ] where the first term is orbital momentum density @xmath163 , the second term is spin momentum density @xmath164 , @xmath165 is the electric field in position space . in momentum space , we can write eq .
( [ momentumdensity ] ) into @xmath166 therefore , the momentum per unit length is @xmath167 where the first term is orbital momentum @xmath168 , the second and third terms are spin momentum @xmath169 . from the above equation , we find that @xmath170 , which means @xmath171 in lhm is opposite to @xmath171 in rhm . for both paraxial and nonparaxial beams , the spin part makes no contribution to the linear momentum @xcite : @xmath172 . in position space ,
the angular momentum density denotes @xmath173 . in momentum space ,
we replace @xmath174 with operator @xmath175 and get the angular momentum @xmath176\hat{\mathbf{z}},\end{aligned}\ ] ] where the first term is orbital angular momentum @xmath177 , the second and third terms are spin angular momentum @xmath178 .
note that the transverse spin angular momenta are null , which means @xmath179 .
we substitute the propagation operator @xmath112 $ ] into eq .
( [ angular_momentum ] ) and get the orbital and spin angular momentum @xcite @xmath180\hat{\mathbf{z}}.\label{spin_angular_momentum}\end{aligned}\ ] ] we find that the transverse orbital angular momenta @xmath181 and @xmath182 are proportional to @xmath183 , which means they have opposite directions in lhm compared with the counterparts in rhm .
equations ( [ orbital_angular momentum ] ) and ( [ spin_angular_momentum ] ) also indicate that the @xmath0-component orbital angular momentum @xmath184 and spin angular momentum @xmath185 have the same directions in lhm and rhm . comparing eq .
( [ beamcentroid_2 ] ) with eqs .
( [ linear_momentum ] ) and ( [ orbital_angular momentum ] ) , we can easily get the following relations @xmath186 where @xmath187 and @xmath188 are spatial shifts , @xmath189 and @xmath190 are angular shifts .
we first offer a qualitative explanation on why spatial shifts are unreversed in lhm but angular shifts are reversed in lhm .
equation ( [ sptail_and_angular_shfits ] ) indicates that spatial shifts have no direct relations with the @xmath0-component orbital angular momentum @xmath184 and spin angular momentum @xmath185 .
they are dependent on the transverse angular momenta @xmath191 and @xmath0-component linear momentum @xmath192 @xcite . since the directions of @xmath191 and @xmath192 are both reversed in lhm , the spatial shifts @xmath193 would remain unreversed as a result .
the angular shifts , however , solely depend on linear momentum .
the reason why angular shifts are reversed in lhm is that the transverse linear momentum @xmath194 is unreversed but the @xmath0-component linear momentum @xmath192 is reversed . from now on
, we will perform a quantitative analysis on the linear and angular momenta and confirm the momentum conservation laws .
therefore , we assume that the lhm is loss - free and the amplitude reflection coefficients @xmath103 and amplitude transmission coefficients @xmath195 are real variables . by substituting eqs .
( [ incidentbeam ] ) , ( [ reflectedbeam ] ) , and ( [ transmittedbeam ] ) into eq .
( [ linear_momentum ] ) , we obtain the linear momenta of incident , reflected , and transmitted beams @xmath196 where @xmath197 is the energy reflection coefficient , @xmath198 .
the energy transmission coefficient is @xmath199 .
note that the linear momenta of incident , reflected , and transmitted beams are presented in three different coordinate systems @xmath200 .
we can verify that the linear momenta along @xmath201 and @xmath202 axes satisfy the conservation law @xcite @xmath203 for incident beam , there is no transverse linear momentum . to satisfy the linear momentum conservation law , transverse linear momenta @xmath204 , @xmath205 , @xmath206 , @xmath207 are produced .
these transverse linear momenta are responsible for angular shifts .
we proceed to analyze the angular momenta . by substituting eqs .
( [ incidentbeam ] ) , ( [ reflectedbeam ] ) , and ( [ transmittedbeam ] ) into eqs .
( [ orbital_angular momentum ] ) and ( [ spin_angular_momentum ] ) , we get the angular momenta of incident , reflected , and transmitted beams .
@xmath208\hat{\mathbf{x}}_r\nonumber\\ & & + \frac{\chi l}{2}(r_p^2-r_s^2)\cot\theta_i\hat{\mathbf{y}}_r+(-l q_r+\sigma r_pr_s)\hat{\mathbf{z}}_r,\nonumber\\ \mathbf{j}_t&=&\bigg[-\frac{nl}{2\eta}\frac{\partial t}{\partial\theta_i}-\frac{n\sigma}{2\eta}(t_p^2+t_s^2 - 2\eta t_pt_s)\cot\theta_i\bigg]\hat{\mathbf{x}}_t\nonumber\\ & & + \frac{n\chi l}{2}(t_s^2-t_p^2)\cot\theta_i\hat{\mathbf{y}}_t+\bigg[\frac{l(1+\eta^2)t}{2\eta^2}+\frac{\sigma t_pt_s}{\eta}\bigg]\hat{\mathbf{z}}_t,\nonumber\\\label{angular_momenta}\end{aligned}\ ] ] where the spin angular momenta of incident , reflected , and transmitted beams are @xmath59 , @xmath209 , and @xmath210 , respectively . for each individual photon ,
the angular momenta of incident , reflected , and transmitted beams are @xmath64 , @xmath211 , @xmath212\hbar$ ] , respectively . though the orbital and spin momenta of reflected photon and transmitted photon depend on the absolute value refractive index , they are independent on the sign of refractive index .
the @xmath0-component angular momenta satisfy the conservation law : @xmath213 from eqs .
( [ angular_momenta ] ) and ( [ angular_momenta_conservation ] ) , we can infer that there are two types of momentum conversions .
the first type is spin - orbit conversion .
the @xmath0-component spin angular momentum of incident beam @xmath214 converts into transverse angular momenta @xmath215 and @xmath216 , resulting in spin - dependent if shifts .
the second type is orbit - orbit conversion .
the @xmath0-component orbital angular momentum of incident beam @xmath217 converts into transverse angular momenta @xmath215 and @xmath216 , resulting in vortex - induced if shifts .
the vortex - induced gh shifts , though relate to @xmath218 , are not governed by angular momenta conservation law . in tir ,
the linear and angular momenta of incident and reflected beams are @xmath219\hat{\mathbf{x}}_r\nonumber\\ & & -\left(|\alpha|^2\frac{\partial \phi_{r_p}}{\partial \theta_i}+|\beta|^2\frac{\partial \phi_{r_s}}{\partial \theta_i}\right)\hat{\mathbf{y}}_r\nonumber\\ & & + \big[-l+\sigma\cos(\phi_{r_s}-\phi_{r_p})+\chi\sin(\phi_{r_s}-\phi_{r_p})\big]\hat{\mathbf{z}}_r,\nonumber\\\end{aligned}\ ] ] we can easily verify that they fulfil the conservation law @xmath220 at this point , we would like to add three comments .
first , it s worth noting that we adopted the minkowski momentum in eqs .
( [ linear_momenta_conservation ] ) and ( [ angular_momenta_conservation ] ) in this paper .
although the abraham - minkowski dilemma has been solved @xcite , why minkowski momentum is a proper form in this study , we believe , is an interesting problem worth further investigation .
second , two important papers @xcite were published while our paper was being peer - reviewed .
one paper @xcite adopted the real - space approach and revealed the role of longitudinal field @xmath221 in if shifts for the first time .
we think additional analysis on the connection between longitudinal field and spin - orbit conversion can also be carried out in the momentum space .
the other paper @xcite unambiguously separated the effects of beam shape and other parameters ( such as polarization , the property of the interface ) on gh and if shifts , which has long escaped researchers attention over the past years .
but the discussions are confined to reflected vortex beam . generalization to transmitted vortex beam , we think , remains quite challenging .
third , owing to the close similarity between light beam and matter waves , scientists have found that electron beams can also possess orbital angular momentum by passing through a spiral phase plate @xcite or nanofabricated diffraction hologram @xcite . in this regard
, we hopefully predict that vortex electron beam might also experience vortex - induced shifts in a potential well . by properly designing the potential well and taking advantage of quantum weak measurements or other measuring technology
, we may even observe the vortex - induced beam shifts of electron beam in experiment .
in conclusion , we have derived the reflected and transmitted fields of vortex beam at air - lhm interface via angular spectrum method . by using this method ,
we have managed to get the formulas of spatial gh shifts , spatial if shifts , angular gh shifts , and angular if shifts .
these formulas suggest that the spatial gh and if shifts remain unreversed at air - lhm interface compared with air - rhm interface . by raising the vortex charge ,
the spatial shifts can be remarkably enhanced . in tir , apart from reversed gh shift , we predict that the if shift would also be reversed when the incident beam is in mixed linearly polarized state .
the physical interpretation of these interesting phenomena lies in the reversed transverse angular momenta and reversed linear momenta .
although the spatial shifts have no direct relations with the @xmath0-component angular momentum ( @xmath0-component angular momentum is not reversed in lhm ) , they are actually the outcomes of spin - orbit and orbit - orbit conversion . therefore , the unreversed spatial shifts are indirect evidence of unreversed angular momentum of lhm . as for angular shifts ,
they are reversed at air - lhm interface and can be amplified by enhancing the vortex charge .
this is direct evidence on the reversed linear momentum of lhm .
besides these qualitative analysis , we also offer concrete expressions of the transverse linear and angular momenta , which explicitly reveal the physical picture of spin - orbit and orbit - orbit conversions .
these momentum conversions are governed by @xmath0-component angular momentum conservation law .
in this appendix , we will give the full analytical expressions of the electric fields of reflected and transmitted beams .
the electric fields in position space are given by inverse fourier transformation @xmath222\nonumber\\ & & \times\exp\left[inkz-\frac{iz}{2nk}(k_x^2+k_y^2)\right].\end{aligned}\ ] ] we first apply eq .
( [ fouriertransformation ] ) to calculate the reflected fields .
substituting eq .
( [ reflectedbeam ] ) into eq .
( [ fouriertransformation ] ) , we get the reflected beams @xmath223\right\}u_r\hat{\mathbf{x}}_r\nonumber\\ & + & \left\{\beta r_s+\frac{i}{k}\left[\beta\frac{\partial r_s}{\partial\theta_i}\frac{\partial}{\partial x_r}+\alpha\left(r_s+r_p\right)\cot\theta_i\frac{\partial}{\partial y_r}\right]\right\}u_r\hat{\mathbf{y}}_r\nonumber\\ & + & \frac{i}{k}\left(\alpha r_p\frac{\partial}{\partial x_r}+\beta r_s\frac{\partial}{\partial y_r}\right)u_r\hat{\mathbf{z}}_r,\label{reflectedbeam_position_space_1}\\ u_r&=&\frac{c_lkw_0}{2(z_r+iz_r)}\left[\frac{kw_0}{\sqrt{2}}\frac{x_r - i~\text{sgn}[l]y_r}{z_r+iz_r}\right]^{|l|}\nonumber\\ & & \times\exp\left[-\frac{k(x_r^2+y_r^2)}{2(z_r+iz_r)}+ikz_r\right].\label{reflectedbeam_position_space_2}\end{aligned}\ ] ] note that the orbital angular momentum of reflected beam is @xmath95 per photon .
( [ reflectedbeam_position_space_1 ] ) and ( [ reflectedbeam_position_space_2 ] ) are the full expressions of reflected electric fields .
we proceed to calculate the transmitted electric fields . substituting eq .
( [ transmittedbeam ] ) into eq .
( [ fouriertransformation ] ) , we get the transmitted fields @xmath224\right\}u_t\hat{\mathbf{x}}_t\nonumber\\ & + & \left\{\beta t_s-\frac{i}{k}\left[\beta\eta\frac{\partial t_s}{\partial\theta_i}\frac{\partial}{\partial x_t}+\alpha\left(\eta t_p - t_s\right)\cot\theta_i\frac{\partial}{\partial y_t}\right]\right\}u_t\hat{\mathbf{y}}_t\nonumber\\ & + & \frac{i}{nk}\left(\alpha t_p\frac{\partial}{\partial x_t}+\beta t_s\frac{\partial}{\partial y_t}\right)u_t\hat{\mathbf{z}}_t.\label{transmittedbeam_position_space_1}\end{aligned}\ ] ] the formula of @xmath225 is rather lengthy . if @xmath226 is an even number , then the formula is @xmath227^{|l|}\exp\left[-\frac{nk}{2}\left(\frac{x_t^2}{z_{r_x}+iz_t}+\frac{y_t^2}{z_{r_y}+iz_t}\right)+inkz_t\right]\nonumber\\ & & \times\bigg\{\sum_{m=0,2,4\cdots}^{|l|}c_{|l|}^{m}\left[\frac{-i\eta}{\sqrt{a}}\right]^{|l|-m}\left[\frac{\text{sgn}[l]}{\sqrt{c}}\right]^{m}\exp[-ab^2-cd^2]\gamma\left[\frac{1+|l|-m}{2}\right]\gamma\left[\frac{1+m}{2}\right]\nonumber\\ & & ~~~\times~\leftidx{_1}f_1\left[\frac{1+|l|-m}{2},\frac{1}{2},ab^2\right]\leftidx{_1}f_1\left[\frac{1+m}{2},\frac{1}{2},cd^2\right]\nonumber\\ & & ~~~+\sum_{m=1,3,5\cdots}^{|l|-1}c_{|l|}^{m}\left[\frac{-i\eta}{\sqrt{a}}\right]^{|l|-m}\left[\frac{\text{sgn}[l]}{\sqrt{c}}\right]^{m}4\sqrt{ac}bd\exp[-ab^2-cd^2]\gamma\left[1+\frac{|l|-m}{2}\right]\gamma\left[1+\frac{m}{2}\right]\nonumber\\ & & ~~~\times~\leftidx{_1}f_1\left[1+\frac{|l|-m}{2},\frac{3}{2},ab^2\right]\leftidx{_1}f_1\left[1+\frac{m}{2},\frac{3}{2},cd^2\right]\bigg\};\label{transmittedbeam_position_space_2}\end{aligned}\ ] ] if @xmath226 is an odd number , then the formula is @xmath227^{|l|}\exp\left[-\frac{nk}{2}\left(\frac{x_t^2}{z_{r_x}+iz_t}+\frac{y_t^2}{z_{r_y}+iz_t}\right)+inkz_t\right]\nonumber\\ & & \times\bigg\{\sum_{m=0,2,4\cdots}^{|l|-1}c_{|l|}^{m}\left[\frac{-i\eta}{\sqrt{a}}\right]^{|l|-m}\left[\frac{\text{sgn}[l]}{\sqrt{c}}\right]^{m}2\sqrt{a}b\exp[-ab^2-cd^2]\gamma\left[1+\frac{|l|-m}{2}\right]\gamma\left[\frac{1+m}{2}\right]\nonumber\\ & & ~~~\times~\leftidx{_1}f_1\left[1+\frac{|l|-m}{2},\frac{3}{2},ab^2\right]\leftidx{_1}f_1\left[\frac{1+m}{2},\frac{1}{2},cd^2\right]\nonumber\\ & & ~~~+\sum_{m=1,3,5\cdots}^{|l|}c_{|l|}^{m}\left[\frac{-i\eta}{\sqrt{a}}\right]^{|l|-m}\left[\frac{\text{sgn}[l]}{\sqrt{c}}\right]^{m}2\sqrt{c}d\exp[-ab^2-cd^2]\gamma\left[\frac{1+|l|-m}{2}\right]\gamma\left[1+\frac{m}{2}\right]\nonumber\\ & & ~~~\times~\leftidx{_1}f_1\left[\frac{1+|l|-m}{2},\frac{1}{2},ab^2\right]\leftidx{_1}f_1\left[1+\frac{m}{2},\frac{3}{2},cd^2\right]\bigg\},\label{transmittedbeam_position_space_3}\end{aligned}\ ] ] where , @xmath228 is the binomial coefficient , @xmath229 is the gamma function , @xmath230 is the kummer confluent hypergeometric function , @xmath231 , @xmath232 , @xmath233 , @xmath234 . equations ( [ transmittedbeam_position_space_1])-([transmittedbeam_position_space_3 ] ) fully describe the transmitted vortex beams .
although eqs .
( [ transmittedbeam_position_space_2 ] ) and ( [ transmittedbeam_position_space_3 ] ) are cumbersome , we still manage to get the electric fields for several low order vortex beams @xmath235,\\ u_t^{l=-2}&\propto&\bigg[\left(\frac{\eta x_t}{z_{r_x}+iz_t}-\frac{iy_t}{z_{r_y}+iz_t}\right)^2\nonumber\\ & & + \frac{iz_t(1-\eta^2)}{nk(z_{r_x}+iz_t)(z_{r_y}+iz_t)}\bigg].\end{aligned}\ ] ] f. goos , and h. hnchen , ann
. phys . * 1 * , 333 ( 1947 ) .
k. artmann , ann .
phys . * 437 * , 87 ( 1948 ) .
h. m. lai , f. c. cheng , and w. k. tang , j. opt .
am . a * 3 * , 550 ( 1986 ) .
f. lima , t. dumelow , e. l. albuquerque , and j. a. p.
da costa , j. opt .
b * 28 * , 306 ( 2011 ) .
f. i. fedorov , dokl .
nauk sssr * 105 * , 465 ( 1955 ) .
h. schilling , ann .
* 16 * , 122 ( 1965 ) . c. imbert , phys .
d * 5 * , 787 ( 1972 ) .
m. a. player , j. phys .
a * 20 * , 3667 ( 1987 ) .
mnard , a. e. mattacchione , h. m. van driel , c. hautmann , and m. betz , phys .
b * 82 * , 045303 ( 2010 ) . c. a. dartora , g. g. cabrera , k. z. nobrega , v. f. montagner , m. h. k. matielli ,
f. k. r. de campos , and h. t. s. filho , phys .
a * 83 * , 012110 ( 2011 ) .
f. pillon , h. gilles , and s. girard , appl . opt . * 43 * , 1863 ( 2004 ) .
m. onoda , s. murakami , and n. nagaosa , phys .
lett . * 93 * , 083901 ( 2004 ) .
k. y. bliokh , and y. p. bliokh , phys . rev
. lett . * 96 * , 073903 ( 2006 ) .
k. y. bliokh , and y. p. bliokh , phys .
e * 75 * , 066609 ( 2007 ) .
o. hosten , and p. kwiat , science * 319 * , 787 ( 2008 ) .
c. c. chan , and t. tamir , opt . lett .
* 10 * , 378 ( 1985 ) .
m. merano , a. aiello , m. p. van exter , and j. p. woerdman , nat . photon . * 3 * , 337 ( 2009 ) .
k. y. bliokh , i. v. shadrivov , and y. s. kivshar , opt . lett . * 34 * , 389 ( 2009 ) .
v. g. fedoseyev , j. opt .
* 13 * , 064025 ( 2011 ) . v. g. fedoseyev ,
. commun . *
282 * , 1247 ( 2009 ) .
j. b. pendry , a. j. holden , w. j. stewart , and i. youngs , phys .
lett . * 76 * , 4773 ( 1996 ) .
r. a. shelby , d. r. smith , and s. schultz , science * 292 * , 77 ( 2001 ) .
h. j. lezec , j. a. dionne , and h. a. atwater , science * 316 * , 430 ( 2007 ) .
r. liu , c. ji , j. j. mock , j. y. chin , t. j. cui , and d. r. smith , science * 323 * , 366 ( 2009 ) .
n. seddon , and t. bearpark , science * 302 * , 1537 ( 2003 ) . j. chen , y. wang , b. jia , t. geng , x. li , l. feng , w. qian , b. liang , x. zhang , m. gu , and s. zhuang , nat .
* 5 * , 239 ( 2011 ) .
h. luo , s. wen , w. shu , z. tang , y. zou , and d. fan , phys .
a * 78 * , 033805 ( 2008 ) .
z. duan , b .-
wu , j. lu , j. a. kong , and m. chen , opt .
express * 16 * , 18479 ( 2008 ) .
p. r. berman , phys .
e * 66 * , 067603 ( 2002 ) .
i. v. shadrivov , a. a. zharov , and y. s. kivshar , appl .
* 83 * , 2713 ( 2003 ) .
c. menzel , c. rockstuhl , t. paul , s. fahr , and f. lederer , phys .
a * 77 * , 013810 ( 2008 ) . c. m. krowne , phys .
a * 373 * , 466 ( 2009 ) .
h. luo , s. wen , w. shu , z. tang , y. zou , and d. fan , phys .
a * 80 * , 043810 ( 2009 ) .
l. allen , m. w. beijersbergen , r. j. c. spreeuw , and j. p. woerdman , phys .
a * 45 * , 8185 ( 1992 ) .
a. t. oneil , i. macvicar , l. allen , and m. j. padgett , phys .
* 88 * , 053601 ( 2002 ) .
j. d. jackson , _ classical electrodynamics _
( wiley , new york , 1999 ) . j. w. goodman , _ introduction to fourier optics _ ( mcgraw - hill , new york , 1996 ) .
h. okuda , and h. sasada , j. opt .
a * 25 * , 881 ( 2008 ) .
r. dasgupta , and p. k. gupta , opt .
commun . * 257 * , 91 ( 2006 ) .
y. qin , y. li , h. he , and q. gong , opt .
lett . * 34 * , 2551 ( 2009 ) .
d. stoler , j. opt .
71 * , 334 ( 1981 ) .
s. j. van enk , and g. nienhuis , j. mod . opt . * 41 * , 963 ( 1994 ) .
a. aiello , n. lindlein , c. marquardt , and g. leuchs , phys .
lett . * 103 * , 100401 ( 2009 ) .
k. y. bliokh , m. a. alonso , e. a. ostrovskaya , and a. aiello , phys .
a * 82 * , 063825 ( 2010 ) . v. g. fedoseyev ,
opt . commun .
* 193 * , 9 ( 2001 ) .
m. merano , n. hermosa , j. p. woerdman , and a. aiello , phys .
a * 82 * , 023817 ( 2010 ) .
a. aiello , and j. p. woerdman , opt
. lett . * 33 * , 1437 ( 2008 ) .
h. luo , w. hu , z. ren , w. shu , and f. li , opt .
* 266 * , 327 ( 2006 ) . s. xiao , v. p. drachev , a. v. kildishev , x. ni , u. k. chettiar , h .- k . yuan , and v. m. shalaev , nature * 466 * , 735 ( 2010 ) . c. garca - meca , j. hurtado , j. mart , a. martnez , w. dickson , and a. v. zayats , phys .
lett . * 106 * , 067402 ( 2011 ) .
m. v. berry , j. opt .
a : pure appl
* 11 * , 094001 ( 2009 ) .
li , phys .
a * 80 * , 063814 ( 2009 ) .
s. m. barnett , phys .
* 104 * , 070401 ( 2010 ) .
p. w. milonni , and r. w. boyd , adv .
* 2 * , 519 ( 2010 ) .
b. a. kemp , j. appl .
* 109 * , 111101 ( 2011 ) .
a. y. bekshaev , phys .
a * 85 * , 023842 ( 2012 ) .
a. aiello , new j. phys .
* 14 * , 013058 ( 2012 ) .
m. uchida , and a. tonomura , nature * 464 * , 737 ( 2010 ) .
j. verbeeck , h. tian , and p. schattschneider , nature * 467 * , 301 ( 2010 ) . | in this paper , we present a systematic study of beam shifts and angular momenta of paraxial vortex beams at air - left - handed material ( lhm ) interfaces .
it is shown that , compared to their counterparts at air - right - handed material ( rhm ) interfaces , the spatial goos - hnchen ( gh ) and imbert - fedorov ( if ) shifts remains the same , while the angular gh and if shifts are reversed , at air - lhm interfaces .
the spatial and angular shifts of paraxial vortex beams have their respective origins in transverse angular momenta and transverse linear momenta .
the spatial gh and if shifts remain unreversed as a result of the both reversions of transverse angular momenta and z - component linear momentum , while the angular gh and if shifts are reversed due to that the z - component linear momentum is reversed and the transverse linear momenta are unreversed , at air - lhm interfaces .
in addition , we perform a quantitative analysis on spin - orbit angular momentum conversion and orbit - orbit angular momentum conversion , which further helps us understand the essence of vortex beam shifts at air - lhm interfaces and their fundamental distinctions with those at air - rhm interfaces . |
the detection of cooper pairs above the superconducting transition temperature @xmath0 has a long history , dating back to 1969.tinkham it implies that the average of the order parameter squared , @xmath12 , does not vanish above @xmath0 , either due to thermal fluctuations or the limited effective spatial extent of the system .
while the regime where thermal fluctuations dominate is reasonably well understood in terms of the scaling theory of critical phenomena subjected to finite size effects , ffh , tsh , kosh , hofer , book , parks , tsjs , larkin novel features have been observed considerably outside the fluctuation dominated regime .
recently , li _ et al_.@xcite have compiled the results of an extended experimental study of the isothermal magnetization of several families of cuprate superconductors over a rather broad range of temperatures and magnetic fields . from these
, they infer that above the transition temperature @xmath0 , the isothermal diamagnetic contribution to the magnetization @xmath13 decreases initially with increasing magnetic field @xmath14 , applied parallel to the @xmath15-axis , consistent with @xmath16 , where @xmath17 is the diamagnetic susceptibility .
however , as @xmath14 increases @xmath13 tends to a minimum at @xmath18 and in excess of this characteristic field the magnetization increases and appears to approach zero , as shown in fig .
[ fig1 ] for la@xmath7sr@xmath8cuo@xmath9 , showing data taken from li _
et al_.@xcite _
vs_. @xmath14 at various temperatures above @xmath19 k for la@xmath7sr@xmath8cuo@xmath9 taken from li _
et al_.@xcite .
the symbol @xmath20 indicates the minimum in the curves for the presented isotherms . ] closely related behavior was reported earlier for oriented powder samples of underdoped y@xmath21ca@xmath22ba@xmath23cu@xmath24o@xmath25,@xcite for pb nanoparticles with particle size larger than the correlation length,@xcite for mgb@xmath23,@xcite and for polycrystalline smba@xmath23cu@xmath26al@xmath25o@xmath27.@xcite the magnetization was either inferred from torque magnetometry,@xcite or probed by squid magnetometery.@xcite however , the torque measurements reveal a temperature dependent paramagnetic background @xmath28 .
the diamagnetic contribution to the magnetization is then derived from@xcite @xmath29 where @xmath30 although this subtraction leads to uncertainties in the high field limit , where the diamagnetic signal becomes small , it appears unlikely that the main feature , the occurrence of the minimum in @xmath31 at fixed temperature @xmath32 , is an artifact of this subtraction . although the crossover from the initial linear ( @xmath16 ) to nonlinear behavior is an expected feature of thermal fluctuations in homogeneous two ( 2d ) and three ( 3d ) dimensional superconductors , the occurrence of the minimum
can not be explained invoking these scenarios.@xcite on the other hand , there is considerable evidence that cuprate and amorphous conventional superconductors are homogeneous over a limited spatial domain only .
@xcite in this case , the growth of the correlation lengths is limited by approaching the transition temperature @xmath0 because it can not exceed the respective extent of the homogenous domains . within a two dimensional superconductor , consisting of a stack of superconducting layers with insulating spacing sheets in between , the adoption of this scenario where the magnetization stems from homogeneous domains with limited extent
only would result in an effective 0d - superconductor . as a consequence
, the isothermal magnetization curves would exhibit a minimum , reminiscent to the observation in nanoparticles.@xcite related behavior was also observed in the isothermal nernst signal of superconducting and amorphous nb@xmath3si@xmath4 films above @xmath0.@xcite in this system , the nernst signal due to normal quasiparticles is particularly low .
this allows to probe the contribution associated with superconductivity without the subtraction of the background due to the normal quasiparticles.@xcite at low magnetic field , the nernst signal @xmath33 increases linearly with field , where @xmath34 is the nernst coefficient . upon increasing the magnetic field , @xmath1 deviates from this linear field dependence , reaches a maximum at @xmath35 and decreases afterwards . in analogy to the position of the minimum in the magnetization , the maximum shifts to higher fields with increasing temperature .
this behavior confirms the evidence that the isothermal nernst signal is proportional to the magnetization in terms of @xmath5.@xcite here we review the properties of the isothermal magnetization curves of a 0d - superconductor , neglecting thermal fluctuations , and explore the consistency with experimental magnetization data of pb nanoparticles,@xcite bulk la@xmath7sr@xmath8cuo@xmath9,@xcite and bisr@xmath10cacu@xmath10o@xmath36 ( bi2212 ) with @xmath37 k and @xmath38 k.@xcite to explore wether this scenario also accounts for the nernst signal @xmath1 in terms of the relation @xmath5 , we consider the data for superconducting nb@xmath3si@xmath4 films taken above @xmath0.@xcite in sec .
ii we sketch the theoretical background including the properties of the 0d - model .
the neglect of thermal fluctuations implies that the model is applicable outside the critical regime only , that is sufficiently above @xmath0 , the regime where the experimental data of la@xmath7sr@xmath8cuo@xmath9 and bi2212 was taken . invoking quantum scaling the doping dependence of the minimum in the isothermal magnetization curves
is also addressed . in sec .
iii we present the analysis of the data based on the 0d - model , neglecting thermal fluctuations .
the remarkable agreement with the measured isothermal magnetization curves , achieved for reasonable values of the model parameters , suggest that the occurrence of the minimum is attributable to a finite extent of the homogeneous domains .
the doping dependence of the bi2212 data is also consistent with the flow to a quantum phase transition in the underdoped limit .
furthermore it is shown that the profile of the isothermal nernst signal of the superconducting nb@xmath3si@xmath4 film , treated as @xmath5 , is fully consistent with the 0d - model .
accordingly , singlet cooper pairs subjected to orbital pair breaking in a 0d system are the main source of the observed diamagnetism and nernst signal in an extended temperature window above @xmath0 . finally we show that the 0d - model provides for a variety of conventional and hole doped superconductors a universal perspective on the interplay between diamagnetism , nernst signal , correlation length and the limited spatial extent of homogeneity .
we close with a brief summary and some discussion .
the fluctuation contribution to the free energy per unit volume of a homogeneous and anisotropic type ii superconductor scales above @xmath0 as@xcite @xmath39 where @xmath40 is a scaling function of its argument and @xmath41 is the magnetic field induced limiting length giving rise to a finite size effect.@xcite we assume that the magnetic field is applied along the @xmath42-axis . @xmath43 denote the correlation length along the respective axis in zero field . in the limit @xmath44 , attainable for sufficiently high fields
, this expression reduces to @xmath45 because the zero field correlation lengths @xmath46 can not grow beyond @xmath47 . in this limit
the magnetization @xmath48 tends to @xmath49 on the other hand , in the opposed limit @xmath50 the scaling function adopts the limiting behavior , @xmath51 , to recover @xmath13 @xmath52 . in this case
we obtain @xmath53 accordingly , in both the 3d and 2d case , where @xmath54 and @xmath55 denotes the thickness of the superconducting sheets , the magnetization saturates at sufficiently high fields due to the magnetic field induced finite size effect , reducing the effective dimensionality @xmath56 of the system from @xmath56 to @xmath57.@xcite in this limit the system corresponds in @xmath58 to independent superconducting cylinders of radius @xmath59 and height @xmath60 and in @xmath61 with height @xmath54 .
detailed calculations in @xmath61 reveal that the crossover from the low to the high field limit occurs monotonically and according to that there is no minimum.@xcite so far we considered homogeneous systems only . in practice
any real and highly anisotropic type ii superconductor is homogenous within _ e.g. _ a cylinder of radius @xmath62 and height @xmath55 .
concentrating on temperatures sufficiently above @xmath0 where thermal fluctuations in the phase and amplitude of the order parameter can be neglected , we are left with a 0d - system with an order parameter @xmath63 which does not depend on the space variables .
the temperature and magnetic field dependence follows then from the ginzburg - landau ( gl ) model for a 0d system as treated by shmidt.@xcite the partition function in this case reads @xmath64\right ) , \label{eq5}\ ] ] with the gl free energy functional @xmath65 & = & \frac{v}{k_{\mathrm{b}}t}\bigg[r_{0}\left(\ln \left(\frac{t}{t_{c}}\right ) + \left ( \frac{2\pi } { \phi _ { 0}}\right)^{2}\xi_{0}^{2}\left\langle \mathbf{a}^{2}\right\rangle \right ) \left\vert\psi\right\vert^{2 } \nonumber \\ & + & \frac{u_{0}}{2}\left\vert \psi\right\vert^{4}\bigg ] , \label{eq6}\end{aligned}\ ] ] and @xmath66 @xmath67 is the vector potential and @xmath68 the correlation length with amplitude @xmath69 .
setting @xmath70 assuming @xmath71 and @xmath72 we obtain for the partition function the expression @xmath73 erfc@xmath74 is the complementary error function .
the magnetization per unit volume follows then from the free energy @xmath75 in terms of @xmath76 yielding @xmath77\right ) .
\label{eq10}\ ] ] using the gauge @xmath78 we obtain for a cylindrical homogenous domain with radius @xmath62 , height @xmath55 and a spherical domain with radius @xmath79 @xmath80 where @xmath81 the magnetization expression ( [ eq10 ] ) can then be rewritten as @xmath82\right ) , \label{eq13}\ ] ] or @xmath83\bigg ) , \label{eq14}\end{aligned}\ ] ] where @xmath84 in terms of the variable @xmath85 , requiring the values of @xmath86 and @xmath87 , eq .
( [ eq13 ] ) adopts the simple scaling form @xmath88 , \label{eq15a}\ ] ] with the limiting behavior @xmath89 in the limit @xmath90 @xmath13 reduces then to @xmath91 consistent with @xmath13 @xmath52 .
contrariwise , for @xmath92 it approaches @xmath93 accordingly , the isothermal magnetization curves adopt for @xmath32 a minimum between the low and high field limits .
this characteristic behavior also appears in fig .
more specific the minimum at @xmath94 follows from @xmath95 yielding for sufficiently large @xmath86 the solution @xmath96 where @xmath97 together with eq .
( [ eq17a ] ) and sufficiently large @xmath86 we obtain for the magnetic field @xmath98 , where the isothermal magnetization curves adopt a minimum , the relation @xmath99 accordingly , @xmath100 does not vanish at @xmath0 and the temperature dependent part is related with @xmath87 in eq .
( [ eq15 ] ) to the amplitude of the correlation length .
this differs from the gaussian approximation , valid for @xmath101 ( @xmath102 ) . in this limit
( [ eq17 ] ) , rewritten in the form @xmath103 applies .
here @xmath104 adopts at fixed temperature a minimum at @xmath105 between the low and high field behavior .
furthermore , in the cylindrical case is the ratio @xmath106 according to eqs .
( [ eq12 ] ) and ( [ eq21 ] ) given by @xmath107 the gaussian expression eq .
( [ eq20 ] ) for the magnetization also implies that there is no particular depairing field .
indeed , the magnetization vanishes as @xmath108 . as aforementioned
, the applicability of the gaussian approximation requires that @xmath109 is very large .
according to this , in systems with non - negligible quartic term @xmath110 it fails close to @xmath0 in the low field limit . considering highly anisotropic cuprates , such as la@xmath111sr@xmath22cuo@xmath9 with @xmath112 and bi2212 , this corresponds to the critical regime where a 3d - xy to 2d - xy crossover occurs and phase fluctuations dominate.@xcite however , in this regime and for sufficiently large @xmath62 even the full model fails because fluctuations are neglected . in the light of these considerations
it is not unexpected that the gaussian version of the model mimics the essential features of the field dependence of the magnetization shown in fig .
[ fig1 ] well , namely the occurrence of the minimum between the low and high field behavior .
the same qualitative agreement also emerges from the magnetization data of la@xmath111sr@xmath22cuo@xmath9 with @xmath113 , @xcite @xmath114,@xcite , bi2212 with @xmath115 k , underdoped bi2212 with @xmath116 k,@xcite optimally and overdoped bi2201la@xmath25 with @xmath117 k and @xmath118 k,@xcite , and pb particles.@xcite to substantiate this qualitative agreement we explore in sec .
iii the consistency of the measured isothermal magnetization curves of pb,@xcite la@xmath111sr@xmath22cuo@xmath9 with @xmath119 and bi2212@xcite with the outlined scenario for a zero dimensional system . because the partition function ( [ eq9 ] ) requires that @xmath71 and with that according to eqs .
( [ eq8 ] ) and ( [ eq12 ] ) @xmath120 our analysis is essentially restricted to temperatures @xmath32 .
one also expects that the diamagnetic contribution to the magnetization exhibits a characteristic doping dependence .
indeed , the phase transition line of la@xmath111sr@xmath22cuo@xmath9 is well described by the empirical relation @xmath121 due to presland _
et al_.@xcite at @xmath122 the systems are expected to undergo a quantum phase transition . here
the amplitude of the correlation length @xmath69 diverges in a homogeneous system as@xcite @xmath123 while @xmath0 scales according to @xmath124 @xmath125 denotes the tuning parameter of the quantum phase transition with dynamic critical exponent @xmath42 and correlation length exponent @xmath126 . combining eqs .
( [ eq23 ] ) and ( [ eq24 ] ) , we obtain @xmath127 expected to apply in both , the underdoped ( @xmath128 ) and overdoped ( @xmath129 ) limits . noting that @xmath69 enters @xmath130 [ eqs .
( [ eq12 ] ) and ( [ eq21 ] ) ] , the magnetic field @xmath131 , where the isothermal magnetization curves exhibit a minimum , the approach to the underdoped or overdoped limit should be observable .
however , this behavior may be masked by means of the doping dependence of @xmath62 , the radius of the homogeneous cylindrical domains .
in this section we explore the consistency of isothermal magnetization and nernst signal data with the predictions of the full 0d - model and its gaussian version .
we concentrate on the magnetization data of pb nanoparticles,@xcite la@xmath7sr@xmath8cuo@xmath9,@xcite and bulk bisr@xmath10cacu@xmath10o@xmath36 ( bi2212 ) with @xmath38 k ( slightly underdoped ) and @xmath37 k ( heavily underdoped),@xcite and on nernst signal data of a nb@xmath4si@xmath4 film.@xcite the diamagnetism in pb nanoparticles with average diameters ranging from @xmath132 to @xmath133 , sizes for which effects of finite - level spacing should be negligible , has been studied by benardi _
et al_.@xcite by means of high - field resolution superconducting quantum interference device ( squid ) measurements , isothermal magnetization curves were obtained for @xmath134 .
[ fig2 ] shows isothermal diamagnetic magnetization curves of the sample containing spherical nanoparticles with average radius @xmath135 . for comparison we included fits to the gaussian approximation eq .
( [ eq20 ] ) with the parameters listed in table [ table1 ] and at @xmath136 k to the @xmath137d - model [ eq .
( [ eq13 ] ) ] yielding the parameters @xmath138 with @xmath139 k in terms of the dotted line . to account for a temperature dependent background contribution we added to eqs .
( [ eq13 ] ) and ( [ eq20 ] ) the parameter @xmath140 , which turns out to be rather small .
the agreement of the @xmath137d - model with the gaussian approximation at @xmath136 k also reveals that for @xmath141 the limit @xmath142 is nearly attained .
indeed , the fit parameters @xmath87 , @xmath143 and @xmath140 as obtained from the 0d - model [ eq .
( [ eq25a ] ) ] and the gaussian counterpart ( see table [ table1 ] ) coincide nearly at @xmath136 k. _
vs_. @xmath14 for pb nanoparticles with average radius @xmath135 taken from benardi _
et al_.@xcite the dashed lines are fits to the gaussian approximation eq .
( [ eq20 ] ) with the parameters listed in table [ table1 ] . the dotted line at @xmath136 k is a fit to eq .
( [ eq13 ] ) with the parameters listed in table [ table2 ] with @xmath139 k. the stars indicate the respective minima of the experimental data at @xmath18 . ] [ cols="^,^,^,^,^",options="header " , ] in fig .
[ fig12 ] we plotted @xmath35 _
vs_. @xmath144 and included a fit to eq .
( [ eq21 ] ) yielding @xmath145 t , in reasonable agreement with the estimates listed in table [ table5 ] .
it then follows that @xmath146 in comparison with @xmath147 @xmath148 for pb corresponding to @xmath149 oe , @xmath150 @xmath148 for la@xmath111sr@xmath22cuo@xmath9 with @xmath19 k [ eq .
( [ eq27 ] ) ] , @xmath151 @xmath148 in bi2212 with @xmath152 k , and @xmath153 @xmath148 with @xmath154 k. to estimate the radius @xmath62 of the homogeneous domains we invoke @xmath155 ,@xcite to obtain @xmath156 in comparison with a limiting lateral length of @xmath157 , obtained from a detailed finite size scaling analysis of the magnetic field dependence of the conductivity in a @xmath158 thick nb@xmath3si@xmath4
film.@xcite this limiting length also implies that the evidence for a magnetic field driven quantum phase transition in this system is constricted by the resulting finite size effect.@xcite _ vs. _
@xmath144 for the nb@xmath3si@xmath4 film with @xmath159 k derived from the data shown in fig .
[ fig11 ] .
the solid line is eq .
( [ eq19b ] ) with @xmath145 t. ] remarkably , the gaussian version of the 0d - model describes the profile of the isothermal nernst signal above @xmath0 and the temperature dependence of @xmath35 in terms of @xmath5 very well . in the low field limit
this relationship transforms with eq .
( [ eq20 ] ) to @xmath160 which differs from the gaussian fluctuation contribution @xmath161 , valid close to @xmath0 and in the zero magnetic field limit.@xcite to substantiate the neglect of thermal fluctuations in the 0d - model further we invoke the ginzburg criterion for a 2d - system , @xmath162 .
@xmath163 is the range of temperatures where thermal fluctuations are essential and @xmath164 is the is the ginzburg - levanyuk parameter for a dirty film with normal state resistance @xmath165.@xcite with @xmath166 k@xmath167@xcite and @xmath168 k we obtain @xmath169 k. as a result the nernst signal curves shown in fig .
[ fig11 ] were taken outside the critical regime where fluctuations dominate .
noting that in cuprate and amorphous conventional superconductors the spatial extent of the homogeneous domains is limited,@xcite we explored the applicability of the 0d - model , neglecting thermal fluctuations , to describe the isothermal magnetization and nernst signal curves above @xmath0 . sufficiently above @xmath0
we observed that for both models , the full 0d - model and its gaussian version , describe the essential features of the curves , including the temperature dependence of the minimum in the magnetization and maximum in the nernst signal curves at @xmath35 , rather well .
the essential difference between the magnetization curves of the pb nanoparticles and the bulk cuprates was traced back to the product between the amplitude @xmath69 of the correlation length and the radius @xmath62 of the spatial restriction .
indeed , the magnitude of @xmath170ln@xmath171 is controlled by the amplitude @xmath172 .
it adopts in the pb nanoparticles the value @xmath173
@xmath174 compared to @xmath175 in the @xmath176 thick nb@xmath3si@xmath4 film , and @xmath177 @xmath174 in the cuprates considered here . indeed , as shown in fig
. [ fig13 ] , the data for @xmath178 vs. @xmath144 , depicted in figs . [ fig3 ] , [ fig5 ] , [ fig7 ] , [ fig9 ] , and [ fig12 ] for pb nanoparticles , la@xmath7sr@xmath8cuo@xmath9 , bi2212 and nb@xmath3si@xmath4 , tend to fall on a single curve , plotted as @xmath179__vs__. @xmath180 .
this curve is even well described by the gaussian version of the 0d - model [ eq .
( [ eq21 ] ) ] .
thus , for a variety of conventional and hole doped cuprate superconductors it gives a universal perspective on the interplay between diamagnetism , nernst signal , correlation length , and the limited spatial extent of homogeneity . _ _
@xmath180 for the data shown in figs .
[ fig3 ] , [ fig5 ] , [ fig7 ] , [ fig9 ] , and [ fig12 ] . the solid curve is @xmath181 corresponding to the gaussian approximation [ eq . ( [ eq21 ] ) ] . ]
although the assumption of an order parameter which does not depend on the spatial variables fails in the fluctuation dominated temperature window close to @xmath0 , we established overall agreement between the 0d - model , the isothermal magnetization and the nernst signal treated as @xmath5 . as a consequence , thermal fluctuations associated with the amplitude and the phase of the order parameter
do not contribute significantly in the temperature and magnetic field regimes considered here .
the agreement also implies that singlet cooper pairs in a 0d system subjected to orbital pair breaking are the main source of the observed diamagnetism and nernst signal in an extended temperature window above @xmath0 .
the monotonic decrease of the magnetization @xmath182 and the nernst signal @xmath1 with magnetic field @xmath14 also reveals that there is no particular depairing field .
indeed , @xmath1 and @xmath183 vanish as @xmath184 [ eq . ( [ eq20 ] ) ] . noting that @xmath185 also holds for gaussian fluctuations close to @xmath0 and in the zero magnetic field limit,@xcite we have shown that it applies even outside the fluctuation dominated regime .
clearly , the outlined approach can not distinguish between intrinsic and extrinsic inhomogeneities , or whether the detected restricted extent of the homogeneous regions reflect an intimate relationship to superconductivity .
however , it implies that the reduced dimensionality is not only responsible for the smeared zero field transitions seen in the specific heat,@xcite , in the temperature dependence of the magnetic penetration depths,@xcite , and the resistive transition,@xcite but also accounts for the observed characteristic minimum in the isothermal magnetization curves and the corresponding maximum in the nernst signal .
the authors acknowledge stimulating and helpful discussions with h. keller .
this work was partly supported by the swiss national science foundation .
s. h. pan , j. p. oeal , r. l. badzey , c. chamon , h. ding , j. r. engelbrecht , z. wang , h. eisaki , s. uchida , a. k. guptak , k. w. ng , e. w. hudson , k. m. lang , and j. c. davis , nature ( london ) * 413 * , 282 ( 2001 ) .
v. v. shmidt , in _ proceedings of the ixx international conference on low temperature physics _ , edited by m. p. malkov , l. p. pitaesvski , and a. shalnikov ( viniti publishing house , moscow , 1967 ) , vol .
ii b , p. 205 . | various superconductors , including cuprate superconductors , exhibit peculiar features above the transition temperature @xmath0 . in particular the observation of a large diamagnetism and nernst signal @xmath1 in a wide temperature window above @xmath0 attracted considerable attention . noting that this temperature window exceeds the fluctuation dominated regime drastically and that in these materials the spatial extent of homogeneity is limited , we explore the relevance of the zero dimensional ( 0d)-model , neglecting thermal fluctuations .
it is shown that both , the full 0d - model as well as its gaussian approximation , mimic the essential features of the isothermal magnetization curves @xmath2 in pb nanoparticles and various cuprates remarkably well .
this analysis also provides estimates for the spatial extent of the homogeneous domains giving rise to a smeared transition in zero magnetic field .
the resulting estimates for the amplitude of the in - plane correlation length exhibit a doping dependence reflecting the flow to the quantum phase transition in the underdoped limit .
furthermore it is shown that the isothermal nernst signal of a superconducting nb@xmath3si@xmath4 film , treated as @xmath5 , is fully consistent with this scenario .
accordingly , the observed diamagnetism above @xmath6 in pb nanoparticles , in the cuprates la@xmath7sr@xmath8cuo@xmath9 and bisr@xmath10ca@xmath10cuo@xmath11 , as well as the nernst signal in nb@xmath3si@xmath4 films , are all in excellent agreement with the scaling properties emerging from the 0d - model , giving a universal perspective on the interplay between diamagnetism , nernst signal , correlation length , and the limited spatial extent of homogeneity .
our analysis also provides evidence that singlet cooper pairs subjected to orbital pair breaking in a 0d system are the main source of the observed diamagnetism and nernst signal in an extended temperature window above @xmath0 . |
a _ wavelet set _ relative to dilation by an expansive ( all eigenvalues greater than 1 in absolute value ) real @xmath3 matrix @xmath4 is a set @xmath5 whose characteristic function @xmath6 is the fourier transform of an orthonormal wavelet .
that is , if @xmath7 then @xmath8 is an orthonormal basis for @xmath9 .
this definition is equivalent to the requirement that the set @xmath10 tiles @xmath11dimensional space ( almost everywhere ) both under translation by @xmath12 and under dilation by the transpose @xmath13 , so that @xmath14 @xmath15 while wavelet set wavelets are not well - localized , and thus not directly useful for applications , they have proven to be an essential tool in developing wavelet theory . in particular , wavelet set examples established that not all wavelets have an associated mra @xcite , and that single wavelets exist for an arbitrary expansive matrix in any dimension @xcite . smoothing and interpolation techniques have also used wavelet set wavelets to produce more well - localized examples .
( see e.g. @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite . )
all of the early examples of wavelet sets for dilation by non - determinant 2 matrices in dimension greater than 1 were geometrically complicated , showing the fingerprints of the infinite iterated process used to construct them .
( see e.g. figure [ dim2](a ) ) .
many early researchers , e.g.@xcite , @xcite , conjectured that a wavelet set for dilation by 2 in dimension greater than 1 could not be written as a finite union of convex sets . in support of this conjecture ,
benedetto and sumetkijakan @xcite showed that a wavelet set for dilation by 2 in @xmath0 can not be the union of @xmath16 or fewer convex sets .
however , in 2004 , gabardo and yu @xcite used self - affine tiles to produce a wavelet set for dilation by 2 in @xmath17 that is a finite union of polygons ( figure [ dim2](b ) ) . in 2008 @xcite we used a technique based on generalized multiresolution analyses @xcite to construct such wavelet sets for arbitrary real ( @xmath18 ) scalar dilations in @xmath17 .
figure [ dim2](c ) shows one of the wavelet sets for dilation by 2 from @xcite . although they were developed independently , and using very different techniques , these two examples are remarkably similar .
in fact , the wavelet sets in figure [ dim2](b ) and [ dim2](c ) are equivalent in the sense that one can be transformed into the other under multiplication by a determinant 1 integer matrix .
the similar shape of these two wavelet sets suggests the general @xmath16-dimensional result produced in this paper .
[ dim2 ] @xmath19{holec.pdf } } \end{picture } & \setlength{\unitlength}{100bp } \begin{picture}(1,1)(.0,-.2 ) \put(0,0){\includegraphics[width=\unitlength]{gabyu.pdf } } \end{picture } & \setlength{\unitlength}{120bp } \begin{picture}(1,0.0)(-.2,-.2 ) \put(0,0){\includegraphics[width=\unitlength]{unsymw.pdf } } \end{picture}\\ \mbox{soardi / wieland 1998\quad\quad\quad}&\mbox{gabardo / yu 2004}&\mbox{\quad\quad merrill 2008}\\ \end{array}$ ] we call wavelet sets that are finite unions of convex sets _ simple wavelet sets_. in 2012 @xcite , we expanded the results in @xcite to produce simple wavelet sets for dilation by any @xmath20 matrix that has a positive integer power equal to a scalar times the identity , as long as its singular values are all greater than @xmath21 . in that paper , we also found examples of expansive @xmath20 matrices that can not have simple wavelet sets .
it is our conjecture that , in any dimension , an expansive matrix whose determinant does not have absolute value equal to 2 can have a simple wavelet set if and only if it has a positive integer power equal to a scalar times the identity . in this paper , we generalize the 2-dimensional examples in @xcite to @xmath16-dimensional space , @xmath1 . we do this using neither the generalized multi - resolution analysis techniques of @xcite , nor the self - affine techniques of @xcite .
rather , we use a remarkable result by sherman stein @xcite on tiling @xmath0 with notched cubes , together with the tiling conditions that are equivalent to the definition of a wavelet set .
section 2 presents stein s result , and then skews and translates the notched n - cubes to produce notched parallelotopes that are simple wavelet sets for dilation by negative scalars . section 3 further modifies these notched parallelotopes by translating out a central parallelotope ( as in figure [ dim2](b ) and [ dim2](c ) ) . using this technique , theorem [ main ] creates simple wavelet sets for dilation by any scalar @xmath22 .
this result establishes counterexamples , in every dimension greater than 1 , to the conjecture that wavelet sets for dilation by 2 can not be finite unions of convex sets .
these counterexamples are composed of @xmath23 convex sets for dimension @xmath16 , as compared to the lower bound of @xmath24 given in the benedetto / sumetkijakan result mentioned above .
theorem [ matrix ] generalizes theorem [ main ] to dilation by matrices that have a positive integer power equal to a scalar , as long as their singular values are not too small .
one consequence of this theorem is to create simple wavelet sets for dilation by a scalar @xmath25 with @xmath26 , thus completing the scalar dilation case of the existence question for simple wavelet sets . for non - scalar dilations in dimension 3 and higher , theorem [ matrix ] offers support to the sufficiency direction of the conjecture above concerning exactly which matrices have associated simple wavelet sets . the examples that end section 3 further support this conjecture by showing that the theorem s additional condition on singular values need not always hold for matrices that have simple wavelet sets .
we begin by establishing some notation .
write @xmath27 for the standard basis of @xmath0 , and @xmath28 for the cyclic permutation matrix with columns @xmath29 .
let @xmath30 stand for the vector @xmath31 , and write @xmath32 for translation by @xmath33
. given a vector @xmath34 in @xmath0 that is not a multiple of @xmath30 , let @xmath35=\{x_0v+x_1c(v)+\dots x_{n-1}c^{n-1}v\;:\;0\leq x_i\leq 1\}\ ] ] be the parallelotope spanned by the vectors @xmath36 .
note that the spanning vectors of @xmath37 $ ] have equal length , since they are permutations of the same vector @xmath38 .
in particular , when @xmath39 , @xmath37 $ ] is a rhombus .
note also that @xmath37 $ ] has two vertices on the line determined by @xmath30 : one at the origin , and the other determined by the sum of the coordinates of @xmath38 , at @xmath40 , where the vector given an @xmath41 , @xmath42 , write @xmath43 $ ] for the notched parallelotope that results from deleting a subparallelotope scaled by @xmath41 from the vertex @xmath44 of @xmath37 $ ] .
that is , let @xmath45=\mathcal p[v]\setminus\;\tau_{(1-\alpha)(\sum v_i)\vec 1}\;\alpha \mathcal p[v],\ ] ] or equivalently , @xmath46=\mathcal p[v]\setminus\;\tau_{(\sum v_i)\vec 1}\;\left(-\alpha \mathcal p[v]\right).\ ] ] we will need the following result about translation tilings by notched cubes due to sherman stein .
[ notchedcube ] given a real number @xmath42 , let @xmath47 be the lattice spanned by the columns of @xmath48 , where @xmath28 is the cyclic permutation matrix .
then the translates of the notched unit cube @xmath49 $ ] by the vectors in @xmath47 tile @xmath0 .
see @xcite we use this result to produce a notched parallelotope that tiles @xmath0 under translation by the lattice @xmath12 : [ notchedpar ] for a fixed real number @xmath41 , @xmath42 , let @xmath50 .
then the translates of @xmath51 $ ] by @xmath12 tile @xmath0 . by lemma [ notchedcube ]
we know that @xmath49 $ ] tiles @xmath0 under translation by @xmath52 the lattice spanned by the columns of @xmath48 .
if we define @xmath4 to be the linear transformation that maps @xmath47 to @xmath12 , we thus have that that @xmath53)$ ] tiles @xmath0 under translation by @xmath12 .
note that @xmath54 so that @xmath55 .
thus , @xmath56\right)&=&a\left(\mathcal p[e_1]\setminus\;\tau_{(1-\alpha)\vec 1}\;\alpha \mathcal p[e_1]\right)\\ & = & \mathcal p[a(e_1)]\setminus\;\tau_{(1-\alpha)a\vec 1}\;\alpha\mathcal p[a(e_1)]\\ & = & \mathcal p[w(\alpha)]\setminus\tau_{\vec 1}\;\alpha \mathcal p[w(\alpha)]\\ & = & \mathcal { n}[w(\alpha ) , \alpha]\end{aligned}\ ] ] to be a wavelet set , a notched parallelotope would have to also tile under dilation . for a scalar dilation @xmath25 ,
this is clearly impossible , since a notched parallelotope is defined to have an extreme point at the origin .
thus , we consider instead the translated notched parallelotope @xmath57 $ ] , for @xmath33 .
for such a set to tile under dilation by @xmath25 would require that the dilated outer parallelotope , @xmath58 $ ] , fit perfectly into the notch , which is of the form @xmath59 $ ] .
first note that this would force the scale of the dilated outer parallelotope to match the scale of the notch , so that @xmath60 . for positive @xmath18 , this perfect fit also would require that @xmath61 so that @xmath62 .
however , the outer parallelotope @xmath63 $ ] has one of its extreme points at @xmath64_j = t+\frac{d}{d-1}\vec 1 $ ] .
thus , using the required value for the translation @xmath65 would again cause the outer parallelotope to have an extreme point at the origin .
hence a set of the form @xmath57 $ ] can not tile by a positive scalar dilation .
however , for negative scalar dilations , a wavelet set that is just a notched parallelotope is possible , as the following theorem shows .
the examples produced by theorem [ negdilate ] generalize the wavelet sets for negative scalar dilations in @xmath17 found in @xcite and @xcite .
[ negdilate ] for @xmath66 , @xmath18 , let @xmath67 , and @xmath68 .
then @xmath69\ ] ] is a wavelet set for dilation by @xmath70 in @xmath0 . we know from lemma [ notchedpar ] that the notched parallelotope @xmath71 $ ] tiles @xmath0 under translation by @xmath12 , and thus that its translate @xmath72 does as well .
it remains to show that @xmath72 tiles under dilation by @xmath70 .
the proposed wavelet set @xmath73 $ ] has its vertices on the line determined by @xmath30 at @xmath74 and @xmath75 , while its outer parallelotope @xmath76 $ ] has its vertices at @xmath74 and @xmath77 .
( see figure [ neg]a . )
if we consider now the dilate by @xmath78 of these two polytopes , we see that @xmath79 has its vertices on the line determined by @xmath30 at @xmath80 and @xmath81 , while @xmath82 $ ] has its vertices at @xmath80 and @xmath83 .
( see figure [ neg]b . )
[ neg ] @xmath84{negdil.pdf } } \end{picture } & \setlength{\unitlength}{200bp } \begin{picture}(1.0,1)(0,0 ) \put(0,0){\includegraphics[width=\unitlength]{negdilb.pdf } } \end{picture}\\ \hskip.5in\mbox{(a)}\quad\mathcal w & \hskip.5 in \mbox{(b)}\quad -\frac 1d \mathcal w \end{array}$ ] using the value @xmath85 given in the statement of the theorem , we see that @xmath86 and @xmath87 , so that the dilation by @xmath78 of the outer parallelotope used to form @xmath72 exactly fits into the notch of @xmath72 .
thus , @xmath88\setminus\left(\tau_{(t+1)\vec
1}\;\frac{1}{d^2}\mathcal p\left[w\left(\frac 1d\right)\right]\right)\\ & = & \tau_{t\vec 1}\mathcal p\left[w\left(\frac 1d\right)\right]\setminus\frac{1}{d^2}\left(\tau_{t\vec 1}\mathcal p\left[w\left(\frac 1d\right)\right]\right),\end{aligned}\ ] ] which tiles under dilation by @xmath89
. therefore , @xmath72 tiles under dilation by @xmath70 .
figure [ neg]a shows the general shape of a wavelet set @xmath72 for dilation by @xmath70 in 3 dimensions .
the size of the notch shown is for @xmath90 ; in general the size of the notch will be @xmath91 times the size of the whole .
for dilation by @xmath70 , the lower left hand vertex of @xmath72 will be at @xmath92 , the inside corner of the notch at @xmath93 , and the outer corner of the notch at @xmath94
even though we have seen that a translate of a notched parallelotope @xmath95 $ ] can not itself tile under dilation by a positive scalar , we will make a simple alteration to such a set that retains the property of tiling under translation , and makes the set tile under dilation as well .
we use the idea behind the wavelet set construction technique in @xcite .
that is , we eliminate the overlap between our proposed wavelet set and its dilate , by translating the dilate out by an integer vector , creating a satellite . in order to avoid the iterated process required in @xcite
, we translate out a little bigger piece than the dilate of @xmath95 $ ] ; that is , we translate out the dilate of the whole outer parallelotope @xmath96 $ ] .
we choose the translation amount such that a dilate of the satellite exactly fills the notch .
the details of this construction are carried out in the following theorem .
[ main ] for @xmath66 , @xmath22 , let @xmath97 .
suppose @xmath98 satisfies @xmath99 , and let @xmath100 then @xmath101\right)\;\setminus\;\left(\frac 1d\tau_{t\vec 1}\ ; \mathcal p\left[w\left(\frac1{d^2}\right)\right]\right)\right)\;\bigcup\;\tau_{k\vec 1}\left(\frac1d\tau_{t\vec 1}\;\mathcal p\left[w\left(\frac1{d^2}\right)\right]\right)\ ] ] is a wavelet set for dilation by @xmath25 in @xmath102 we claim that @xmath103\subset\tau_{t\vec 1 } \;\mathcal n\left[w(\frac1{d^2 } ) , \frac 1{d^2}\right]$ ] . to see this , first note that @xmath104 and @xmath105_i,$ ] so that @xmath106\subset\tau_{t\vec 1}\ ; \mathcal p\left[w\left(\frac1{d^2}\right)\right]$ ] .
thus , to establish the claim , it will suffice to show that the vertex of the notch that is closest to the origin , namely @xmath107_i)\vec 1 $ ] , lies outside of @xmath108 $ ] .
that is , we must show that @xmath109 . substituting @xmath110
we see that this is equivalent to @xmath111 which follows from the given conditions @xmath112 and @xmath22 .
lemma [ notchedpar ] impies that @xmath113 $ ] tiles under translation by the integer lattice , and thus that @xmath114 $ ] does as well .
by the claim , we have that @xmath72 is formed by removing @xmath115 $ ] from inside @xmath114 $ ] , and shifting it by a vector of the integer lattice . thus @xmath72 tiles @xmath0 under translation by @xmath12 . to establish tiling under dilation by @xmath25 ,
we first show that @xmath116 $ ] is disjoint from @xmath117\right)$ ] .
that is , we must show that @xmath118 , which follows from the definition of @xmath65 together with the condition @xmath112 .
now , note that @xmath119\setminus\frac1d\tau_{t\vec 1 } \;\mathcal p\left[w\left(\frac1{d^2}\right)\right]$ ] tiles @xmath0 under dilation by @xmath25 .
the definition of @xmath65 together with the disjointness of the pieces of @xmath72 shows that @xmath72 is formed from this set by dilating the notch @xmath120\right)$ ] by @xmath25 .
thus , @xmath72 also tiles @xmath0 under dilation by @xmath25 .
see figure 9.7 of @xcite for illustrations of these tilings under both translation and dilation in the case @xmath39 .
[ var]the wavelet sets produced by theorem [ main ] are a natural generalization of the 2 dimensional simple wavelet sets produced for scalar dilations in @xcite .
we can also alter these examples to produce natural generalizations of the 2-dimensional example for dilation by 2 that appears in @xcite .
note that if @xmath72 is a wavelet set for dilation by the scalar @xmath25 , then so is @xmath121 for any integer matrix @xmath122 of determinant @xmath123 .
if we take @xmath122 to be the @xmath3 matrix that has 1 s on the diagonal , -1 s on the subdiagonal and 0 s elsewhere , then @xmath121 is a simple wavelet set for dilation by @xmath25 in dimension @xmath16 that is centered on the @xmath124 axis rather than the line @xmath125 .
other variations are easily produced using other choices for the matrix @xmath122 .
figure [ 3d ] shows one of the wavelet sets produced by theorem [ main ] for dilation by 2 in @xmath126 , as well as the variation described in remark [ var ] .
[ 3d ] @xmath127{dil2wav3d.pdf } } \end{picture } & \setlength{\unitlength}{150bp } \begin{picture}(1.0,1)(-.2,0 ) \put(0,0){\includegraphics[width=\unitlength]{dil2wav3dx.pdf } } \end{picture } \end{array}$ ] the final theorem uses the same technique as theorem [ main ] to produce simple wavelet sets for dilation by an expansive matrix @xmath4 that has an integer power @xmath128 equal to a scalar multiple @xmath25 of the identity . in this case , the notched parallelotope @xmath129 $ ] fails to tile under dilation by @xmath13 because it is formed from the outer parallelotope @xmath130 $ ] by removing its dilate by @xmath131 instead of its dilate by the matrix @xmath132 .
we would like to remedy the resulting overlap of @xmath129 $ ] and all of its negative dilates @xmath133 $ ] , by translating @xmath134 $ ] out by an appropriately chosen integer vector , to form a satellite that will be clear of @xmath129 $ ] and will perfectly dilate into the notch .
this is possible only if @xmath135 $ ] is completely contained in @xmath136 $ ] .
( for an example where this containment does not hold , see figure 5(a ) . ) to overcome this difficulty , we put a restriction on @xmath4 , and sometimes replace @xmath131 with a higher power @xmath137 . [ matrix ] let @xmath4 be an @xmath3 integer matrix such that @xmath138)$ ] , where @xmath18 , @xmath139 , @xmath140 , and @xmath141 $ ] is the @xmath3 identity matrix .
suppose further that all of the singular values of @xmath4 are greater than @xmath2 .
let @xmath142 .
let @xmath143 be the closest vector in @xmath12 to @xmath144 , and let @xmath145 then for @xmath146 sufficiently large , @xmath147\right)\subset\tau_t\mathcal n\left[w(\frac 1{d^q}),\frac 1{d^q}\right].$ ] for such @xmath148 @xmath149\right)\;\setminus\;\left(a^{*-1}\tau_{t}\ ; \mathcal p\left[w\left(\frac 1{d^q}\right)\right]\right)\right)\;\bigcup\;\left(\tau_{k}\left(a^{*-1}\tau_{t}\;\mathcal p\left[w\left(\frac 1{d^q}\right)\right]\right)\right)\ ] ] is a wavelet set for dilation by @xmath4 in @xmath102 by lemma [ notchedpar ] , @xmath150 $ ] , tiles under translation by @xmath12 , and thus its translate by @xmath65 does as well .
thus , @xmath72 will also tile by translation as long as the set @xmath151 $ ] , which is translated out by the integer vector @xmath143 , is a subset of @xmath152 $ ] , and moves to a position disjoint from @xmath152 $ ] . as @xmath153 , the vector @xmath154 and @xmath155 ,
so that the set @xmath156 $ ] approaches the unit @xmath16-cube centered at the origin .
thus , by taking @xmath146 sufficiently large , we can make the longest vector in @xmath156 $ ] to be arbitrarily close to @xmath2 times as long as the shortest vector .
then , since the singular values of @xmath4 are greater than @xmath2 , we will have @xmath147\right)\subset\tau_t \mathcal p\left[w(\frac 1{d^q})\right]$ ] .
the size of the notch also shrinks to 0 as @xmath153 , so that by taking @xmath146 larger if necessary , we will also have @xmath147\right)\subset\tau_t\mathcal n\left[w(\frac 1{d^q}),\frac 1{d^q}\right]$ ] . by the definition of @xmath143 ,
@xmath157 $ ] and @xmath158\right)$ ] are clearly disjoint for large @xmath146 . to establish tiling under dilation ,
note that for @xmath146 large enough that @xmath147\right)\subset\tau_t\mathcal n\left[w(\frac 1{d^q}),\frac 1{d^q}\right]$ ] , we have that @xmath159\setminus\left(a^{*-1}\tau_{t}\ ; \mathcal p[w(\frac 1{d^q})]\right)$ ] tiles @xmath0 under dilation by @xmath160 . to show that @xmath72 tiles under dilation as well
, we must show that the outlier piece of @xmath72 exactly fits into the notch of @xmath161 $ ] under dilation by some integer power of @xmath13 .
we have @xmath162\right)\right)&=&a^{*(-pq)}\left(\tau_{a^*k+t}\;\mathcal p\left[w\left(\frac 1{d^q}\right)\right]\right)\\ & = & \frac 1{d^q}\left(\tau_{a^*k-(d^q-1)t+d^qt}\;\mathcal p\left[w\left(\frac 1{d^q}\right)\right]\right)\\ & = & \frac
1{d^q}\left(\tau_{d^q\vec1+d^qt}\;\mathcal p\left[w\left(\frac 1{d^q}\right)\right]\right)\\ & = & \tau_{t+\vec 1}\left(\frac1{d^q}p\left[w\left(\frac 1{d^q}\right)\right]\right ) , \end{aligned}\ ] ] which is exactly the notch of @xmath163 $ ] .
thus we have that @xmath72 also tiles @xmath0 under dilation by @xmath13 .
theorem [ matrix ] also produces simple wavelet set for scalar dilations
@xmath26 , which were not covered by theorem [ main ] . for scalar dilations by @xmath22 , theorem [ matrix ] produces a series of alternative wavelet sets to those of theorem [ main ] .
as @xmath146 increases in this series , the parallelotope becomes closer to cubic , the notch becomes smaller , and the satellite becomes farther removed .
[ mat1 ] let @xmath164
. then @xmath165 and @xmath166 , so theorem [ matrix ] applies . with @xmath167 , @xmath168 and @xmath169
, we have @xmath170\subset\tau_t \mathcal p\left[w(\frac 19)\right]$ ] .
thus , we have a simple wavelet set @xmath171\right)\;\setminus\;\left(a^{*-1}\tau_{t}\;\mathcal p\left[w\left(\frac 19\right)\right]\right)\;\bigcup\;\left(\tau_{(1,1,-1)}\left(a^{*-1}\tau_{t}\;\mathcal p\left[w\left(\frac 19\right)\right]\right)\right),\ ] ] which is pictured in figure 4 . [ matrix1 ] ( 1,1)(.1,0 ) ( 0,0 ) of example [ mat1].,title="fig : " ]
the hypothesis in theorem [ matrix ] that the singular values of @xmath4 be greater than @xmath2 is sufficient but not necessary , as the next example shows .
[ mat ] let @xmath172 .
then @xmath173 and @xmath174 , so theorem [ matrix ] does not apply . with @xmath175 , @xmath176 and @xmath177 , we do not have @xmath178\subset\tau_t \mathcal p\left[w(\frac 14)\right]$ ] .
( see figure 5(a ) . )
however , using @xmath179 , with @xmath180 , @xmath181 , and @xmath182 , the required containment does hold , yielding the wavelet set @xmath183\right)\;\setminus\;\left(b^{*-1}\tau_{t}\ ; \mathcal p\left[w\left(\frac 1{16}\right)\right]\right)\;\bigcup\;\left(\tau_{k}\left(b^{*-1}\tau_{t}\;\mathcal p\left[w\left(\frac 1{16}\right)\right]\right)\right).\ ] ] figure 5(b ) shows the central part of the wavelet set .
( the complete wavelet set also includes a translation of the missing inner parallelotope by @xmath184 . )
[ mat ] @xmath185{mat1.pdf } } \end{picture } & \setlength{\unitlength}{130bp } \begin{picture}(1.0,1)(0,0 ) \put(0,0){\includegraphics[width=\unitlength]{mat2.pdf } } \end{picture}\\ \mbox{(a ) } b^{*-1}\mathcal p\left[w\left(\frac14\right)\right]\not\subset\mathcal p\left[w\left(\frac14\right)\right]\qquad\qquad&\mbox{(b ) } \mathcal n\left[w\left(\frac1{16}\right),\frac1{16}\right]\setminus b^{*-1}\mathcal p\left[w\left(\frac1{16}\right)\right ] \end{array}$ ]
the author wishes to thank the referee for several suggestions that clarified the presentation .
k. d. merrill ( 2008 ) .
simple wavelet sets for scalar dilations in @xmath190 . in , _ wavelets and frames : a celebration of the mathematical work of lawrence baggett _
( p. jorgensen , k. merrill and j. packer eds . ) , birkhauser , boston , pp.177 - 192 . | wavelet sets that are finite unions of convex sets are constructed in @xmath0 , @xmath1 , for dilation by any expansive matrix that has a power equal to a scalar times the identity and also has all singular values greater than @xmath2 . in particular , we produce simple wavelet sets in every dimension for dilation by any real scalar greater than 1 . |
the interactions of surfaces are often affected by nanoparticles or macromolecules in the surrounding medium .
non - adhesive particles cause attractive depletion interactions between the surfaces , since the excluded volume of the molecules depends on the surface separation @xcite .
adhesive particles , on the other hand , can directly bind two surfaces together if the surface separation is close to the particle diameter @xcite . in a recent letter @xcite
, we have presented a general , statistical - mechanical model for two surfaces in contact with adhesive particles . in this model ,
the space between the surfaces is discretized into columns of the same diameter @xmath0 as the particles .
the approximation implied by this discretization is valid for small bulk volume fractions of the particles , since three - dimensional packing effects relevant at larger volume fractions are neglected . for short - ranged particle - surface interactions ,
the gas of particles between the surfaces is as dilute as in the bulk for large surface separations , except for the single adsorption layers of particles at the surfaces . in this article , we present an exact solution of the one - dimensional gas of hard - sphere particles in a single column between two ` surfaces ' .
our aim here is two - fold .
first , the exact solution presented here corroborates our previous , approximate solution for this one - dimensional gas obtained from a virial expansion in the particle concentration @xcite .
second , the exactly solvable , one - dimensional model considered here is a simple toy model to study the interplay of surface adhesion and particle adsorption . exactly solvable ,
one - dimensional models have played an important role in statistical mechanics @xcite .
one example is the kac - baker model @xcite , which has shed light on the statistical origin of phase transitions of the classical van der waals type .
more recent examples are models for one - dimensional interfaces , or strings , which have revealed the relevance of entropy and steric interactions in membrane unbinding and wetting transitions @xcite .
other examples are the tonks model @xcite and its various generalizations @xcite , which have influenced our understanding of the relations between short - ranged particle interactions , thermodynamics , and statistical correlations in simple fluids . the tonks model has been exploited also in soft - matter physics to investigate structures of confined fluids @xcite , depletion phenomena in two - component mixtures @xcite , thermal properties of columnar liquid crystals @xcite and the phase behavior of polydisperse wormlike micelles @xcite .
a recent biophysical modification of the tonks model addresses the wrapping of dna around histone proteins @xcite .
the model considered here is a novel extension of the tonks model . in our model , a one - dimensional gas of hard - sphere particles
is attracted to the system boundaries , or ` surfaces ' , by short - ranged interactions .
we calculate the effective , particle - mediated interaction potential between the surfaces , @xmath1 , by explicit integration over the particles degrees of freedom in the partition function .
the potential @xmath1 is a function of the surface separation @xmath2 and exhibits a minimum at surface contact , which reflects depletion interactions , and a second minimum at separations close to the diameter of the adhesive particles .
the effective , particle - mediated adhesion energy of the surfaces , @xmath3 , can be determined from the interaction potential @xmath1 .
the adhesion energy is the minimal work that has to be performed to bring the surfaces apart from the equilibrium state corresponding to the deepest well of the potential @xmath4 .
interestingly , the adhesion energy @xmath3 attains a maximum value at an optimal particle concentration in the bulk , and is considerably smaller both for lower and higher particle bulk concentrations .
this article is organized as follows . in section 2 ,
we introduce our model and define the thermodynamic quantities of interest . in section 3 , we calculate the particle - mediated interaction potential @xmath4 of the surfaces . the global minimum of this potential is determined in section 4 , and the effective adhesion energy of the surfaces in section 5 . in section 6 , we show that the interaction potential @xmath4 exhibits a barrier at surface separations slightly larger than the particle diameter , because a particle bound to one of the surfaces ` blocks ' the binding of a second particle to the apposing surface .
the particle binding probability is calculated and analyzed in section 7 .
we consider a one - dimensional gas of particles with attractive boundaries , see figure [ cartoon ] .
the particles are modeled as hard spheres , and the attractive interaction between the particles and the boundaries , or ` surfaces ' , is described by a square - well potential with depth @xmath5 and range @xmath6 .
the length @xmath2 of the gas ` column ' corresponds to the separation of the surfaces and the width of the column is chosen to be equal to the particle diameter @xmath0 .
the particles in the column exchange with a bulk reservoir of particles .
the position of the center of mass of particle @xmath7 is denoted by @xmath8 , and its momentum by @xmath9 . for the system of @xmath10 hard particles confined in the column of length @xmath11
, one has @xmath12 .
we assume that the @xmath13-st and @xmath10-th particle interact with the surfaces , i.e. with the bases of the columns , _ via _ the square - well potential @xmath14 where @xmath15 and @xmath16 are the potential depth and range , respectively .
we also assume that @xmath17 . here and below
, @xmath18 denotes the heaviside step function with @xmath19 for @xmath20 and @xmath21 for @xmath22 .
the configuration energy for the system of @xmath10 particles in the column is @xmath23 and the corresponding canonical partition function can be written as @xmath24 after integration over the momenta of the particles , see , e.g. , @xcite . here
, @xmath25 is the thermal de broglie wavelength , and @xmath26 denotes the temperature times the boltzmann constant . in other words
, @xmath26 is the basic energy scale .
since the particles can exchange with the bulk solution , the number @xmath10 of particles in the column is not constant .
such a system is described by the grand - canonical ensemble in which the temperature , the column length @xmath2 , and the particle chemical potential @xmath27 are fixed .
the corresponding grand - canonical partition function @xmath28 is a sum of a finite number of elements , where @xmath29 denotes the largest integer less than or equal @xmath30 .
the upper limit of the sum on the right hand side of equation ( [ gcpf ] ) is the largest number of hard particles with diameter @xmath0 that can be accommodated in a column of length @xmath2 .
the partition function @xmath31 given by equation ( [ gcpf ] ) determines the grand potential @xmath32 the bulk density of the grand potential @xmath33 and , hence , the surface contribution to the grand potential @xmath34 .
the effective interaction potential of the surfaces was called the ` effective adhesion potential ' in reference @xcite . ] , @xmath35 is defined as the density of the surface contribution to the grand potential @xmath36 . for consistency with our previous model for particle - mediated surface interactions
@xcite , the column bases are chosen here to be squares of side length @xmath0 .
thus the @xmath37 in the denominator of equation ( [ sfe ] ) is the column base area .
the surface potential @xmath1 defined by equation ( [ sfe ] ) is the main quantity of interest here and will be determined in the next section .
equations ( [ gcpf ] ) - ( [ sfe ] ) imply that @xmath38 to determine the surface interaction potential @xmath1 , we thus have to calculate the canonical partition function @xmath39 and the bulk density of the grand potential , @xmath40 , defined in equation ( [ bfed ] ) .
the @xmath10-particle partition function @xmath39 is defined in equation ( [ cpf ] ) .
the change of variables @xmath41 in equation ( [ cpf ] ) , with @xmath42 , leads to @xmath43 where @xmath5 is the binding energy of the particles .
the integral ( [ integral_zn ] ) can be evaluated , see [ appendix_cpf ] , and after some computation we arrive at @xmath44 with @xmath45 for any length @xmath46 . for @xmath47 , equation ( [ zn ] ) ,
reduces to the partition function @xmath48 of the classical tonks gas @xcite .
the grand potential density @xmath40 can be derived from this exact result as shown in the following subsection .
the canonical and grand - canonical ensembles are equivalent in the thermodynamic limit , i.e. for infinite surface separation @xmath2 . in this limit
, the grand potential density @xmath40 therefore can be obtained from the canonical potential density _ via _ legendre transformation .
first , we define the canonical free energy @xmath49 and the free energy density in the bulk , @xmath50 where @xmath51 denotes the thermodynamic limit in which both the column length @xmath2 and the particle number @xmath10 go to infinity while the particle ` volume fraction ' @xmath52 remains constant .
the particle volume fraction in one dimension defined by equation ( [ phi_definition ] ) attains values @xmath53 , and could also be called a ` length fraction ' . in the one - dimensional model considered here ,
close packing corresponds to @xmath54 .
the free energy density @xmath55 defined by equation ( [ fca_definition ] ) is an intensive quantity in the thermodynamic limit and , therefore , does not depend on the boundary conditions . in particular , the free energy density @xmath55 and its derivatives do not depend on the boundary potential ( [ square - well - pot ] ) , which is characterized by the binding energy @xmath5 and range @xmath6 . with the exact expression for the canonical partition function @xmath39 given in equation ( [ zn ] ) and the sterling formula
@xmath56 we get @xmath57 .
\label{ffb}\ ] ] as expected , the particle - surface interactions characterized by the binding energy @xmath5 and range @xmath6 do not affect the free energy density @xmath55 in the thermodynamic limit . from the canonical free energy density @xmath55 , we obtain the chemical potential @xmath58 for the particles in the bulk .
equations ( [ ffb ] ) and ( [ mu ] ) lead to @xmath59 which can be rewritten as @xmath60 finally , the grand potential density @xmath40 follows from the legendre transformation @xmath61 equations ( [ ffb ] ) , ( [ muu ] ) and ( [ legendre ] ) lead to @xmath62 with the dependence between the chemical potential @xmath27 and the particle bulk volume fraction @xmath63 given in equation ( [ muu ] ) .
basic thermodynamics implies that the gas pressure in the bulk is @xmath64 .
this leads to the pressure @xmath65 $ ] here , which is the correct equation of state for the tonks gas @xcite . by combining equations ( [ v ] ) , ( [ muuu ] ) and ( [ omega ] ) ,
the effective , particle - mediated interaction potential @xmath1 of the surfaces can be expressed as a function of the particle bulk volume fraction @xmath63 and the separation @xmath2 between the surfaces : @xmath66 + \frac { \ell \ , t}{d^3 } \frac{\phi}{1 - \phi } , \label{vv}\ ] ] with the exact expression for the canonical partition function @xmath39 given in equations ( [ zn ] ) and ( [ phi_n ] ) , the potential @xmath1 can be evaluated numerically for any finite surface separation @xmath2 , see figures [ v1 ] and [ v2 ] . for large bulk volume fractions @xmath63 ( dashed curve in figure [ v1 ] ) , the potential @xmath1 exhibits oscillations up to surface separations @xmath2 of the order of several particle diameters before approaching a constant , asymptotic value @xmath67 for large surface separations .
the oscillations are related to successive layers of particles formed in the space between the surfaces . for small bulk volume fractions @xmath63 , in contrast
, the interaction potential @xmath1 attains an approximately constant value for surface separations @xmath68 , see solid curve in figure [ v1 ] .
since the interactions between the particles and the surfaces are short - ranged , the potential @xmath1 has a horizontal asymptote , i.e. @xmath69 approaches a constant value @xmath67 for large surface separation @xmath2 , see figures [ v1 ] and [ v2 ] . in this subsection
, we calculate the position of the asymptote . to simplify the notation
, we first introduce the auxiliary variable @xmath70 and the function @xmath71 defined for an arbitrary length @xmath46 . by combining equations ( [ v ] ) , ( [ zn ] ) , ( [ muuu ] ) and ( [ omega ] )
, we then express the potential @xmath1 as @xmath72 using the saddle - point approximation and stirling s formula ( [ stirling ] ) , one can prove that @xmath73 see [ appendix_proof ] .
we now apply this result to equation ( [ vinf1 ] ) and arrive at @xmath74 hence , the asymptotic value of the potential @xmath4 is given by the following exact expression : @xmath75 .
\label{vinfty}\ ] ] for non - adhesive particles with @xmath47 or @xmath76 , the asymptotic value of the potential @xmath4 is @xmath77 . for small bulk volume fractions @xmath78 of the particles and large binding energy @xmath5 with @xmath79
, we obtain @xmath80 we will use equations ( [ vinfty ] ) and ( [ vinfty2 ] ) in section [ section_adhesion ] to calculate the effective adhesion energy of surfaces . in the following section [ section_transition ] ,
we determine the global minimum of the potential @xmath4 .
in the present calculation we have chosen the free energy reference state in such a way that the effective surface interaction potential vanishes at surface contact @xmath81 , i.e. @xmath82 , see equation ( [ sfe ] ) . for surface separations
@xmath83 , the potential @xmath4 increases linearly with @xmath2 , i.e. @xmath84 $ ] , since @xmath85 for these separations .
the potential @xmath4 decreases for separations @xmath86 , and increases again for @xmath87 .
the potential thus attains a minimum at @xmath88 , see figures [ v1 ] and [ v2 ] .
the value @xmath89 at this minimum can be calculated again from equation ( [ vv ] ) . for @xmath90 , equation ( [ zn ] )
reduces to @xmath91 and @xmath92 for @xmath93 since only a single particle fits into the column .
insertion of these results into equation ( [ vv ] ) leads to @xmath94 + \frac{t}{d^2 } \left ( \frac{l_r}{d } + 1 \right ) \frac{\phi}{1- \phi } .
\label{v_d_plus_r}\ ] ] the potential @xmath4 thus has two local minima , one located at @xmath81 with @xmath95 and the other at @xmath96 with @xmath97 . the two minima result from the interplay of depletion interactions and adhesive interactions .
the global minimum of @xmath4 is located at @xmath88 if @xmath98 , i.e. for @xmath99 with @xmath100 , see equation ( [ zeta ] ) .
the inequality ( [ inequality ] ) is fulfilled for sufficiently large particle binding energies @xmath5 . for small binding energies @xmath5 ,
in contrast , the potential @xmath4 has its global minimum at surface contact @xmath81 , see figure [ v2 ] . in the experimentally relevant case of small particle
bulk volume fractions @xmath78 and large binding energy @xmath5 with @xmath79 , equation ( [ v_d_plus_r ] ) reduces to @xmath101 and the inequality ( [ inequality ] ) simplifies to @xmath102 and , thus , to a relation that is independent of the particle bulk volume fraction @xmath63 .
in this section , we assume that the binding energy @xmath5 of the particles is sufficiently large so that the inequality ( [ inequality ] ) is fulfilled .
the global minimum of the interaction potential @xmath69 then is located at @xmath88 .
the minimum value @xmath103 is given by equation ( [ v_d_plus_r ] ) .
for large surface separations @xmath104 , the potential @xmath4 attains a constant value @xmath67 given by equation ( [ vinfty ] ) .
the difference between the asymptotic and the minimum value of the potential @xmath4 is the effective adhesion energy @xmath105 of the surfaces .
the effective adhesion energy is the minimal work that has to performed to bring the two surfaces far apart from the separation @xmath88 . from equations ( [ vinfty ] ) , ( [ v_d_plus_r ] ) , and ( [ gamma ] )
, we obtain the exact result @xmath106 - \frac{t}{d^2 } \left ( \frac{l_r}{d } + 1 \right ) \frac{\phi}{1- \phi } \\ - \frac{t}{d^2 } \ln \left ( 1- \phi \right ) - 2 \frac{t}{d^2 } \ln \left [ e^{u / t } - ( e^{u / t } -1 ) \exp \left ( - \frac{l_r}{d } \frac{\phi}{1- \phi } \right ) \right ] .
\label{gamma3}\end{aligned}\ ] ] in figure [ g ] , the adhesion energy @xmath3 is plotted as a function of the particle bulk volume fraction @xmath63 .
interestingly , the adhesion energy @xmath3 exhibits a maximum at an optimal bulk volume fraction @xmath107 of the particles . for small bulk volume fractions @xmath108 and large binding energy @xmath5 with @xmath79 , the asymptotic value and minimum value of @xmath69 are approximately given by equations ( [ vinfty2 ] ) and ( [ vmin2 ] ) , respectively .
the adhesion energy @xmath109 then simplifies to @xmath110 this expression is identical with our previous result obtained from a virial expansion in @xmath63 up to second order terms @xcite . for @xmath78 and @xmath79 , the adhesion energy ( [ gamma2 ] ) is a good approximation of the exact result ( [ gamma3 ] ) , see figures [ g ] and [ w0w ] . from equation ( [ gamma2 ] ) , we obtain the approximate expression @xmath111 for the optimum bulk volume fraction @xmath107 at which the adhesion energy @xmath3 attains its maximum value .
the adhesion energy ( [ gamma2 ] ) can be understood as the difference of two langmuir adsorption free energies per fluid column , or pair of apposing binding sites @xcite : ( i ) the adsorption free energy @xmath112 for small surface separations at which a particle binds both surfaces with total binding energy @xmath113 , and ( ii ) the adsorption free energy @xmath114 for large surface separations , counted twice in ( [ gamma2 ] ) because we have two surfaces .
these langmuir adsorption free energies result from a simple two - state model in which a particle is either absent ( boltzmann weight @xmath13 ) or present ( boltzmann weights @xmath115 and @xmath116 , respectively ) at a given binding site , see e.g. @xcite .
the factor @xmath117 depends on the degrees of freedom of a single adsorbed particle . in our model
, we obtain @xmath118 . to assess the quality of approximate expression ( [ gamma2 ] ) , we analyze its relative error in reference to the exact result ( [ gamma3 ] ) .
the relative error is the magnitude of the difference between the exact result ( [ gamma3 ] ) and the approximate expression ( [ gamma2 ] ) divided by the magnitude of the exact result ( [ gamma3 ] )
. figure [ w0w ] shows parameter regions in which the relative error of the expression ( [ gamma2 ] ) is smaller or larger than 1% , 2% and 5% , respectively . in this example , the binding range is @xmath119 . for intermediate and large binding energies with @xmath120 ,
we find that the relative error of the approximate expression ( [ gamma2 ] ) is smaller than 1% in a broad range of volume fractions @xmath63 .
for large binding energies @xmath5 with @xmath79 , the effective interaction potential has a barrier at surface separations @xmath121 , see figure [ v1 ] . at these separations
, only a single particle fits between the surfaces , but this particle can just bind one of the surfaces .
the particle thus blocks the binding site at the apposing surface .
the potential barrier attains its maximum value @xmath122 at the separation @xmath123 , see figure [ v1 ] . from equation ( [ vinf1 ] ) , we obtain @xmath124 + 2 \frac{t}{d^2 } \frac{\phi}{1- \phi}\ ] ] for @xmath108 and @xmath79 , we get @xmath125 the barrier height @xmath126 then is @xmath127 since the asymptotic value @xmath67 is given by equation ( [ vinfty2 ] ) in this limiting case .
the width of the barrier is approximately @xmath128 , see figure [ v1 ] .
equation ( [ u_ba ] ) is again identical with our previous result obtained from a virial expansion in @xmath63 up to second order terms @xcite .
another quantity of interest here is the binding probability @xmath129 defined as the probability that the separation of the closest particle from a column base is smaller than the binding range @xmath6 . in other words , the binding probability @xmath130 is the probability of finding a particle bound to one of the bases .
the binding probability corresponds to the surface coverage in the case of a three - dimensional gas of particles between two parallel attractive surfaces .
equations ( [ square - well - pot ] ) - ( [ fe ] ) imply that the binding probability can be calculated by differentiation of the grand potential @xmath36 with respect to binding energy @xmath5 , i.e. @xmath131 . since the grand potential density @xmath40 given by equation ( [ omega ] ) does not depend on the binding energy @xmath5 , the binding probability can also be obtained from the effective surface interaction potential _ via _
@xmath132 with the exact expression ( [ vinf1 ] ) for the interaction potential @xmath1 , the binding probability @xmath130 can be determined numerically for any finite separation @xmath2 . in figure [ c ] ,
the binding probability @xmath130 is plotted as a function of surface separation @xmath2 for three different volume fractions @xmath63 around the optimal volume fraction @xmath107 at which the adhesion energy @xmath3 is maximal . in the vicinity of @xmath107 , the binding probability at large separations
@xmath133 is sensitive to small variations of @xmath63 , while the binding probability at separations @xmath2 in the surface binding range @xmath134 remains practically constant at almost 100% . for small bulk volume fractions @xmath63 of the particles and large particle binding energies @xmath5 ,
the asymptotic and minimum value of the interaction potential @xmath4 are given by equations ( [ vinfty2 ] ) and ( [ vmin2 ] ) , respectively .
from these equations and relation ( [ cov_def ] ) , we obtain the approximate expressions @xmath135 for the binding probability at large surface separation @xmath2 and @xmath136 for the particle binding probability at the binding separation @xmath88 of the surfaces , with the optimum bulk volume fraction @xmath107 given in equation ( [ x_b^star ] ) .
these expressions correspond to the well - known langmuir adsorption equation @xcite . at the optimal volume fraction @xmath107 ,
the particle binding probability for unbound and bound surfaces is @xmath137 and @xmath138 , respectively . bringing the surfaces from large separations
@xmath68 within binding separations @xmath134 thus does not require desorption or adsorption of particles at @xmath139 .
we have considered one - dimensional gas of hard - sphere particles with attractive boundaries , a novel extension of the tonks model @xcite .
we have solved this model analytically in the whole range of parameters by explicit integration over the particles degrees of freedom in the partition function .
in contrast to other studies on one - dimensional models for hard spheres @xcite , we have focused on the boundary contribution to the free energy of the system , which corresponds to the effective , particle - mediated interaction potential between the boundaries , or surfaces , see figures [ v1 ] and [ v2 ] .
the effective adhesion energy obtained from the interaction potential depends non - monotonically on the volume fraction @xmath63 of the particles in the bulk , see figure [ g ] .
the adhesion energy exhibits a maximum at an optimum volume fraction , which can lead to reentrant transitions in which the surfaces first bind with increasing volume fraction @xmath63 , and unbind again when the volume fraction @xmath63 is increased beyond its optimum value .
a lattice of such one - dimensional gas columns represents a discrete approximation of a three - dimensional gas of particles between two adsorbing surfaces , see figure [ cartoon ] and reference @xcite . for small volume fractions @xmath63 and short - ranged particle - surface interactions considered here , the gas of particles between two well - separated surfaces is as dilute as in the bulk , except for the single adsorption layers of particles at the surfaces . at larger volume fractions ,
three - dimensional packing effects become relevant .
these effects are not captured correctly in the one - dimensional model .
however , it has been pointed out @xcite that approximations based on one - dimensional models do well in comparison to density functional theories for three - dimensional hard sphere fluids confined in planar , non - adsorbing pores @xcite . in principle , the quality of the one - dimensional approximation can be tested by monte carlo or molecular dynamics simulations , which have been used to study various three - dimensional systems of hard spheres confined between non - adsorbing surfaces @xcite . for simplicity , and for consistency with our previous publication @xcite ,
we have considered here a square lattice of columns between the surfaces . in particular , the factor @xmath37 in the denominator of equation ( [ sfe ] ) is the column base area in the square lattice . for a hexagonal lattice of columns ,
the corresponding area is @xmath140 , and the corresponding effective interaction potential of the surfaces is thus obtained by multiplying the right hand side of equation ( [ vv ] ) with a factor @xmath141 .
the adhesion energy of the surfaces has to be rescaled with the same factor in the case of hexagonal lattice of columns , but its functional dependence on the bulk volume fraction @xmath63 , binding energy @xmath5 , and binding range @xmath6 remains unchanged .
we have considered an equilibrium situation in which the particles exchange with a bulk solution . for polymers between surfaces ,
such an equilibrium has been termed ` full equilibrium ' . in a ` restricted equilibrium ' ,
in contrast , the polymers are trapped between the surfaces @xcite , which is less likely for the spherical particles considered here .
in this section we calculate the @xmath10-particle partition function @xmath39 as given by equation ( [ integral_zn ] ) for @xmath11 .
first , if one notices that @xmath142 and @xmath143 the integral ( [ integral_zn ] ) can be written as a sum of four terms @xmath144 with @xmath145 @xmath146 @xmath147 @xmath148 the first integral @xmath149 and second integral @xmath150 can be easily calculated . to calculate the third integral , we start from @xmath151 = y_2 - \left ( y_2 - l_r \right ) \theta \left ( y_2 - l_r \right ) .
\label{help_int_1}\ ] ] in the next steps , we find @xmath152 and @xmath153 iterating these results leads to @xmath154 { \rm d } y_n .\ ] ] the integral @xmath155 can now be evaluated as @xmath156 note that @xmath157 .
the fourth integral , @xmath158 , can be brought to the form @xmath159 { \rm d } y_n \label{i4a}\ ] ] if one uses again ( [ help_int_1 ] ) and iterates the integration as in ( [ help_int_2 ] ) and ( [ help_int_3 ] ) .
the first term on the right hand side of equation ( [ i4a ] ) is equal to @xmath160 , see equation ( [ i2a ] ) .
thus @xmath161 , where @xmath162 to determine the integral @xmath163 , one has to distinguish three cases : ( i ) for @xmath164 , we obtain @xmath165 ( ii ) for @xmath166 , we obtain @xmath167 and ( iii ) for @xmath168 one gets @xmath169 . in summary @xmath170
if one now gathers the results ( [ i1a ] ) , ( [ i2a ] ) , ( [ i3a ] ) , ( [ i4b ] ) and returns to equation ( [ i_split ] ) , one obtains the partition function @xmath171\ ] ] with @xmath172 @xmath173 @xmath174 this result can be written as @xmath175\end{aligned}\ ] ] which simplifies to equation ( [ zn ] ) .
here , we explore the asymptotics of the function g@xmath176 defined in equation ( [ function_f ] ) and , hence , prove the equality ( [ limit_toprove ] ) . to simplify the notation , let @xmath177 , where @xmath178 is a large integer number , and @xmath179
. then @xmath180 in the next step , we introduce the auxiliary function @xmath181 and rewrite the function g@xmath182 given by equation ( [ function_ff ] ) in the form @xmath183 using stirling s formula ( [ stirling ] ) . for large @xmath178 , one can replace the sum on the right hand side of equation ( [ function_fff ] ) by an integral and write @xmath184 the function @xmath185 has a global maximum at @xmath186 with @xmath187 for large @xmath178 . note that for large numbers @xmath178 , the location @xmath188 of the global minimum scales linearly with @xmath178 .
if we now expand the function @xmath185 around the point @xmath189 up to second order terms and apply the saddle - point approximation , we get @xmath190 { \rm d } n .\ ] ] a simple change of variables @xmath191 leads to @xmath192 where we have used @xmath188 , which follows from equation ( [ n0 ] ) and relation ( [ zeta ] ) between variables @xmath193 and @xmath63 . in the limit of large @xmath178
, we thus obtain @xmath194 because @xmath195 .
now , we can calculate the gaussian integral in ( [ function_f_sp ] ) to get @xmath196 from the definition ( [ function_phi ] ) of the auxiliary function @xmath185 and equation ( [ n0 ] ) for the point @xmath186 at which function @xmath185 has its global maximum , we get @xmath197 and @xmath198 note that @xmath199 , and that the function @xmath185 has indeed a maximum at @xmath186 .
combining equations ( [ function_f_g ] ) , ( [ phi_n0 ] ) and ( [ phi_bis_n0 ] ) leads to @xmath200 for large @xmath178 values and , thus , to equation ( [ limit_toprove ] ) _ quod erat demonstrandum_. | we present here exact results for a one - dimensional gas , or fluid , of hard - sphere particles with attractive boundaries .
the particles , which can exchange with a bulk reservoir , mediate an interaction between the boundaries .
a two - dimensional lattice of such one - dimensional gas ` columns ' represents a discrete approximation of a three - dimensional gas of particles between two surfaces .
the effective particle - mediated interaction potential of the boundaries , or surfaces , is calculated from the grand - canonical partition function of the one - dimensional gas of particles , which is an extension of the well - studied tonks gas @xcite .
the effective interaction potential exhibits two minima .
the first minimum at boundary contact reflects depletion interactions , while the second minimum at separations close to the particle diameter results from a single adsorbed particle that crosslinks the two boundaries .
the second minimum is the global minimum for sufficiently large binding energies of the particles .
interestingly , the effective adhesion energy corresponding to this minimum is maximal at intermediate concentrations of the particles . |
in physics , the study of colloidal brownian motion has a long history beginning with einstein s famous paper @xcite in 1905 , and the understanding of its mechanism has been systematically developed in molecular kinetic theory @xcite .
recently , experimental developments have enabled researchers to observe a single trajectory itself of brownian motion @xcite , which engages physicists in quantitative modelling of sub - micrometer systems , such as molecular motors @xcite .
likewise , recent breakthroughs for computational technologies have enabled physicists to study microstructure of financial brownian motion in detail .
they have applied their knowledge beyond material science and into studying in particular price movements in financial markets for about a quarter century . there
consequently appeared new approaches as mentioned below , in contrast to conventional mathematical finance where a priori theoretical dynamics is assumed @xcite .
there are three physical approaches to financial brownian motions as shown in figure [ fig : micro - meso - macro ] .
the microscopic approach focuses on the dynamics of traders in the financial markets ( figure [ fig : micro - meso - macro]a ) .
traders correspond to molecules in the modelling of materials , which enables both a numerical and a theoretical analysis of the macroscopic motion of prices @xcite .
another approach is based on macroscopic empirical analyses of price time series ( figure [ fig : micro - meso - macro]c ) and the direct empirical modelling of price dynamics focusing on fat - tailed distributions and long - time correlations in volatility @xcite .
the third approach focuses on mesoscopic dynamics concerning order - books ( figure [ fig : micro - meso - macro]b ) , which are accumulated buy sell orders initiated by traders in the price axis where deals occur either at the best bid ( buy ) or ask ( sell ) price defining the market prices
. numerical simulations of markets become available by introducing models of order - book dynamics @xcite .
recently , the mesoscopic approach developed considerably with the analysis of high frequency financial market data in which the whole history of orders is tractable using a direct analogy between order - books and colloids @xcite . from the order - book data
@xcite , the importance of inertia was reported for market price , implying the existence of market trends . the langevin equation was then found to hold most of time showing that the fluctuation
dissipation relation can be extended to order - book dynamics .
however , microscopic mechanisms behind market trends have not been clarified so far because direct information is required on individual traders strategies . in the present paper , we analyse more informative order - book data in which traders can be identified in an anonymised way for each order so that we can estimate each trader s dynamics directly from the data . here
we present a minimal model validated by direct observation of individual traders trajectories for financial brownian motion .
we first report a novel statistical law on the trend - following behaviour of foreign exchange ( fx ) traders by tracking the trajectories of all individuals .
we next introduce a corresponding microscopic model incorporated with the empirical law .
we reveal the mesoscopic and macroscopic behaviour of our model by developing a parallel theory to that for molecular kinetics : boltzmann - like and langevin - like equations are derived for the order - book and the price dynamics , respectively . a quantitative agreement with empirical findings
is finally presented without further assumptions .
we analysed the high - frequency fx data between the us dollar ( usd ) and the japanese yen ( jpy ) from the 5th 16.00 to the 10th 20.00 gmt september 2016 on electronic broking services , one of the biggest fx platforms in the world .
all trader activities were recorded for our dataset with anonymised trader ids with one - millisecond time - precision .
the minimum price - precision was 0.005 yen for the usd / jpy pair at that time , and the currency unit in this paper is 0.001 yen , called the tenth pip ( tpip ) .
the minimum volume unit for transaction was one million usd , and the total monetary flow was about @xmath0 billion usd during this week .
the market adopts the double - auction system , where traders quote bid or ask prices . in this paper , we particularly focused on the dynamics of high - frequency traders ( hfts ) , who frequently submit or cancel orders according to algorithms ( see appendix [ app : def_hft ] for the definition ) .
the presence of hfts has rapidly grown recently @xcite and @xmath1 of the total orders were submitted by the hfts in our dataset .
we first illustrate the trajectories of bid and ask prices quoted for the top 3 hfts in figure .
[ fig : trajectory_traders]a c .
we observed that with the two - sided quotes typical hfts tend to play the role of liquidity providers ( called market - makers @xcite ) .
we also observed that buy sell spreads ( i.e. , the difference between the bid and ask prices for a single market - maker ) fluctuated around certain time - constants , showing a strong coupling between these prices .
indeed , the buy sell spread distributions exhibit sharp peaks for individual hfts as shown in the insets in figure [ fig : trajectory_traders]a c ( see also appendix [ app : buy - sell ] ) .
we next report the empirical microscopic law for the trend - following strategy of individual traders .
let us denote the bid and ask prices of the top @xmath2th hft by @xmath3 and @xmath4 ( see appendix [ app : trend - following ] for the definitions ) .
we investigated the average movement of the mid - price @xmath5 between transactions conditional on the previous market price movement @xmath6 ( figure [ fig : trend_follow]a ) . for the top 20 hfts ( figure [ fig : trend_follow]b and c ) ,
we find that the average movement is described by @xmath7 where the conditional average @xmath8 is taken when the previous price movement is @xmath6 and @xmath9 ( see appendix [ app : trend - following ] for the detail ) . @xmath10 and @xmath11 are constants characterizing the price movement and the saturation threshold against the market trend . here
, typical values are given using @xmath12 [ tpip ] and @xmath13 [ tpip ] .
the empirical law ( [ eq : trendfollow ] ) implies that the reaction of traders is linear for small market trends but saturates for large market trends .
remarkably , a similar behaviour was reported from a full macroscopic analysis of market price data at one - month precision @xcite .
here we introduce a minimal microscopic model incorporating the empirical law ( [ eq : trendfollow ] ) .
we make four assumptions : ( i ) the number of traders is sufficiently large .
( ii ) they always quote both bid and ask prices ( for the @xmath2th trader , @xmath3 and @xmath4 ) simultaneously with a unit volume as market - makers .
( iii ) buy sell spreads are time - constants unique to traders with distribution @xmath14 .
the trader dynamics is then characterized by the mid - price @xmath15 .
( iv ) trend - following random walks are assumed in the microscopic dynamics ( see figure [ fig : dealermodel]a c ) : @xmath16 with a constant strength for trend - following @xmath17 , white gaussian noise @xmath18 with constant variance @xmath19 , and a requotation jump @xmath20 after transactions ( figure [ fig : dealermodel]c ) .
@xmath20 is defined by @xmath21 with jump size @xmath22 and the @xmath23th transaction time @xmath24 between traders @xmath2 and @xmath25 .
the transaction condition at time @xmath26 is given by @xmath27 ( figure [ fig : dealermodel]b ) .
this model can be assessed using parallel tools employed in molecular kinetic theory . in this theory
@xcite , the boltzmann equation is first derived for the one - body velocity distribution from the hamiltonian dynamics by the method of bogoliubov , born , green , kirkwood , and yvon assuming molecular chaos .
the langevin equation is derived in turn from the boltzmann equation for massive brownian particles @xcite .
here we have followed the same mathematical procedure to elucidate the dynamics behind order - book profiles and financial brownian motion . for relative position @xmath28 from the centre of mass @xmath29 ,
the boltzmann - like equation is first derived for the one - body probability distribution density @xmath30 conditional for a trader with a buy sell spread @xmath31 from the multi - agent dynamics ( [ eq : dealermodel ] ) : @xmath32\label{eq : financialboltzmann}\ ] ] with @xmath33 and @xmath34 . here , @xmath35
( @xmath36 ) represents transactions as bid ( ask ) orders .
the average order - book profile is given by @xmath37 for the ask side .
the integral term in equation ( [ eq : financialboltzmann ] ) corresponds to the collision integral in the conventional boltzmann equation .
as the langevin equation is derived from it @xcite , then similarly the langevin - like equation is derived from the boltzmann - like equation ( [ eq : financialboltzmann ] ) , @xmath38 where @xmath39 and @xmath40 are the market price movement and transaction time interval at the @xmath41th tick time .
the tick time is an integer time incremented by every transaction and the mean time interval between transactions is @xmath42 seconds in this week .
the first and second terms on the right - hand side of eq .
( [ eq : financiallangevin ] ) describe trend - following and random noise , respectively ; the trend - following term corresponds to the momentum inertia in the conventional langevin equation .
equations ( [ eq : financialboltzmann ] ) and ( [ eq : financiallangevin ] ) can be solved for @xmath43 under an appropriate boundary condition ( see appendix [ app : boundaryconditions ] ) .
we first set the buy sell spread distribution as @xmath44 with decay length @xmath45}$ ] , empirically validated in our dataset ( figure [ fig : pricediff]a and appendix [ app : buy - sell ] ) .
the average order - book profile @xmath46 is given for @xmath47 by @xmath48 .
\label{eq : orderbook}\ ] ] the tail of the price movement is approximately given by @xmath49 with decay length @xmath50 , average movement from trend - following @xmath51 , average transaction interval @xmath52 , and complementary cumulative distribution @xmath53 .
further technical details are to be published in a forthcoming paper .
we next investigated the consistency between our microscopic model and our dataset .
the empirical daily profile was first studied for the average order - book @xmath46 in figure [ fig : pricediff]b ( see appendix [ app : order - book ] for the detail ) .
surprisingly , we found an agreement with our theoretical lines ( [ eq : orderbook ] ) without fitting parameter that strongly supports the validity of our description .
we also empirically evaluated the two - hourly segmented cumulative distribution for the price movement in one - tick precision @xmath54 ( figure [ fig : pricediff]c ) , which obeys an exponential law that is consistent with our theoretical prediction ( [ eq : theory_pricemove ] ) . in our dataset ,
the decay length @xmath55 is approximately constant over a two - hour period but varies over time during the week .
to remove this non - stationary feature , we introduced the two - hourly scaled cumulative distribution @xmath56 with scaling parameters @xmath55 and @xmath57 ( figure [ fig : pricediff]d ) , thereby incorporating the two - hourly exponential - law for the whole week .
the price movements obey an exponential law for short periods but simultaneously obeys a power - law over long periods with exponent @xmath58 ( figure [ fig : pricediff]e ) .
this apparent discrepancy is explained by the power - law nature of the decay length @xmath55 . because @xmath55 approximately obeys a power - law cumulative distribution @xmath59 over the week with @xmath60 ( figure [ fig : pricediff]f ) , the one - week distribution @xmath61 obeys the power - law as a superposition of the two - hourly segmented exponential distribution , @xmath62 with @xmath63 .
remarkably , @xmath55 tends to be long when the market is inactive ( figure [ fig : pricediff]g ) .
we therefore obtain a consistent result with the previously reported power - law @xcite as a non - stationary property of @xmath55 .
we note that our model can exhibit super - diffusion under an appropriate parameter set as @xmath64 with @xmath65 for short periods ( figure [ fig : pricediff]h ) , which is consistent with previous reports @xcite . here
the mean squared displacement is defined by @xmath66 as a function of tick time @xmath41 with the ensemble average @xmath67 .
we also note that our model asymptotically shows ballistic behaviour @xmath68 when trend - following is sufficiently large , which implies that it plays the role of a momentum inertia " in financial markets .
we further note that our model can show sub - diffusion ( @xmath69 ) when trend - following is sufficiently small .
in this article , we have presented an intensive data analysis of anonymised traders in a foreign exchange market to directly observe strategies of hfts .
we first report a simple empirical law characterizing trend - following behaviour of individual traders against market trends .
a trend - following random walk model is correspondingly introduced as a microscopic dynamics of the financial market .
the mesoscopic and macroscopic behaviours of this model are systematically analysed in a parallel calculation to molecular kinetic theory .
our theoretical model reproduces the average order - book profile and the price movement distributions empirically .
this work would be an important step toward unified description of financial markets from individual traders dynamics .
we discuss here a possible reason behind the success of our kinetic - like theory for our model . in material physics ,
mean - field approximations are invalid for low - dimensional systems because the low - dimensional geometry does not allow two - body correlations to disappear after collision .
in contrast , in our model , traders are separated compulsorily after transactions , and there is little possibility for the same pair to enter successive transactions .
the two - body correlation then quickly decays after collision assuming molecular chaos " is valid .
this scenario implies that the kinetic - like description may work well in various socio - economic systems , in addition to the previously studied examples , such as opinion formation and wealth distribution @xcite .
our report dealt mainly with short - duration trends ; their correlation with long - duration trends @xcite is a topic to future studies .
of interest is the study of traders behaviour in unstable markets triggered by external shocks , as those that occur in financial crises and flash crashes .
the economics reason behind the hyperbolicity ( [ eq : trendfollow ] ) in trend - following needs to be pursed further .
we greatly appreciate icap for their provision of the ebs data .
we also appreciate m. katori , h. hayakawa , s. ichiki , k. yamada , s. ogawa , f. van wijland , d. sornette , t. sano , and t. ito for fruitful discussions .
this work was supported by jsps kakenhi ( grand no .
16k16016 ) and jst , strategic international collaborative research program ( sicorp ) on the topic of ict for a resilient society " by japan and israel .
99 einstein , a. ber die von der molekularkinetischen theorie der wrme geforderte bewegung von in ruhenden flssigkeiten suspendierten teilchen . _
_ 322 , 549 - 560 ( 1905 ) .
chapman , s. & cowling , t. g. _ the mathematical theory of non - uniform gases _ ( cambridge univ .
press , cambridge , 1970 ) .
van kampen , n. g. _ stochastic processes in physics and chemistry _
( amsterdam , north - holland , 2007 ) .
van den broeck , c. , kawai , r. & meurs , p. microscopic analysis of a thermal brownian motor .
_ 93 , 090601 - 090604 ( 2004 ) .
huang , r. , chavez , i. , taute , k. m. , luki , b. , jeney , s. , raizen , m. g. & florin , e .-
direct observation of the full transition from ballistic to diffusive brownian motion in a liquid . _ nat .
_ 7 , 576 - 580 ( 2011 ) .
li , t. , kheifets , s. , medellin , d. & raizen , m. g. measurement of the instantaneous velocity of a brownian particle .
_ science _ 328 , 1673 - 1675 ( 2010 ) .
noji , h. , yasuda , r. , yoshida , m. & kinosita , k. jr direct observation of the rotation of f1-atpase .
_ nature _ 386 , 299 - 302 ( 1997 ) .
black , f. & scholes , m. s. the pricing of options and corporate liabilities .
_ j. political econ .
_ 81 , 637 - 654 ( 1973 ) .
engle , r. dynamic conditional correlation : a simple class of multivariate generalized autoregressive conditional heteroskedasticity models
. _ j. bus .
_ 20 , 339 - 350 ( 2002 ) .
takayasu , h. , miura , h. , hirabayashi , t. & hamada , k. statistical properties of deterministic threshold elements the case of market price . _
physica a _ 184 , 127 - 134 ( 1992 ) .
bak , p. , paczuski , m. & shubik , m. price variations in a stock market with many agents .
_ physica a _ 246 , 430 - 453 ( 1997 ) .
lux , t. & marchesi , m. scaling and criticality in a stochastic multi - agent model of a financial market .
_ nature _ 397 , 498 - 500 ( 1999 ) .
yamada , k. , takayasu , h. , ito , i. & takayasu , m. solvable stochastic dealer models for financial markets .
e _ 79 , 051120 - 051131 ( 2009 ) .
mantegna , r. n. & stanley , h. e. scaling behaviour in the dynamics of an economic index .
_ nature _ 376 , 46 - 49 ( 1995 ) .
mantegna , r. n. & stanley , h. e. _ an introduction to econophysics _
( cambridge univ .
press , 2000 ) .
lux , t. the stable paretian hypothesis and the frequency of large returns : an examination of major german stocks .
financial economics _ 6 463 - 475 ( 1996 ) .
plerou , v. , gopikrishnan , p. , amaral , l. a. , meyer , m. & stanley , h. e. scaling of the distribution of price fluctuations of individual companies .
e _ 60 , 6519 - 6528 ( 1999 ) .
guillaume , d. m. , dacorogna , m. m. , dav , r. r. , mller , u. a. , olsen , r. b. & pictet , o. v. from the bird s eye to the microscope : a survey of new stylized facts of the intra - daily foreign exchange markets .
_ finance stochast .
_ 1 , 95 - 129 ( 1997 ) .
longin , f. m. the asymptotic distribution of extreme stock market returns .
_ j. business _ 69 , 383 - 408 ( 1996 ) .
takayasu , m. , mizuno , t. & takayasu , h. potential force observed in market dynamics .
_ physica a _ 370 , 91 - 97 ( 2006 ) .
maslov , s. simple model of a limit order - driven market .
_ physica a _ 278 , 571 - 578 ( 2000 ) .
smith , e. , farmer , j. d. , gillemot , l. & krishnamurthy , s. statistical theory of the continuous double auction .
_ quantitative finance _ 3 , 481 - 514 ( 2003 ) .
bouchaud , j .-
, mzard , m. & potters , m. statistical properties of stock order books : empirical results and models .
_ quantitative finance _ 2 , 251 - 256 ( 2002 ) .
yura , y. , takayasu , h. , sornette , d. & takayasu , m. financial brownian particle in the layered order - book fluid and fluctuation - dissipation relations . _ phys .
_ 112 , 098703 - 098707 ( 2014 ) .
hendershott , t. , jones , c. m. & menkveld , a. j. does algorithmic trading improve liquidity ? _ the journal of finance _ 66 , 1 - 33 ( 2011 ) .
menkveld , a. j. high frequency trading and the new market makers .
_ journal of financial markets _ 16 , 712 - 740 ( 2013 ) .
ebs dealing rules appendix
ebs market ( at the time of june 2016 ) .
lemprire , y. , deremble , c. , seager , p. , potters , m. & bouchaud , j .-
two centuries of trend following .
_ journal of investment strategies .
_ 3 , 41 - 61 ( 2014 ) .
slanina , f. _ essentials of econophysics modelling .
_ ( oxford university , oxford , 2014 ) .
pareschi , l. & toscani , g. _ interacting multiagent systems : kinetic equations and monte carlo methods . _
( oxford university press , oxford , 2013 ) .
for this paper , we define a high frequent trader ( hft ) as a trader who submits more than 500 times a day on average ( i.e. , more than 2500 times for the week ) . as
several traders are unwilling to transact and often interrupt orders at the instant of submission ( called flashing ) , we excluded traders with live orders of less than @xmath70 percent of the transaction time . with this definition ,
the number of hfts was 134 during this week , whereas the total number of traders was 1015 .
we note that the total number of traders who submitted limit orders was 922 ; the other 93 traders submitted only market orders .
we calculated the percentage of two - sided quotes as follows : when a bid ( ask ) order is submitted by a trader , we check whether corresponding ask ( bid ) orders exist .
we then count the number of two - sided quotes for all traders at every order submission and finally divide it by the total number of submissions .
the difference in the median bid and ask prices was studied as a buy sell spread for an hft .
samples where only both bid and ask prices exist are taken in the one - second time - precision for figure [ fig : trajectory_traders ] and [ fig : pricediff]a .
we plotted standard deviations of the averages as error bars for each point .
we remark on the precise definition of the bid ( ask ) price of individual hfts for the analysis of trend - following .
if a trader quotes both single - bid and single - ask orders at any time , the bid and ask prices are defined literally . in the presence of multiple bid or
ask orders , we use the value of the median for the bid or ask orders as @xmath3 or @xmath4 . in the absence of any bid or ask orders , we use the previous bid or ask price as @xmath3 or @xmath4 for interpolation for figure .
[ fig : trend_follow]b and c. we also remark that exceptional samples where the bid or ask price is far from the market price by 10 yen ( 0.0659% of the total ) are excluded from the calculation of the conditional ensemble average @xmath71 .
we note that the standard deviations of the conditional averages are plotted for each point as error bars .
also , median values in the top 20 hfts are given using @xmath12 [ tpip ] and @xmath72 [ tpip ] .
for the boltzmann - like equation ( [ eq : financialboltzmann ] ) , we first introduce sufficiently large cutoffs at @xmath73 .
the limit @xmath43 is taken with reflecting boundaries assumed at @xmath73 . a large cutoff limit @xmath74 is taken finally .
for the langevin - like equation ( [ eq : financiallangevin ] ) , we make the following two assumptions .
( i ) trend - following has the same order as random noise : @xmath75 .
( ii ) saturation in trend - following ( [ eq : trendfollow ] ) is dominant : @xmath76 .
we note that equation ( [ eq : trendfollow ] ) can be approximated for large fluctuations as @xmath77 under these conditions with the signature function @xmath78 defined by @xmath79 for @xmath80 and @xmath81 for @xmath82 .
the daily average order - book profile is calculated for the hfts .
we took snapshots of the order - book every second and its ensemble average every day .
we also plotted standard deviations of the averages as error bars for each point .
we take snapshots of the order - book after every transaction and count the total number of different trader ids for both bid and ask sides . the counting weight for an hft quoting both sides
is set to 1 and that for an hft quoting one side is 1/2 .
we then plot the average of the number of trader ids for both bid and ask sides every two hours in figure [ fig : pricediff]g .
the typical number of hft ids was about @xmath83 in our dataset with this definition .
the number of total volumes quoted by hfts is typically about @xmath84 .
admittedly , there is room for debate on which number is appropriate for the calibration of the total number of traders in our model ; it remains a topic for future study . | brownian motion has been a pillar of statistical physics for more than a century , and recent high - frequency trading data have shed new light on microstructure of brownian motion in financial markets .
though evidences of trend - following behaviour of traders were indirectly shown in such trading data , the microscopic model has not been established so far by direct observation of trajectories for individual traders . in this paper
, we present a minimal microscopic model for financial brownian motion through an intensive analysis of trajectory data for all individuals in a foreign exchange market .
this model includes a novel empirical law quantifying traders trend - following behaviour that can create the inertial motion in market prices over short durations .
we present a systematic solution paralleling molecular kinetic theory to reveal mesoscopic and macroscopic dynamics of our model .
our model exhibits quantitative agreements with empirical results strongly supporting our analysis . |
complex networks can be observed in a wide variety of natural and man - made systems @xcite , and an important general problem is the relationship between the connection structure and the dynamics of these networks . with graph - theoretical approaches
, networks may be characterized using graphs , where nodes represent the elements of a complex system and edges their interactions . in the study of brain dynamics @xcite
, a node may represent the dynamics of a circumscribed brain region determined by electrophysiologic @xcite or imaging techniques @xcite .
then two nodes are connected by an edge , or direct path , if the strength of their interaction increases above some threshold . among other structural ( or statistical ) parameters ,
the average shortest path length @xmath0 and the cluster coefficient @xmath1 are important characteristics of a graph @xcite .
@xmath0 is the average fewest number of steps it takes to get from each node to every other , and is thus an emergent property of a graph indicating how compactly its nodes are interconnected .
@xmath1 is the average probability that any pair of nodes is linked to a third common node by a single edge , and thus describes the tendency of its nodes to form local clusters .
high values of both @xmath0 and @xmath1 are found in regular graphs , in which neighboring nodes are always interconnected yet it takes many steps to get from one node to the majority of other nodes , which are not close neighbors . at the other extreme ,
if the nodes are instead interconnected completely at random , both @xmath0 and @xmath1 will be low .
recently , the emergence of collective dynamics in complex networks has been intensively investigated in various fields @xcite .
it has for example been proposed that random , small - world , and scale - free networks , due to their small network distances , might support efficient and stable globally synchronized dynamics @xcite .
synchronized dynamics , however , depends not only on statistical but also on spectral properties of a network , which can be derived from the eigenvalue spectrum of the laplacian matrix describing the corresponding network @xcite .
although a number of studies reported on a correlation between statistical network properties ( such as degree homogeneity , cluster coefficient , and degree distribution ) and network synchronizability , the exact relationship between the propensity for synchronization of a network and its topology has not yet been fully clarified .
one of the most challenging dynamical systems in nature is the human brain , a large , interacting , complex network with nontrivial topological properties @xcite .
anatomical data , theoretical considerations , and computer simulations suggest that brain networks exhibit high levels of clustering combined with short average path lengths , which was taken as an indication of a small - world architecture @xcite .
a disorder of the brain that is known to be particularly associated with changes of neuronal synchronization is epilepsy along with its cardinal symptom , recurrent epileptic seizures .
seizures are extreme events with transient , strongly enhanced collective activity of spatially extended neuronal networks @xcite . despite considerable progress in understanding the physiological processes underlying epileptic dynamics , the network mechanisms involved in the generation , maintenance , propagation , and termination of epileptic seizures in humans are still not fully understood .
there are strong indications that seizures resemble a nonlinear deterministic dynamics @xcite , and recent modeling studies @xcite indicate the general importance of network topology in epilepsy . clinical and anatomic observations together with invasive electroencephalography and functional neuroimaging now provide increasing evidence for the existence of specific cortical and subcortical _ epileptic networks _ in the genesis and expression of not only primary generalized but also focal onset seizures @xcite .
an improved understanding of both structure and dynamics of epileptic networks underlying seizure generation could improve diagnosis and , more importantly , could advice new treatment strategies , particularly for the 25% of patients
whose seizures can not be controlled by any available therapy . in order to gain deeper insights into the global network dynamics during seizures we study in a time resolved manner statistical and spectral properties of functionally defined seizure networks in human epileptic brains .
we observe that , while seizures evolve , statistical network properties indicate a concave - like movement between a more regular ( during seizures ) and a more random functional topology ( prior to seizure initiation and already before seizure termination ) .
network synchronizability , however , is drastically decreased during the seizure state and increases already prior to seizure end .
we speculate that network randomization , accompanied by an increasing synchronization of neuronal activity may be considered as an emergent self - regulatory mechanism for seizure termination .
we retrospectively analyzed multichannel ( @xmath2 channels ) electroencephalograms ( eeg ) that were recorded prior to , during , and after one - hundred focal onset epileptic seizures from 60 patients undergoing pre - surgical evaluation for drug - resistant epilepsy .
seizure onsets were localized in different anatomical regions .
all patients had signed informed consent that their clinical data might be used and published for research purposes .
the study protocol had previously been approved by the ethics committee of the university of bonn .
eeg data were recorded via chronically implanted strip , grid , or depth electrodes from the cortex and from within relevant structures of the brain , hence with a high signal - to - noise ratio .
signals were sampled at 200 hz using a 16 bit analog - to - digital ( a / d ) converter and filtered within a frequency band of 0.5 to 70 hz . in order to minimize the influence of a particular referencing of electrodes on spatially extended correlations
, we here applied a bipolar re - referencing prior to analyses by calculating differences between signals from nearest neighbor channels .
we defined _ functional _ links @xcite between any pair of bipolar eeg channels @xmath3 and @xmath4 ( @xmath5 $ ] ) regardless of their anatomical connectivity using the cross - correlation function as a simple and most commonly used measure for interdependence between two eeg signals @xcite , which is both computationally efficient and , in light of the known correlation - based changes of synaptic strengths @xcite , also physiologically plausible .
we used a sliding window approach ( see fig .
[ fig : figure1 ] ) to estimate the elements @xmath6 of the normalized maximum - lag correlation matrix @xmath7 @xmath8 where @xmath9 denotes the cross - correlation function .
this function yields high values for such time lags @xmath10 for which the signals @xmath11 and @xmath12 have a similar course in time .
the sliding window @xmath13 had a duration @xmath14 = 2.5 s ( 500 sampling points ; no overlap ) , which can be regarded as a compromise between the required statistical accuracy for the calculation of the cross - correlation function and approximate stationarity within a window s length .
@xmath15 denotes the normalized ( zero mean and unit variance ) eeg signal at channel @xmath3 . from time - resolved matrices
@xmath16 we derived adjacency matrices @xmath17 by thresholding : @xmath18 elements on the main diagonal of @xmath19 were set to 0 in order to exclude self - connections of nodes . since our aim is to characterize the evolving _
global _ network dynamics , we chose , for each time window , the highest possible threshold for which the resulting graph represented by @xmath17 was connected . starting from a threshold @xmath20 we gradually decreased @xmath21 , and we calculated , at each step , the second smallest eigenvalue @xmath22 of the corresponding laplacian matrix @xmath23 , whose elements are @xmath24 , where @xmath25 is the kronecker delta , and @xmath26 denotes the degree of node @xmath3 .
@xmath22 is positive if and only if the graph is connected @xcite . following ref .
@xcite we used @xmath19 to compute average shortest path length @xmath0 , cluster coefficient @xmath1 , and normalized edge density @xmath27 ( i.e , the actual number of edges in @xmath19 divided by the number of possible edges between @xmath28 nodes ) and assigned their values to the time point at the beginning of each window @xmath13 . to detect deviations from a random network topology we consider the ratios @xmath29 and @xmath30 and computed @xmath31 and @xmath32 for a random graph with a preserved degree distribution and an identical average number of edges per node @xcite . in order to investigate synchronizability during seizures
, we analyzed again in a time - resolved manner the spectrum of the laplacian matrix @xmath23 .
as a measure for the stability of the globally synchronized state @xcite of a connected graph we consider the eigenratio @xmath33 , where @xmath22 denotes the smallest non - vanishing eigenvalue and @xmath34 the maximum eigenvalue of @xmath23 .
a network is said to be less synchronizable for larger values of @xmath35 and better synchronizable for smaller values of @xmath35 , the latter indicating a more stable globally synchronized state @xcite .
in fig . [ fig : figure2 ] we show typical time courses of @xmath36 and @xmath37 . given our thresholding , which can yield a different number of links for every time window , we show , in addition , the time course of @xmath38 . already prior to the electrographic seizure onset ( see ref .
@xcite for a fully automated detection of electrographic seizure onset and seizure end ) both average shortest path length and cluster coefficient indicate a slight deviation from a random network , while edge density of the seizure network remains almost constant .
approximately in the middle of the seizure this deviation is most pronounced and indicates a movement toward a more regular functional topology .
interestingly , already prior to the electrographic seizure end we observe a movement away from the more regular functional topology , which extends into the post - seizure period .
edge density slowly increases during the second half of the seizure and reaches an average value during the post - seizure period that is almost twofold the average value of the pre - seizure period .
we note that this temporal evolution is also reflected in the dynamics of maximum degree @xmath39 .
the difference between maximum and average degree ( @xmath40 ) remains almost constant at one quarter of the number of nodes while the minimum degree @xmath41 is 1 for all windows @xmath13 ( data not shown ) .
the variability of the concave - like temporal evolution of network characteristics @xmath1 and @xmath0 for different seizures from the same patient was low , and moreover , was a consistent finding for all investigated seizures independently of the anatomical location of their onset ( cf . figs .
[ fig : figure4]a and b ) .
typical time courses for @xmath42 , @xmath43 , and @xmath44 during a seizure are shown in fig .
[ fig : figure3 ] .
again we observed a concave - like temporal evolution , with highest values of @xmath35 ( i.e. , lowest synchronizability ) in the middle of the seizure , followed by a decline ( i.e. , an increasing synchronizability ) already prior to the electrographic seizure end .
although this behavior varied from patient to patient , it was a consistent finding for all seizures ( cf .
[ fig : figure4]c ) . a comparison of fig .
[ fig : figure3]b with fig .
[ fig : figure3]c shows that @xmath44 is largely dominated by the dynamics of the smallest non - vanishing eigenvalue @xmath43 , and its decrease in the middle of the seizure may indicate a reorganization of the network into local sub - structures @xcite . in this case , sparsely occurring links between local sub - structures can significantly affect @xmath22 @xcite .
the relative change of largest eigenvalue @xmath42 during the seizure is less pronounced as compared to that of @xmath43 and resembles the time course of edge density @xmath38 ( cf .
[ fig : figure3]d ) .
this similarity can be expected , at least to some extent , since edge density constitutes a lower bound for the largest eigenvalue , @xmath45 ( cf .
we could , however , not observe such a clear cut influence of @xmath27 on @xmath22 and hence on @xmath35 . as regards the degree distribution , we again observed @xmath46 and @xmath47 for all windows and a temporal evolution of @xmath39 quite similar to the dynamics of @xmath38 ( data not shown ) indicating that the degree distribution of the seizure network does not appear to determine its synchronizability ( cf .
@xcite ) . fig .
[ fig : figure4 ] summarizes the dynamics of functional network properties and synchronizability for all one - hundred focal onset seizures .
irrespective of the anatomical location of seizure origin , both the normalized cluster coefficient and the normalized average shortest path length rapidly increased during the first half of the seizures then gradually decreased again .
interestingly , this temporal evolution was more pronounced for the normalized cluster coefficient than for the normalized average shortest path length .
this indicates a relative shift toward a less random functional topology of the seizure state .
seizures are usually associated with massively synchronized brain activity @xcite , and the significantly decreased synchronizability of the underlying functional topology may catalyze the emergence of a globally synchronized state of the epileptic brain .
once such a state has been established , synchronizability increases again , as observed in our data .
if the observed changes of functional topology were simply a consequence of enhanced volume conduction during the seizures , i.e. , due to direct propagation of distant sources to remote sensors , the covariance of the eeg signals would be expected to occur with zero time lag . in order to exclude the effect of a linear superposition of sources leading to a more regular - looking graph , we estimated the normalized frequency distribution of absolute time lag of maximal correlation @xmath48 ( cf .
( [ eq : rho ] ) ) for all seizures , partitioned into 10 equidistant time bins ( cf . fig .
[ fig : figure4 ] ) .
we observed @xmath48 to peak in the range of 550ms ( see fig .
[ fig : figure5 ] ) , which indicates that the observed changes of functional topology are not due to passive electromagnetic field effects in the extracellular space , but rather due to propagation of electrical activity along anatomical pathways .
we have presented findings obtained from a time - resolved analysis of statistical and spectral properties of functionally defined networks underlying human epileptic seizures . despite the many influencing factors ( number of nodes , non - uniform arrangement of sensors , focus on particular brain regions , choosing a threshold for the extraction of functional networks , etc . )
that impede an interpretation of graph - theoretical measures in a strict sense when analyzing field data , our results indicate that seizure dynamics irrespective of the anatomical onset location can be characterized by a relative transient shift toward a more regular and then back toward a more random functional topology .
this is consistent with recent observations reported in ref .
@xcite , where a small number of seizures that originated from a circumscribed brain region have been analyzed .
we here observed that the changing functional network topology during seizures was accompanied by an initially decreased stability of the globally synchronized state , which increased already prior to seizure end . in a previous study @xcite we analyzed the same data set using multivariate time series analysis techniques from random matrix theory and observed that
surprisingly global neuronal synchronization ( derived from the eigenvalue spectrum of the zero - lag correlation matrix ) significantly increased during the second half of the seizures , and before seizures stopped .
our present findings indicate that such a global increase of neuronal synchronization prior to seizure end may be promoted by the underlying functional topology of brain dynamics .
this corroborates the hypothesis @xcite that increasing synchronization of neuronal activity may be considered as an emergent self - regulatory mechanism for seizure termination .
thus , our result can provide clues as to how to control seizure network , e.g. via pinning @xcite .
while the aforementioned interpretation would indicate that the transient evolution in graph properties is an active process of the brain to abort a seizure , our findings could also be understood as a passive consequence of the seizure itself .
the extremely intense firing of neurons during a seizure might lead to a saturation of the capacity of neurons to fire , particularly in brain areas with a high number of functional links .
if such hubs are saturated and become silent , then the connections between local sub - structures are affected to a larger extent , which could lead to a segregation of the global network and to a decrease of global synchronizability .
thus , a reliable identification of local sub - structures or hubs could improve understanding of mechanisms underlying the generation , maintenance , propagation , and termination of epileptic seizures . at present ,
our findings are restricted to interdependences that can be assessed by the cross - correlation function .
future studies will clarify whether additional information can be gained from analyses invoking other time series analysis techniques , including nonlinear ones , as well as techniques that take into account the direction of interactions .
nevertheless , disentangling the interplay between connection structure and dynamics of the complex network human brain may advance our understanding of epileptic processes .
we thank philip h. goodman and christof cebulla for helpful comments .
k. s. was supported by a scholarship of the ssmbs ( schweizerische stiftung fr medizinisch - biologische stipendien ) donated by roche .
s. b. was supported by the german national academic foundation ( studienstiftung ) .
h. , c. e. e. , and k. l. acknowledge support from the deutsche forschungsgemeinschaft ( grant nos . sfb - tr3 sub - project a2 and le660/4 - 1 ) . | we assess electrical brain dynamics before , during , and after one - hundred human epileptic seizures with different anatomical onset locations by statistical and spectral properties of functionally defined networks .
we observe a concave - like temporal evolution of characteristic path length and cluster coefficient indicative of a movement from a more random toward a more regular and then back toward a more random functional topology .
surprisingly , synchronizability was significantly decreased during the seizure state but increased already prior to seizure end .
our findings underline the high relevance of studying complex systems from the view point of complex networks , which may help to gain deeper insights into the complicated dynamics underlying epileptic seizures .
* epilepsy represents one of the most common neurological disorders , second only to stroke .
patients live with a considerable risk to sustain serious or even fatal injury during seizures . in order to develop more efficient therapies the pathophysiology
underlying epileptic seizures should be better understood . in human epilepsy , however , the exact mechanisms underlying seizure termination are still as uncertain as are mechanisms underlying seizure initiation and spreading .
there is now growing evidence that an improved understanding of seizure dynamics can be achieved when considering epileptic seizures as network phenomena . by applying graph - theoretical concepts
, we analyzed seizures on the eeg from a large patient group and observed that a global increase of neuronal synchronization prior to seizure end may be promoted by the underlying functional topology of epileptic brain dynamics .
this may be considered as an emergent self - regulatory mechanism for seizure termination , providing clues as to how to efficiently control seizure networks . * |
d - branes are very important non - perturbative objects in string theory .
their existence is essential since they are needed , among other things , for various string theory dualities to work .
the most primitive definition of a d - brane is as a hyper surface on which open strings end . in this work
we are interested only in d0-branes .
they play a particularly important role since they are the basic constituents of matrix theory @xcite , which , being a suggestion for a non - perturbative definition of string theory ( or m - theory @xcite ) , deserves particular attention .
it is also possible to think about them as bound states of higher dimensional unstable d - branes @xcite .
furthermore , as was shown by myers @xcite , they can also form bound states with all the properties of higher dimensional branes .
it is therefore clear that d0-branes have many interesting properties that make them worth studying in detail .
concretely this means studying supersymmetric quantum mechanics which is an interesting topic in itself .
although this topic has been studied before @xcite-@xcite we feel that there are still unresolved issues .
in particular one could use a computer to compute properties of d0-branes which are not possible to address analytically because of the complexity of the theory .
this is the goal of this work , to develop a `` 0-brane quantum chemistry '' .
it should be mentioned that similar issues have been addressed in @xcite .
however , we use a different method which we compare to the method developed in @xcite ( to which we also suggest certain improvements ) . more concretely ,
in this work we try to find the bound state of two d0-branes in three dimensional minkowski space ( which is really a toy - model for the real situation , d0-branes in ten dimensional minkowski space ) .
it should also be noticed that the really interesting cases where the d0-brane theory is thought to describe macroscopic supergravity states is achieved by taking the number of zero branes to infinity .
the low energy physics of @xmath0 parallel d@xmath1-branes is governed by the dimensional reduction of @xmath2 dimensional @xmath3 supersymmetric yang - mills theory with @xmath4 gauge group to @xmath5 dimension @xcite .
the center of mass motion is governed by the overall @xmath6 factor so if we are interested in relative motion only , we can choose the gauge group to be @xmath7 .
in the case we are interested in , the relative motion of two 0-branes , we thus choose the gauge group to be @xmath8 . the action is @xcite @xmath9 \label{full}\end{aligned}\ ] ] and the hamiltonian derived from this is @xcite @xmath10 together with the constraint one gets from varying ( [ full ] ) with respect to @xmath11 @xmath12 on quantum level , we restcrict our hilbert space to vectors which satisfy @xmath13 that is , our physical space is gauge invariant because @xmath14 are gauge generators . we will only study motion of two @xmath15-branes in three dimensional minkowski space . we hope that this gives us the basic behavior of @xmath15-branes and also an understanding of the full problem .
this problem of two branes is also described in @xcite but we study it in a different way . in the three dimensional case
the action is given by dimensional reduction of @xmath3 supersymmetric yang - mills theory with @xmath8 gauge group in @xmath16 dimension to @xmath17 dimension .
the hamiltonian takes the slightly simpler form @xmath18 where @xmath19 are fields in the @xmath8 adjoint representation and @xmath20 is a complex fermion also in the adjoint representation .
of course we still have to impose gauge invariance ( [ phys ] ) .
in fact , the gauge invariance complicates things somewhat since we would like to separate out gauge invariant degrees of freedom from pure gauge degrees of freedom in our basic quantum mechanical operators @xmath21 and @xmath22 .
let us focus on physical content of the @xmath23 .
it contains six components ( the gauge index runs over three values and the space index @xmath24 runs from 1 to 2 ) .
we know that we can remove three of these variables using gauge transformations so only three variables are observable .
these three variables should describe the relative position of two pointlike objects in two space dimensions .
we draw the conclusion that one of the physical variables do not have the interpretation of a coordinate but rather as some internal auxilliary degree of freedom .
to get some further insight into this problem it is neccesery to investigate the bosonic vacuum of the theory .
it is possible to explicitly separate the gauge degrees of freedom from @xmath25 by decomposition it in matrix form @xcite @xmath26 here the matrix @xmath27 is an group element in the adjoint representation of @xmath8 .
thus when the gauge group acts on @xmath21 , @xmath27 just changes by ordinary gauge group multiplication ( from the left ) .
we will parametrise the group element @xmath27 by the `` angles '' @xmath28 ( [ psi ] ) .
this decomposition has the advantage that all the gauge dependence sits in @xmath27 and all the other matrices are gauge invariant . in an analogous way we have separated out the dependence on rotations in space .
namely , performing an @xmath29 rotation in space we have an element of @xmath29 acting from the right on the matrix @xmath30 .
thus we can separate out the dependence on the angle in space ( we will call it @xmath31 ) by saying that @xmath32 is a group element of @xmath29 .
we are left with the matrix @xmath33 ( [ rt ] ) which by construction is both gauge and space rotation invariant @xmath34 the bosonic potential in ( [ toy ] ) is gauge and rotation invariant and in the new decomposition coordinates depends only on two @xmath35 which have length dimension ( fig .
[ picpot ] ) .
the parametrisation @xmath36 is the only way how to obtain exactly one variable , @xmath37 , with the dimension length , which could represent relative distance of two branes .
the dimensionless @xmath38 is the auxilliary coordinate .
the potential in this coordinate reads @xmath39 looking at the picture we can draw some interesting conclusions . if we fix a point on the bosonic vacuum ( a classical static configuration with minimum energy ) , that is on the axes , we can study the behavior of the potential for small fluctuations of the auxilliary variable @xmath38 .
we see that for large @xmath37 the @xmath38 fluctuation are very much suppressed but at small @xmath37 , @xmath38 will be allowed to fluctuate .
this can be interpreted to mean that when the branes get close to each other , they can start to move also in the @xmath38 direction .
thus , @xmath38 is an auxilliary coordinate which is visible only when the branes come close together .
the above discussion included only the bosonic degrees of freedom , we should keep in mind that the fermionic degrees of freedom can ( and will ) change this behavior somewhat .
in essence , the pauli repulsion will try to spread out the wavefunction as much as possible .
when we use the standard operator representation of @xmath40 and @xmath21 the first term in the hamiltonian ( [ toy ] ) is proportional to the laplacian which we have to rewrite in the decomposition coordinates above to separate out gauge using formula @xmath41 where the @xmath42 is the metric which is trivial in @xmath21 coordinates .
it is also good idea to rewrite the laplacian in terms of gauge angular momenta @xmath43 : @xmath44 and physical angular momentum @xmath45 since their action on states with given gauge and rotational properties is simple .
in particular , the ( bosonic ) ground state should have total spin equal to zero and be gauge invariant .
all angular momenta are killing vectors of the metric . in appendix [ applaplace ]
we derive the lagrangian expressed in terms of gauge invariant variables and angular momenta @xmath46 here @xmath47 @xmath48 is a compact notation for the ( physical and gauge ) angular momenta defined above .
furthermore , we have defined @xmath49 and @xmath50 @xmath51 to find the ( bosonic ) ground state we will find all gauge invariant states with spin zero .
the first state is the vacuum state @xmath52 .
then we may act with the fermionic creation operators @xmath53 on the vacuum to find new states .
the following states has total spin zero and they are gauge invariant @xmath54 that is , they satisfy @xmath55 the most general gauge invariant wavefunction with total spin zero can then be written @xmath56 it would also possible to construct the superpartner ground state with the help of @xmath57 now we study what happens when we act with the laplacian ( [ laplacian ] ) on this wavefunction and using the result we write the hamiltonian matrix in this base . let us start with second part of the laplacian which is the relevant bosonic piece @xmath58 where @xmath59 is only on the gauge indexes .
finally the fermionic interaction in the hamiltonian gives us @xmath60 so the full hamiltonian matrix elements are @xmath61 where @xmath62 notice that the hamiltonian we have obtained is the same as in @xcite but differs from the one used in @xcite .
this section gives an overview of solving the hamiltonian eigenvalue problem ( [ toy ] ) using the numerical renormalized numerov method @xcite . with it
one can solve for the discrete spectra of a one dimensional operator of the form @xmath63 which acts on @xmath64\otimes{\bf c}^n$ ] with dirichlet boundary conditions .
however , our hamiltonian depends on the _ two _ coordinates @xmath37 and @xmath38 ( the other coordinates , the angles , are fixed by the requirement that we are studying only gauge invariant states with spin zero ) .
the `` kinetic '' term in our hamiltonian ( i.e. the term which contains the derivatives ) is @xmath65 and we see that it is naively not of the form required above . let us sketch briefly how to modify our problem to be able to apply the method .
an arbitrary wavefunction can be written in the form @xmath66 where @xmath67 is a complete basis of functions ( with apropriate boundary conditions ) in @xmath38 ( the explicit choice of basis does not have to be the same for different values of the index @xmath24 but can be chosen to optimize the numerics ) .
the functions @xmath68 ( which we will compute by the numerov method ) one can think of as being `` combination coefficients '' depending continuously on @xmath37 .
@xmath69 is the standard base spanning @xmath70 , in our case it is the four dimensional complex vector space on which the hamiltonian matrix acts . choosing a concrete basis depends only on the hamiltonian domain which depends on one of the selfconjugated extensions of the hilbert space ( an extension of the hilbert space such that the hamiltonian operator is hermitian ) but we will rather apply a physical principle which will be described later .
the expectation values of the hamiltonian in the base @xmath71 gives the same number of coupled equations for the radial part as the number of @xmath72 in ( [ state ] ) .
thus the problem is now correctly defined and the matrix representation of the expectation value of the hamiltonian in the basis above forms the potential which is used in the one dimensional hamiltonian ( [ op ] ) . to get @xmath73 in the kinetic term we have to orthonormalize the base .
furthermore , to be able to use the numerov method on a computer we need a finite basis which means that we need to `` cut off '' or restrict the base to be finite .
the rescaling of the wavefunction @xmath74 by the factor @xmath75 transforms the radial part of ( [ kinetic ] ) to the operator @xmath76 and changes boundary condition to dirichlet .
let us now describe the physical principle we use to choose the basis functions @xmath77 for our hamiltonian ( [ toy ] ) .
we define the wavefunction on the interval @xmath78 $ ] ( [ domain ] ) .
it is not necessary to choose this particular interval , one could , for instance , select the interval @xmath79 $ ] instead of the above mentioned .
using the identification @xmath80 the hamiltonian defined on the interval @xmath81 $ ] acting on states with total spin zero is connected to our original hamiltonian on the interval @xmath78 $ ] by the unitary transformation @xmath82 where @xmath83 it is also possible to consider other interval @xmath84 $ ] . using the identification @xmath85 the corresponding hamiltonian can be also obtained by a unitary transformation with the matrix @xmath86 the wavefunctions of course also transform under the unitary transformation @xmath87 if we require that the wavefunctions be everywhere smooth , the above condition severely restricts the possible wavefunctions and in particular the basis wavefunctions that we can use .
we note that this principle was not used in @xcite and hence the @xmath38 derivative of their groundstate wavefunction at @xmath88 is not well defined . for the bosonic part of the wavefunction we will use the basis states @xmath89 and for the fermionic part we will use @xmath90 they nicely cancel the divergences in the hamiltonian coming from the laplacian .
of course these basis functions satisfy the boundary condition above and they also form a complete basis for the functions with the given boundary conditions . however , the basis is not orthonormal due to the non trivial @xmath38 part in the measure ( [ measure ] ) .
so we need to orthonormalize these basis functions to get @xmath73 in ( [ op ] ) .
doing so , using the gramm - schmidt procedure , we diagonalize the @xmath38 part of the laplacian .
notice that these two sets of basis functions also correspond to the bases which are implicitly used in the wosiek method @xcite when one are doing calculations with states with total spin zero ( the basis functions above depend on the total spin ) .
this will be shown in section [ comparison ] .
thus we will be able to compare our results and our method very directly with the results obtained by the wosiek method @xcite .
it is not obvious how much our results for a fixed number of basis functions fit the exact solutions which one would get using the complete basis .
to get some intuition for how the general solution would look like we will repeat the calculations increasing the number of basis functions each time and hopefully one can extrapolate the result to the exact case .
at least we should be able to make an intelligent guess at the properties of the exact solution . to study the groundstate of our hamiltonian ( which ,
because of supersymmetry , should have energy zero @xcite ) is a good test for the method described above .
there are some results for coupling constant @xmath91 in the table [ tab ] ( p. ) where @xmath0 is the number of the test functions ( [ boson],[fermion ] ) and @xmath92 is their corresponding groundstate energy .
the dependence of the energy on the number of functions we include in the basis given in the last two columns of the table is as follows @xmath93 as was claimed in @xcite , one can therefore predict the groundstate energy as a function of the number of basis functions with very high accuracy .
we therefore see that in any concrete numerical calculation ( using a finite basis ) we do not expect to get zero energy .
only in the ( numerically unobtainable ) case of infinite basis do we get zero energy .
the picture ( fig . [ picden ] ) is the probability density for the case with the highest number of basis functions in the table .
the following two pictures show the bosonic contribution coming from ( [ boson ] ) and the fermionic one coming from ( [ fermion ] ) .
the domain of these plots is @xmath94\times[-\pi/4,\pi/4]$ ] .
the hill of the probability density is located at the boson potential valley ( [ potential ] ) and isolines represents sections for fixed @xmath37 , @xmath38 and one for fixed density on each picture .
the maximum of the probability density of any constant @xmath37 section is in the potential valley ( @xmath95 ) .
notice that the global maximum is not at @xmath96 .
this is probably an effect of the fermion pauli repulsion which can be seen from the purely fermionic contribution to the probability density which is zero at @xmath96 . increasing the number of basis functions
, the only thing that happens is that the global maximum moves slowly to larger and larger @xmath37 at the same time as the whole wavefunction becomes more spread out in @xmath37 but more peaked in @xmath38 .
one can expect that considering the complete base ( [ base ] ) the ground state density will be the same near the origin as in the picture and also will have the hills on the valleys of the potential which will be sharper and sharper when we follow the potential valley to large @xmath37 .
one can not conjecture in this case the behavior of the hills by this numerical method .
rather one has to use other methods @xcite for the asymptotic behavior of the wavefunction at large @xmath37 .
here we briefly review the body of paper @xcite where a different approach to solving the eigenvalue problem of a hamiltonian is given .
the main idea is to choose a finite subset of a complete basis of the hamiltonian hilbert space and then find the combination which minimizes the lowest lying energy for instance .
a lot of hamiltonians have the form ( [ op ] ) which can be easily rewritten in terms of the harmonic oscillator @xmath97 creation and annihilation operators @xmath98 . this is true even in the case where the potential @xmath99 is not quadratic .
the complete base above is formed by eigenvectors of this linear harmonic oscillator which are constructed by creation operators acting on the vacuum .
if we choose a finite subset of this basis , calculating hamiltonian matrix elements is very easy . in the harmonic oscillator hamiltonian ( [ hamosc ] ) which we used to define the basic creation and annihilation operators
there is an arbitrary parameter @xmath100 in the potential which we may use to optimize the numerics .
the option to use @xmath100 to optimize the calculations was invented by us and should be regarded as suggestion for an improvement of the method presented in @xcite .
concretely we will do it like this : first we choose a subset of the basis functions , we then compute the energy ( or whatever else we would like to calculate ) for different values of @xmath100 .
we then find the value of @xmath100 which gives the minimum value for the energy .
it turns out that it is much more effective to minimize the energy in terms of @xmath100 than to try to use a larger basis since the computer time used for a calculation increases exponentially with the number of basis functions included .
there is one extra complication in our problem . because of the gauge symmetry one should also restrict the basis to include only gauge invariant states .
the generalization to the fermionic part and to the higher dimensional problem is straightforward .
let us illustrate the method above on our hamiltonian .
first of all we have to form all gauge invariant states which can be obtained by acting with a combination of creation operators on the vacuum .
the @xmath8 algebra has only two tensors which can form scalars from vectors @xmath101 these vectors are formed by creation operators @xmath102 and a combination of them has to be contracted in gauge indexes with the special tensors above to get a gauge scalar .
let us write all non zero fundamental possibilities @xmath103 where any combination of them acting on a gauge invariant state gives also a gauge invariant state .
one has to start with the vacuum to generate all gauge invariant state .
however , if we are looking for the groundstate we are interested only in states which can form a state with total spin zero . that is only states which consist of the fermion vacuum or two creation fermion operators acting on it because these are the only cases where the fermionic contribution to the total angular momentum can be canceled by the bosonic angular momentum .
so using the operators @xmath104 acting on @xmath52 and on the fermion states @xmath105 gives the appropriate complete basis for our hamiltonian .
choosing a finite subset of the basis states allows us to calculate matrix elements of any operator on the computer .
it is a good idea to orthonormalize the states in the basis to avoid problems in the eigenvalue calculation when the metric is non trivial .
see the next section for the explicit form of first few states in the basis .
we are able to illustrate the effect of the optimized @xmath100 now .
let us show what happens with the ground state energy when we calculate with fixed @xmath100 but with different coupling constants @xmath106 $ ] in our hamiltonian ( [ toy ] ) . in this calculation
we have chosen the restriction of the basis to include up to six creation operators in the bosonic basis and up to five creation operators in the fermion part .
the result is given by the dotted line in the figure [ zeta ] .
it seems that the energy of the ground state _ diverges _ when the coupling constant goes to zero which is clearly wrong !
the way out of this dilemma is to remember that we also have the parameter @xmath100 at our disposal .
if we for each @xmath107 use the @xmath100 which gives the minimal energy , we get the dependence of the ground state energy on the coupling constant as drawn in bold line .
this is clearly consistent with the @xmath108 dependence one expects from dimensional arguments .
this is a very nice illustration of the power of our suggestion for improving the wosiek method @xcite by including the @xmath100 parameter .
one can guess that there has to exist a coupling constant such that the optimalization is not needed .
this is also viewable from the picture .
there are also some results for coupling constant @xmath91 in the ( tab .
[ tab ] , p. ) where the meaning of the columns is described above the table .
the energy of the bold line for @xmath91 corresponds to the first two columns in the table for the highest order obtained by the wosiek method .
we draw the density plot ( fig .
[ figw ] . ) for this calculation on which we can see the bump which for a larger number of basis functions develops into the long and thin hill in the picture on page . its domain is @xmath109\times[-\pi/4,\pi/4]$ ] .
to compare our method with the wosiek method it is important to know the correspondence between the base ( [ boson],[fermion ] ) and the base formed by creation operators .
since we have written our hamiltonian in terms of angular momentum operators it is useful to classify all states in terms of angular momenta .
in fact , this is even more useful since when we are discussing the ground state with total spin zero .
we may rewrite the angular momentum operator ( [ momentum ] ) as @xmath110 using the correspondence between creation / annihilation and @xmath111 operators @xmath112 let us write down how the lowest lying angular momentum states look like .
first we do the purely bosonic sector .
the ( zero angular momentum ) ground state @xmath113 in the coordinate representation looks like @xmath114 next we can write down all gauge invariant states created by two bosonic oscillators .
@xmath115 the following linear combinations @xmath116 are orthonormal eigenvectors of the angular momentum operator with eigenvalues @xmath117 . using formulas for the creation / annihilation operators in the @xmath118 representation we may write the coordinate representation of these states as @xmath119
the angular momenta spectrum of the next six states all formed by four creation operators is @xmath120 where the two orthonormal states with zero angular momenta are given by @xmath121 they give the first non trivial wave function dependence on @xmath38 in the zero angular momentum sector .
the @xmath38 part in higher order states with zero angular momenta reproduce the basis ( [ boson ] ) which we guessed as the fourier basis for the functions with the correct boundary conditions .
we see that a new @xmath122 appears when there appears a new @xmath123 in the angular momenta spectrum . for reference
we also give the spectra for the states created by six and eight creation operators @xmath124 and @xmath125 . with this
we close the discussion about the purely bosonic part . in the same way we can study the fermionic part .
the wave function that interest us most are the ones with angular momentum one since they are the ones which contribute to the ground state with total angular momentum zero . on the lowest level in the creation / annihilation operator basis
we have two gauge invariant orthonormal states @xmath126 they can be combined into the orthonormal states @xmath127 which are eigenstates of angular momenta with eigenvalues @xmath128 .
it is again possible to rewrite these states in the coordinate basis with help of the orthonormal and gauge invariant purely fermionic states ( [ states ] ) as @xmath129 the @xmath38 part of the second state gives the first function in the basis ( [ fermion ] ) .
the states in the next level consisting of the six states we get by acting with all gauge invariant combinations of two bosonic creation operators on the previous two states have angular momentum spectrum @xmath130 where the two orthonormal states with angular momentum one are given by @xmath131 this agrees with the second function in our fermionic basis .
continuing we find that the @xmath38 part of the angular momentum one states at higher levels reproduce the basis ( [ fermion ] ) which we have used .
writing the angular momentum spectrum for the next level ( states obtained by acting with all gauge invariant combinations of four bosonic oscillators on the original fermionic states ) we have @xmath132 .
this closes the discussion about the fermionic part and again gives a hint how the general spectrum looks like .
we have now achieved our main goal to rewrite the states in the wosiek method in terms of angular momentum eigenstates in order to be able to compare the two methods . since we now know at what level a particular basis function appears in the wosiek method , we know where we should `` cut off '' the set of basis functions used in our method in order to get an equivalent calculation .
furthermore , doing the calculation using the two different methods at the same level we should still expect that our method should give better results since in our method the radial part of the wave function is not a priori fixed while in the wosiek method the radial part comes together with the angular part , see for instance ( [ fixed ] ) .
also , our method should be more effective for fixed spin calculation on the computer since we have the possibility to use only the angular momentum eigenstates we need while in the wosiek method one is using all possible states which leads to a huge number of states not contributing to the wavefunction but absorbing space and time in the computer . to illustrate this we compose the following table .
.results [ cols="^,^,^,^,^,^",options="header " , ] we compare the ( improved ) wosiek method with @xmath133 boson and fermion creation / annihilation states of the type ( [ first],[second ] ) with our method using only the test functions ( [ boson],[fermion ] ) to the corresponding level for @xmath91 .
the @xmath100 given in the table is the optimized value for the ( improved ) wosiek method , the @xmath0 counts the number of the test functions corresponding to the level used in the wosiek method and the groundstate energies obtained by both methods are also given .
we see that indeed the energy calculated using our method is lower than in the wosiek method .
furthermore it is clear that the number of states used in the wosiek method increases uncontrollably as compared to the increase of basis functions used in our method .
finally we also compare probability density sections for @xmath95 in the case of the third line in the table ( the last case where both methods have been used ) on the domain @xmath134 $ ] .
the states which generate these sections have the same norm to compare them correctly .
we have also included higher order wave functions calculated using our method .
the labels indicate which order they correspond to .
these wave functions are not equally normalized but have rather been chosen to have the same value at @xmath96 in order to illustrate the behavior of the wavefunction when the order is increased .
we see that the maximum moves outwards at the same time as the wavefunction becomes wider .
the aim of the present work was to find an approach how to calculate the boundstate of two 0-branes .
this we have achieved with good results .
in the course of our work we also found a way to improve the wosiek method but our method still gives better results .
as this work was being completed the closely related paper @xcite by wosiek ( which improve his original method ) appeared .
we have also calculated the groundstate probability density near the origin with high accuracy which gives a basic intuition about the physics of branes on the string scale .
in particular about the meaning of the auxilliary coordinates which become important at small distances between the branes .
it is for example interesting to observe that the most probable position of the branes is not on top of each other but rather at some small distance away from each other .
this we understand as an effect of the pauli fermionic repulsion .
it is possible to apply the method described here for branes in higher dimensional minkowski space .
however there are a lot of complications which do not allow numerical results of high precision .
namely , there are many more auxilliary coordinates in the higher dimensional case ( or in the case where there are more than two branes ) .
since our basis functions are functions of the auxilliary coordinates , this implies that our basis functions will be multi dimensional functions and thus the number of components that the computer has to handle increases . naively speaking
the number of effective radial wavefunction in the state computed by the numerov method for a given dimension is the number in our toy model raised to the power of the number of auxilliary coordinates .
this fact of course taxes the computer very heavily .
another important point that we wanted to study was how the branes behave in external ( generalized electric - magnetic ) fields .
the potential coming from these background fields represents this interaction . however , our two dimensional toy model is too simple to display any interesting effects .
again one would have to go to higher dimensional minkowski space to do be able to do some interesting calculations . as a first attempt to go further it would therefore be nice to consider three dimensional minkowski space which would allow more general background fields which coming from other types of branes .
however , we do feel that doing _ realistic _
calculations of the full ten dimensional ( or even four dimensional ) theory or including more than two 0-branes would not be possible even on a super computer .
this is one of the reasons why we have not pursued this issue further .
i am very grateful to rikard von unge for continuous support and i also thank to milan indelka which told me of his experiences in numerical methods which can solve quantum mechanics problems .
it is possible to decompose @xmath25 @xcite in matrix form as ( [ decomposition ] ) @xmath135 where @xmath27 is a group element in the adjoint representation of the gauge group and @xmath32 is a group element of the rotation group @xmath29 .
the matrix @xmath33 is then gauge and rotation invariant .
we may now choose our physical coordinates as @xmath136 together with @xmath137 where @xmath138 is the generator of spatial rotation @xmath139 the matrix @xmath27 which describes the gauge degrees of freedom is parametrized by euler angles @xmath28 @xmath140 where the matrices are generators of @xmath8 in the adjoint representation .
@xmath141 the flat metric in the original @xmath142 coordinates is trivial and can be written in matrix notation as @xmath143 the decomposition ( [ decomposition ] ) leads to the expression @xmath144 and then the metric is @xmath145 we see here that the differentials in @xmath146 are not mixed with the other ones and thus the metric will be block diagonal @xmath147 where the @xmath148 block is associated with the @xmath146 differential and @xmath149 is associated with the differentials @xmath150 .
now we compute the metric term by term .
the @xmath33 part in the case of two branes is @xmath151 or explicitly in variables @xmath152 from ( [ rt ] ) @xmath153 next we take care of the terms which contribute to the @xmath149 part of the metric .
using the fact that @xmath154 the purely gauge part @xmath155 can be written with help of the left - invariant one forms defined by the @xmath27 being a group element in the adjoint representation of @xmath8 @xmath156 where @xmath157 are the generators . in our coordinates
we explicitly have the expressions @xmath158 introducing the matrix @xmath159 of the coefficient of the @xmath160 1-forms @xmath161 the gauge part ( [ gaugepart ] ) is @xmath162 where @xmath163 and the diagonal matrix @xmath164 has elements @xmath165 the calculation of the contribution from terms which contain @xmath166 is very similar . introducing the right - invariant one form defined by @xmath167 and using our parametrisation ( [ eta ] ) we have @xmath168 and @xmath169 in this expression it is necessary to write the @xmath170 ( [ leftforms ] ) explicitly . since we know that the metric is block diagonal ( [ diag ] ) we will write the @xmath149 part of the metric compactly
this is done by writing the matrix ( [ boldpi ] ) @xmath171 and the structure matrix @xmath172 then it is possible to write the sum of ( [ gauge],[angular ] , [ mixing ] ) as @xmath173 where @xmath174 .
so @xmath175 and the full metric is @xmath176 it is now easy to find the square root of the metric determinant @xmath177 which also gives us the integration measure . on the domain of @xmath178 @xmath179\times(0,2\pi ]
\times(0,2\pi]\times(0,2\pi]\times(0,\pi ] \label{domain}\end{aligned}\ ] ] the transformation rule ( [ decomposition ] ) is one to one and on its inside the jacobian is nonzero and thus regular which is important in calculation of derivatives with respect to @xmath25 in terms of these coordinates .
the metric in the new coordinates is ( [ metric ] ) @xmath180 where @xmath174 , @xmath181 is defined by ( [ omega ] ) and @xmath182 by ( [ boldpi ] ) . from the metric or
directly from ( [ expanded ] ) one can see that the metric has the block diagonal structure ( [ diag ] ) @xmath183 where @xmath148 is a diagonal matrix being the metric in the @xmath152 variables and @xmath149 is the metric in the other variables .
thus we are able to calculate the inverse of the full metric in terms of the inverses of @xmath148 and @xmath149 separately .
since @xmath148 does not depend on the gauge variables , the laplacian also splits into two pieces @xmath184 where the latin index runs over the @xmath185 variables and the greek indexes over the angular variables .
using the square root of the metric determinant ( [ measure ] ) the first part of the laplacian generated by @xmath148 is @xmath186 at this point we would like to rewrite the second part of the laplacian in terms of the angular momenta ( [ momentum],[gaugemomenta ] ) @xmath187 we notice that the angular momenta are equal to the vector field @xmath188 being dual to the _ right _ invariant one form @xmath189 @xmath190 since the right invariant one forms induce left translations , they are killing fields of the metric . also , since we know the relation between the left and right invariant one forms @xmath191 we may transform the expression for the metric , which is expressed in terms of @xmath192 into an expression in terms of @xmath188 which is then equivalent to an expression in terms of the angular momenta .
explicitly we write the vector field @xmath193 in terms of coefficient matrices @xmath194 inverse to the matrices @xmath195 introduced in ( [ omegadef ] ) . the relation ( [ lrrel ] ) then implies @xmath196 to treat all the angles on equal footing , we incorporate also the physical angle @xmath31 into this formalism by extending the three by three matrices @xmath197 to four by four matrices by putting @xmath198 in the `` zeroth '' row and column .
these new matrices we denote by bold letters , for instance @xmath199 now all vector fields obtained from @xmath200 are equal to @xmath201 times the physical and the gauge angular momenta and they are all killing vectors of the metric
. we can now write @xmath202 ; @xmath203 .
we may now rewrite the angular part of the laplacian in terms of the angular momenta which are given by @xmath204 .
the appropriate part in the laplacian is @xmath205 first we expand the formula for the laplacian @xmath206 the first term is reduced to @xmath207 where we have used the killing equation for @xmath208 . in the same way the second term becomes @xmath209 and
finally the third part is @xmath210 so the sum of these three contributions gives @xmath211= -({\bf\psi\pi}^{-1}{\bf\psi}^{-1})_{\mu\nu}l^\mu l^\nu\nonumber\\\end{aligned}\ ] ] where we have used that the angular momenta are defined in terms of the matrix @xmath200 .
then the whole laplacian is @xmath212 d. bigatti and l. susskind , `` review of matrix theory , '' .
t. banks , w. fischler , s. h. shenker and l. susskind , `` m theory as a matrix model : a conjecture , '' phys .
d * 55 * ( 1997 ) 5112 .
hoava , `` type iia d - branes , k - theory , and matrix theory , '' adv .
theor . math .
* 2 * ( 1999 ) 1373 [ arxiv : hep - th/9812135 ] .
e. witten , `` d - branes and k - theory , '' jhep * 9812 * ( 1998 ) 019 .
r. c. myers , `` dielectric - branes , '' jhep * 9912 * ( 1999 ) 022 .
j. wosiek , `` spectra of supersymmetric yang - mills quantum mechanics , '' nucl .
b * 644 * ( 2002 ) 85 [ arxiv : hep - th/0203116 ] .
d. kabat and p. pouliot , `` a comment on zero - brane quantum mechanics , '' phys .
* 77 * ( 1996 ) 1004 [ arxiv : hep - th/9603127 ] . m. trzetrzelewski and j. wosiek , `` quantum systems in a cut fock space , '' [ arxiv : hep - th/0308007 ] . m. campostrini and j. wosiek , `` exact witten index in d = 2 supersymmetric yang - mills quantum mechanics , '' phys .
b * 550 * ( 2002 ) 121 [ arxiv : hep - th/0209140 ] .
j. wosiek , `` supersymmetric yang - mills quantum mechanics , '' . b. de wit , m. luscher and h. nicolai , `` the supermembrane is unstable , '' nucl .
b * 320 * ( 1989 ) 135 .
m. b. halpern and c. schwartz , `` asymptotic search for ground states of su(2 ) matrix theory , '' int .
j. mod .
a * 13 * ( 1998 ) 4367 .
u. h. danielsson , g. ferretti and b. sundborg , `` d - particle dynamics and bound states , '' int .
j. mod .
phys . a * 11 * ( 1996 ) 5463 .
v. g. kac and a. v. smilga , `` normalized vacuum states in n = 4 supersymmetric yang - mills quantum mechanics with any gauge group , '' nucl .
b * 571 * ( 2000 ) 515 [ arxiv : hep - th/9908096 ] .
e. witten , `` bound states of strings and p - branes , '' nucl .
b * 460 * ( 1996 ) 335 [ arxiv : hep - th/9510135 ] .
b. r. johnson , `` the renormalized numerov method applied to calculating bound state of the coupled - channel schroedinger equation , '' j. chem
* 69 * ( 1978 ) 10 | we apply two different numerical methods to solve for the boundstate of two 0-branes in three dimensions .
one method is developed by us in this work and we compare it to a method existing in the literature . in spite of considering only three dimensional minkowski space
we obtain interesting results which should give some basic understanding of the behaviour of 0-branes . |
it is compelling that most of the matter in the universe is in a form of non - baryonic cold dark matter .
if it exists , it would play an important role in the structure formation of the universe @xcite .
axions , the pseudo - goldstone bosons , are among the most promising candidates for the non - baryonic cold dark matter .
they arise from the spontaneous breaking of a global @xmath0 symmetry of peccei and quinn ( pq ) , which is introduced to solve the strong cp problem of qcd @xcite . in the standard big - bang cosmology ,
after the spontaneous breakdown of the pq symmetry , the expectation value of the axion field ( i.e. the axionic condensate ) takes some random value on the interval @xmath1 $ ] and , is approximately constant over length scales which are smaller than the horizon size @xcite .
if inflation occurs either after or during the pq symmetry breaking , then the expectation value can be nearly constant throughout the entire universe @xcite . at high temperatures above the @xmath2 scale ,
the axion is massless ; however , at low temperatures , the axion develops a mass due to qcd instanton effects @xcite .
once the axion mass becomes greater than the universe expansion rate , the expectation value of the axion field begins to oscillate coherently around the minimum of its effective potential that is near the origin .
the oscillating axion field then dissipates mainly due to the universe expansion as well as particle production @xcite . in the original papers @xcite , simple estimates of the thermal dissipation of the homogeneous axionic condensate were given .
they considered instabilities arising from the parametric amplification of quantum fluctuations that could pump the energy of the homogeneous axionic condensate into its quantum fluctuations via self couplings , as well as into quantum fluctuating photon modes via a coupling of the axion to electromagnetism due to the color anomaly of the pq symmetry .
this dissipational dynamics via quantum particle production exhibits the feature of unstable bands , and an exponential growth of the quantum fluctuating modes that are characteristics of parametric resonance .
the growth of the modes in the unstable bands translates into profuse particle production .
a given unstable mode will grow as long as it lies within the unstable band .
however , eventually it will be red - shifted out of the band as the universe expands , and then the instabilities of parametric resonance are shut off . in ref .
@xcite , it has been shown that for the pq symmetry breaking scale @xmath3 , because the axion is very weakly coupled , the time it takes to be red - shifted out of the unstable band is too short to build up an appreciable growth of the quantum fluctuating modes .
thus , all of these effects are insignificant .
the condensate is effectively nondissipative and pressureless .
it would survive in the expanding universe , and it behaves like cold dust at the present time . interestingly ,
if @xmath4 , it could constitute a major component of the dark matter of the universe .
recently , the authors of ref .
@xcite were motivated by the recent understanding of the important role of the spinodal instability and parametric resonance that provide the nonlinear and nonperturbative mechanisms in the quantum particle production driven by the large amplitude oscillations of the coherent field @xcite .
they re - examined the issue of the dissipation of the axion field resulting from the production of its quantum fluctuations .
they confirmed that the presence of the parametric resonance would lead to an explosive growth of quantum fluctuations if the universe was minkowskian .
taking account of the expansion of the universe , quantum fluctuations of the axion do not become significant .
this result confirms the conventional wisdom . in this paper
, we will re - examine the damping dynamics of the axion arising from photon production in an expanding universe in the context of the non - equilibrium quantum field theory . the goal of this study is to present a detailed and systematical study of the above - mentioned problem using a fully non - equilibrium formalism @xcite .
we will derive the coupled nonperturbative equation for the axion field and the mode equations for the photon field in a flat robertson - walker spacetime within the nonperturbative hartree approximation that is implemented to consistently take the back reaction effects into account .
we then try to study both numerically and analytically how the nonperturbative effects of spinodal instability and parametric amplification of quantum fluctuations trigger photon production from the oscillations of the axion field . at this stage
, it is worthwhile to mention that our approach can be generalized to any pseudo - scalar field coupled to the photon field in a more general curved spacetime . since the pseudo - scalar nature of the coupling between the axion and the photon , the axion field affects the left- and right - handed circularly polarized photons differently .
this leads to producing the two polarized photons in different amounts .
this polarization asymmetry , if it survives , may have interesting effects on the polarization of the cosmic microwave background . to consider the fermionic plasma effect on photon production ,
one must systematically obtain the non - equilibrium in - medium photon propagators and the off - equilibrium effective vertices between the axion and the photon by integrating out the fermionic field to deal with this problem@xcite . in a plasma , the transverse photons
are dynamically screened @xcite . however , in the literatures @xcite , the arguments stated to include the fermionic plasma effect in support of their conclusions amount to adding by hand the electron plasma frequency into the propagating photon mode equations .
this is problematic when we consider propagating photon modes in the presence of a thermal background .
in fact , the consequence of the abelian ward identities reveals that the transverse photons have the vanishing static magnetic mass in all orders of the perturbation theory @xcite .
this means that the in - medium transverse photon propagators must be nonlocal in nature , and can not be approximated by the local propagator as suggested in ref .
@xcite even in the low energy limit @xcite .
besides , in a fermionic plasma , the effective coupling between the axion and the photon resulting from integrating out the fermionic thermal loop can be modified both at finite temperature as well as out of equilibrium @xcite .
therefore , to fully consider the plasma effect , the non - equilibrium in - medium photon propagators as well as the off - equilibrium effective vertices play the essential roles .
however , incorporating these non - equilibrium effects is a challenging task that lies beyond the scope of this paper , but certainly deserves to be taken up in the near future . in the following
, we will totally ignore the fermionic plasma effect in order to focus on understanding whether the particle production due to spinodal instability as well as parametric amplification is effective or not in the cosmological context . in sec .
[ axi ] , we introduce the axion physics and its coupling to the photon . an effective action of the axion - photon system in the expanding universe is derived .
[ equ ] is devoted to the formalism of the problem in terms of non - equilibrium quantum field theory .
we obtain the equation of motion for the classical axion field and the photon mode equations . in sec .
[ num ] , we present the numerical results .
[ con ] is our conclusions .
the physics of axion and its implication to astrophysics and cosmology can be found in the review articles @xcite .
the axion does not couple directly to the photon .
however , at tree level , the axion field @xmath5 has a coupling with the fermionic field @xmath6 , @xmath7 therefore , the axion can couple to the photon via a fermionic loop . as a consequence of the color anomaly of the pq current , similar to the pion - photon system , the effective lagrangian density for the axion - photon coupling is @xmath8 where @xmath9 . in eq .
( [ l_phia ] ) , the scale @xmath10 , where @xmath11 is the pq symmetry breaking scale , and @xmath12 is the color anomaly of the pq symmetry .
the coupling constant @xmath13 , where @xmath14 is the electromagnetic anomaly of the pq symmetry , and @xmath15 is the fine structure constant .
henceforth , we assume an axion incorporated into the simplest guts with @xmath16 , such that @xmath17 @xcite .
now we write down the effective action for the axion - photon system in the expanding universe , @xmath18 where @xmath19 is added in front of @xmath20 , given by eq .
( [ l_phia ] ) , because @xmath21 is a tensor density of weight @xmath22 @xcite . in the following , for simplicity
, we will assume a flat robertson - walker metric @xmath23 where the signature is @xmath24 , and @xmath25 is the cosmic scale factor .
( [ action ] ) , @xmath26 where the axion potential has a temperature - dependent mass term due to qcd instanton effects , being of the form @xcite @xmath27 where @xmath28 is the zero - temperature axion mass , satisfying @xmath29 . also , we use @xmath30 .
it is well known that the minimal coupling of photons to the metric background is conformally invariant @xcite . as such , in the conformally flat metric ( [ metric ] )
, it is convenient to work with the conformal time , @xmath31 . hence , defining @xmath32 , the action ( [ action ] ) becomes @xmath33 , \label{newact}\end{aligned}\ ] ] where @xmath34 is the minkowski metric . in terms of the conformal time
, the effective action now has analogy with the effective action in minkowski spacetime with the time dependent mass term and interactions .
the non - equilibrium effective lagrangian in the closed time path formalism@xcite is given by @xmath35 - { \cal l}\left[\chi^-,a_\mu^-\right ] , \label{lneq}\ ] ] where + ( - ) denotes the forward ( backward ) time branches .
we then decompose @xmath36 into the axionic mean field and the associated quantum fluctuating fields : @xmath37 with the tadpole conditions , @xmath38 we will implement the tadpole conditions to all orders in the corresponding expansion to obtain the non - equilibrium equations of motion . to take account of the back reaction effects on the dynamics of the axion field from quantum fluctuating photon modes , we adopt the following hartree factorization which is implemented for both @xmath39 components@xcite : @xmath40
as seen later , the expectation value can be determined self - consistently .
it must be noted that there is no _ a priori _ justification for such a factorization .
however , this approximation provides a nonperturbative framework that allows us to treat photon fluctuations self - consistently @xcite . on the contrary
, we will ignore the quantum fluctuations of the axion , which can be produced via self - couplings , as the study of ref .
@xcite has shown that these effects are insignificant . with eq .
( [ mean ] ) , we first expand the non - equilibrium lagrangian density ( [ lneq ] ) in powers of @xmath6 and keep the term up to linear @xmath6 to ignore its quantum fluctuation effects . together with eq .
( [ har ] ) , the hartree - factorized lagrangian then becomes @xmath41 - { \cal l}\left[\varphi(\eta)+\psi^-,a_\mu^-\right ] & & = \left\ { - u ( \eta ) \psi^+
-{1\over4}f^{+}_{\mu\nu}{\tilde{f}}^{+\mu\nu } + \frac{c}{a f_a } \varphi(\eta ) f^+_{\mu\nu } { \tilde{f}}^{+\mu\nu } \right .
\nonumber \\ & & \left .
+ \frac{c}{a f_a } \psi^+ \langle f^{+\mu\nu}{\tilde{f}}^+_{\mu\nu } \rangle \right\ } -\left\ { + \rightarrow -\right\ } , \end{aligned}\ ] ] where @xmath42\ , .
\end{aligned}\ ] ] the dot means the time derivative with respect to the conformal time . with the tadpole conditions ( [ tad ] ) , we obtain the following equation of motion for the axionic mean field given by @xmath43 where we define the dimensionless field , @xmath44 . within the hartree approximation , the photon production processes do not involve photons in the intermediate states @xcite . to avoid the gauge ambiguities , we will work in the coulomb gauge and concentrate only on physical transverse gauge field , @xmath45 @xcite .
then , the heisenberg field equation for @xmath45 can be read off from the quadratic part of the lagrangian in the form @xmath46 it is more convenient to decompose the field @xmath47 into the fourier mode functions @xmath48 in terms of circularly polarized states , @xmath49 e^{i\vec k\cdot \vec x } + { \rm h.c.}\right\},\end{aligned}\ ] ] where @xmath50 are destruction operators , and @xmath51 are circular polarization unit vectors defined in ref .
then the mode equations are @xmath52 with the expectation values given by @xcite @xmath53 \frac{d}{d\eta } \left ( |v_{1 k}(\eta)|^2- |v_{2 k}(\eta)|^2 \right ) , \label{qf}\ ] ] where we have assumed that at initial time @xmath54 , the photons are in local equilibrium at the initial temperature @xmath55 .
clearly , the photon mode equations ( [ meq ] ) are decoupled in terms of the circular polarization mode functions .
the axion field acts as the time dependent mass term that s triggers photon production .
the effective mass terms have opposite signs for the two polarizations due to the pseudo - scalar nature of the axion - photon coupling .
this will lead to producing the different polarized photons in different amounts , resulting in a polarization asymmetry in photon emission .
the expectation value of the number operator for the asymptotic photons with momentum @xmath56 is given by@xcite @xmath57 - 1 \nonumber \\ & = & { 1\over 4k^2 } \coth \left [ \frac{k}{2 t_i } \right ] \left[|\dot v_{1 k}(\eta)|^2+k^2 & & { 1\over 4k^2 } \coth \left [ \frac{k}{2 t_i } \right ] \left[|\dot v_{2 k}(\eta)|^2+k^2 & = & n_{+}(k,\eta ) + n_{-}(k,\eta ) , \label{numop}\end{aligned}\ ] ] which is the number of photons with momentum @xmath56 per unit comoving volume .
in this section , we will compute the photon production from the axionic condensate with @xmath58 , which has a zero - temperature mass @xmath59 . as we will show below
, the photon production takes place mainly during the radiation - dominated epoch .
so , we simply assume a radiation - dominated universe . at temperature @xmath60 ,
the hubble parameter is @xmath61 where @xmath62 is the number of effective degrees of freedom at temperature @xmath60 , @xmath63 is the planck scale , and the cosmic scale factor is @xmath64 where @xmath65 is the time when the axion field starts to oscillate , defined by a temperature @xmath66 such that @xmath67 . as such , @xmath68 . for @xmath58 , from eqs .
( [ mat ] ) and ( [ hubble ] ) we find @xmath69 and @xmath70 . changing the variable @xmath71 into @xmath72 ,
the equations of motion ( [ eom ] ) and ( [ meq ] ) become @xmath73 where @xmath74 and @xmath75^{7.4\over 4 } & t > > \lambda_{qcd } , \\
10 ^ 2 \left(\frac{t_1}{\lambda_{qcd}}\right)^{7.4 } & t < < \lambda_{qcd}. \end{array } \right . \label{mass}\ ] ] note that @xmath62 does not change significantly from @xmath66 to @xmath76 . henceforth , we approximate @xmath77 , where @xmath78 .
it is worth to point out that the mode equations ( [ meq2 ] ) have unstable modes via the spinodal instability for sufficiently low - momentum modes with @xmath79 , where the effective mass becomes negative . to solve eqs .
( [ eom2 ] ) and ( [ meq2 ] ) , we have to specify the initial conditions for the axion and photon fields .
the amplitude of the axion field is frozen for @xmath80 , i.e. , @xmath81 for the photon mode functions , we propose @xmath82 these initial conditions are physically plausible and simple enough for us to investigate a quantitative description of the dynamics . to evaluate the @xmath83 in eq .
( [ qf ] ) and the photon number operator in eq .
( [ numop ] ) , we approximate the bose enhancement factor by @xmath84 = \coth \left [ \frac{\xi}{2\gamma } \right ] \simeq \frac{2\gamma}{\xi},\ ] ] where @xmath85 , and we are interested in @xmath86 . in fig .
[ fig1 ] , we plot the temporal evolution of the axion field and its time derivative by choosing @xmath87 that corresponds to @xmath88 . due to the expansion of the universe , the field amplitude decreases with time .
however , the rate of change of the amplitude increases with time . to understand this , we redefine @xmath89 in eq .
( [ eom2 ] ) .
the @xmath83 term can be neglected , being extremely small as shown in fig .
[ fig2 ] , where we have evaluated the last term of eq .
( [ eom2 ] ) denoted by @xmath90 .
then , the equation of motion for @xmath91 when @xmath92 is given by @xmath93 from eq .
( [ mass ] ) , the solutions for @xmath91 are bessel functions . for @xmath94 , asymptotically @xmath95 . for @xmath96 , @xmath97 .
the latter shows that the axions behave like non - relativistic matter . for the photon production , we calculate the spectral photon number density @xmath98 , equal to @xmath99 in eq .
( [ numop ] ) .
it is convenient to define a ratio , @xmath100 which is the excess photons above the thermal background .
three snapshots of the ratio at @xmath101 and @xmath102 are shown in fig .
it is interesting to see that the production duration of each non - zero mode is short , and higher - momentum modes are produced at later times .
this in fact demonstrates a brief exponential growth of the unstable mode due to the parametric resonance instability , which is shut - off when the unstable mode has been red - shifted out of the unstable band . as a consequence ,
the photon production is limited with an excess photon ratio typically at a level of @xmath103 . as we have mentioned above
, the spinodal instability happens for low - momentum photon modes .
we demonstrate this numerically in fig .
[ fig4 ] , where we have chosen @xmath104 that lie within the spinoidal region where @xmath79 .
the production ratio is also at a level of @xmath103 , but it is apparent that these low - momentum modes are produced during the first oscillating cycle of the axion field . the ratio in fig . [ fig3 ] can be actually estimated as follows .
first , let us find out the approximate form for @xmath105 from eq .
( [ eom3 ] ) and fig .
[ fig1 ] , which is given by @xmath106 where @xmath107 is the scale factor when @xmath108 . at instant @xmath72 , @xmath105 is oscillating with an effective frequency @xmath109 . inserting the approximate form ( [ dtheta ] ) with @xmath110 treated as a constant into the mode equation of @xmath111 in eq .
( [ meq2 ] ) , and changing variable to @xmath112 , we have @xmath113 this is the standard mathieu equations @xcite .
the widest and most important instability is the first parametric resonance that occurs at @xmath114 with a narrow bandwidth @xmath115 .
but actually @xmath110 is changing with time .
as such , the unstable mode will grow exponentially only during a brief period roughly given by @xmath116 consequently , this instability leads to the growth of the occupation numbers of the created photons by a growth factor , @xmath117 where the growth index @xmath118 .
hence the mode number density is increased by an amount approximately given by @xmath119 where @xmath72 is the scale factor when the mode @xmath111 enters into the resonance band .
this estimation is of the same order of magnitude as found in the numerical results .
interestingly , the @xmath120 dependence of the mode production can also be seen in fig .
[ fig3 ] . similar estimation can also be done for the mode function @xmath121 .
the polarization asymmetry in the produced photons is defined as @xmath122 we have input the mode solutions for @xmath111 and @xmath121 to eq .
( [ numop ] ) to calculate @xmath123 ( [ asymm ] ) . a plot of the asymmetry versus the momentum at @xmath124 is shown in fig .
the numerical result shows that the asymmetry is fluctuating about zero as the photon momentum varies .
the fluctuating amplitude is about @xmath125 of the thermal background . although the asymmetry averaged out over a wide range of momenta
is nearly zero , at certain momenta the produced photons are about @xmath126 circularly polarized .
however , subsequently this polarization asymmetry will be damped out by photon - electron scatterings in the plasma .
first we basically confirm that the photon production via parametric resonance is invalidated by the expansion of the universe . besides
, we find that in addition to parametric resonance , for long - wavelength photon modes , a new dissipative channel via spinodal instability is open .
this open channel results in the long - wavelength fluctuations of the photon modes . again
, this production is suppressed in the expanding universe .
we also observe the polarization asymmetry in the produced circularly polarized photons as a result of the pseudo - scalar nature of the coupling .
however , it will be damped out effectively by the plasma .
but it is very interesting to see whether it is possible to generate a circular polarization asymmetry in the production of photons from certain pseudo - scalar fields such that it may leave an imprint on the polarization of the cosmic microwave background . as to the plasma damping on photon production ,
we have pointed out the problem in the naive approximation that simply introduces the electron plasma frequency to the photon modes .
we have thus proposed a dynamical and non - equilibrium treatment which should be a better approach to consider the plasma effects .
the actual calculations are rather difficult , but they certainly deserve further studies .
we would like to thank d. boyanovsky and h. j. de vega for their useful discussions . the work of d.s.l .
( k.w.n . ) was supported in part by the national science council , roc under the grant nsc89 - 2112-m-259 - 008-y ( nsc89 - 2112-m-001 - 001 ) .
d. boyanovsky and h. j. de vega , phys .
d * 47 * , 2343 ( 1993 ) ; d. boyanovsky , d .- s .
lee , and a. singh , phys .
d * 48 * , 800 ( 1993 ) ; d. boyanovsky , h. j. de vega and r. holman , phys .
d * 51 * , 734 ( 1995 ) ; d. boyanovsky , h. j. de vega and r. holman , phys .
d * 49 * , 2769 ( 1994 ) ; d. boyanovsky , h. j. de vega , r. holman , d .- s .
lee , and a. singh , phys .
d * 51 * , 4419 ( 1995 ) ; d. boyanovsky , m. dattanasio , h. j. de vega , r. holman , and d .- s .
lee , phys . rev .
d * 52 * , 6805 ( 1995 ) .
d. boyanovsky , h. j. de vega , d .- s .
lee , y. j. ng and s .- y .
wang , phys .
d * 59 * , 125009 ( 1999 ) ; s .- y .
wang , d. boyanovsky , h. j. de vega , d .- s .
lee and y. j. ng , `` damping rates and mean free paths of soft fermion collective excitations in a hot fermion gauge scalar theory '' , hep - ph/9902218 .
d. boyanovsky , h. j. de vega , r. holman , s. prem kumar and r. d. pisarski , phys . rev .
d * 58 * , 125009 ( 1998 ) ; s. prem kumar , d. boyanovsky , h. j. de vega and r. holman , `` anomalous pseudosalar vertex in and out of equilibrium '' , hep - ph/9905374 , and references therein .
three snapshots of the spectral number density ratio @xmath129 , defined in eq .
( [ ratio ] ) , of the photons produced via parametric amplification at @xmath101 and @xmath102 . the dimensionless quantity @xmath74 , where @xmath130 is the photon momentum and @xmath65 is the conformal time when the axion field starts to oscillate . | using the non - equilibrium quantum field theory , photon production from the coherently oscillating axion field in a flat robertson - walker cosmology is re - examined . first neglecting the debye screening of the baryon plasma to photons , we find that the axions will dissipate into photons via spinodal instability in addition to parametric resonance . as a result of the pseudo - scalar nature of the axion - photon coupling
, we observe a circular polarization asymmetry in the produced photons .
however , these effects are suppressed to an insignificant level in the expanding universe .
we then briefly discuss a systematic way of including the plasma effect which can further suppress the photon production .
we note that the formalism of the problem can be applied to any pseudo - scalar field coupled to photon in a thermal background in a general curved spacetime . |
the monolayer systems of group - vi dichalcogenides , @xmath2 ( @xmath3=mo , w ; @xmath4=s , se , te ) @xcite have received considerable interest as unique alternatives of graphene for their various intriguing properties @xcite : direct band gap of 2 ev @xcite , high photoluminescence yield @xcite , high on / off switching ratio in field effect transistors @xcite , and electric field - induced superconductivity @xcite .
most notably , their staggered - honeycomb - like lattice structure ( the structure and the unit cell are shown in fig . [
fig : mos2-struct ] ( a)(c ) ) hosts a model system for valley - dependent phenomena @xcite originally proposed for graphene @xcite .
the bottom of the conduction band and the top of the valence band are located at the k points of the hexagonal brillouin zone in @xmath2 .
since the k and @xmath5@xmath6k points are not equivalent , the electronic states in the opposite pockets can carry an additional quantum number , the valley index .
furthermore , the spin - orbit coupling results in sizable valley - dependent spin splitting at the valence top ( vt ) . on the basis of the strong coupling between the spin , orbital and valley degrees of freedom , the control of the carrier population of each valley by orbital / spin - sensitive probes has been proposed .
indeed , valley - selective excitation of the electrons and excitons by circularly polarized light has been demonstrated @xcite . further understanding and stable control of the valley - dependent electronic properties
could bring us to the _ valleytronics_a way of encoding information into the valley degree of freedom , which can be much faster and more efficient than conventional optoelectronics @xcite .
the above progress led to renewed attention to the rich polymorphism of multilayered @xmath2 @xcite in view of the valley physics . in inversion - symmetric bilayers with the 2h stacking [ fig .
[ fig : mos2-struct ] ( d ) ] , the net valley - dependent spin polarization is absent .
this property has been proposed to be utilizable for switching of the polarization with a symmetry - breaking electric field @xcite or mechanical bending @xcite .
it has also been stated that a layer degree of freedom ( upper / lower layer ) couples to the valley degrees of freedom @xcite and this coupling should cause magnetoelectric effects @xcite . on the other hand , very recently , suzuki _
et al . _
@xcite observed valley - dependent spin polarization in multilayered mos@xmath1 by utilizing the noncentrosymmetric 3r stacking [ fig .
[ fig : mos2-struct ] ( e ) ] .
this success paves a very different way to the control of the valley carriers : valleytronics by engineering the stacking geometry .
however , knowledge of the valley electronic states in the 3r systems is still scarce , which hampers further advances in the field and applications of multilayered mos@xmath1 . ;
( b)(c ) side and top views of the monolayer , where the trigonal prisms are depicted as shaded triangles .
@xmath7 , @xmath8 and @xmath9 represent inequivalent three - fold rotational axes and @xmath10 and @xmath11 are primitive lattice vectors .
( d)(e ) side view ( top ) and top view ( bottom ) of the 2h and 3r stackings , respectively .
rotational axes for each layer are also indicated . ] in this article , we study the valley electronic states in mos@xmath1 with the 3r stacking and compare them with those in 2h - mos@xmath1 . combining _
ab initio _ band structure calculations and group - theoretical analysis , we show that the interlayer hopping amplitude of the valley states is exactly zero in 3r - mos@xmath1 , i.e. the electrons are confined within the two - dimensional ( 2d ) layers .
furthermore , we study how this confinement affects the exciton spectrum with an anisotropic hydrogen atom model . finally , we compare the theoretical results to the reflectivity spectra measured on both 3r - mos@xmath12 and 2h - mos@xmath12 compounds .
the revealed mechanism of the single - layer confinement of the valley electrons respects only the crystal and orbital symmetries and therefore is commonly applicable to the family of 3r-@xmath2 , which should facilitate the dimensionality control of the valley states in the transition - metal dichalcogenides . with the 2h and 3r stacking ,
respectively , where we employed special points of the conventional hexagonal brillouin zone ( bz ) defined by @xmath13@xmath14@xmath15 , @xmath16@xmath14@xmath17 , @xmath18@xmath14@xmath19 and @xmath20@xmath14@xmath21 with @xmath22@xmath14@xmath23@xmath14@xmath24 and @xmath25 being the interlayer distance . note that the the present bz is twice as large as the primitive bz of 2h - mos@xmath1 .
solid ( dashed ) lines denote the result with ( without ) spin - orbit coupling . ]
first , we calculated the band structures of bulk 2h- and 3r - mos@xmath12 using the wien2k code employing the full - potential linearized augmented plane - wave method @xcite .
we used the perdew - burke - ernzerhof exchange - correlation functional @xcite and included the scalar - relativistic effects @xcite and spin - orbit coupling @xcite as implemented in the wien2k code .
experimental lattice parameters and atomic configurations were taken from refs . and .
the muffin - tin radii for mo and s atoms , @xmath26 and @xmath27 , were set to 2.44 and 2.10a.u .
, respectively .
the maximum modulus for the reciprocal lattice vectors @xmath28 was chosen so that @xmath27@xmath29@xmath28 = 7.00 .
the calculated band structures are shown in fig .
[ fig : bulk - band ] .
the apparent difference seen around the @xmath13 point in valence bands is due to brillouin - zone folding for the 2h case with respect to @xmath30@xmath14@xmath31 plane and hence trivial .
notably , along the k
h path , the band at the conduction bottom ( cb ) is flat for the both polytypes . on the other hand ,
the vt bands for the 2h polytype shows small but nonzero dispersion , whereas those for 3r is flat .
this feature has been first found in ref .
but its origin has remained unexplored .
below , we clarify its mechanism analytically . in order to understand the microscopic origin of the flat bands along the k
h path , let us first look into the symmetries of the electronic states in a monolayer of type @xmath32 in fig .
[ fig : mos2-struct ] .
the little group of k point includes three fold rotation ( @xmath33 ) . under the rotation around the axis passing through the center of the honeycomb unit [ @xmath7 in fig .
[ fig : mos2-struct](c ) ] , the vt ( cb ) state , which is formed by mo 4@xmath34@xmath35@xmath36 ( 4@xmath37 ) orbital having orbital angular momentum @xmath38@xmath14@xmath39 ( @xmath40 ) , gets phase 1 ( @xmath41 ) @xcite . ] is represented , where circle , triangle and square represent the three phase values @xmath42 , @xmath43 and @xmath41 , respectively .
large shaded triangles correspond to those in fig [ fig : mos2-struct ] .
phase configurations before ( after ) the corresponding permutation @xmath44 are represented by solid ( shaded ) symbols .
crosses indicate the rotational centers @xmath45 , @xmath8 and @xmath9 , whereas arrows represent how the sites permute with @xmath44 . ]
[ tab : phases ] = 1 mm .phases @xmath46 of the eigenvalues @xmath47 of @xmath48 around different rotational axes ( @xmath49@xmath7 , @xmath8 and @xmath9 ) for the k - point states ( @xmath50=vt , cb ) . [ cols="<,^,^,^,^,^,^,^,^,^,^,^,^ " , ] as a matter of fact , there are two more rotational axes inequivalent to @xmath7 : @xmath8 and @xmath9 passing through s and mo sites , respectively [ fig .
[ fig : mos2-struct](c ) ] . in order to calculate the phase factors for these @xmath33 axes ,
let us decompose the @xmath33 rotation into two successive operations : permutation of the atomic sites @xmath51 and rotation of the local coordinates at each lattice point @xmath52 .
the @xmath33 around @xmath53 axis is defined as @xmath48 ( @xmath54 ) , where @xmath44 denotes the corresponding site permutation .
the contribution to the total phase from @xmath52 is simply @xmath55 with @xmath56=@xmath57 .
the contribution from @xmath44 can be calculated by examining how the bloch wave function transforms under the site permutation . in fig .
[ fig : rotate - scheme ] , the phase of the k - point bloch state is depicted , where three values ( @xmath42 , @xmath43 and @xmath41 ) are represented by circle , triangle and square , respectively . by operating @xmath58 ,
e.g. , the configuration depicted with solid symbols is transformed to that with shaded symbols [ panel ( a ) ] . comparing these symbols on each site , the phase contribution @xmath41 for @xmath58 is deduced . applying similar arguments to @xmath53=@xmath8 and @xmath9 , and to the monolayer of type @xmath59 , we get the results summarized in table [ tab : phases ] .
note that the phases for @xmath59 are inverse of those for @xmath32 because their valley states are related by spatial inversion and time reversal . on the basis of the axis - dependent phases derived above @xcite ,
let us next consider the interlayer hopping amplitude defined by @xmath60 for the bulk material . here
, @xmath61 denotes the k - point bloch functions at @xmath62th layer , which are eigenstates in the limit where the layer is isolated from others .
@xmath63 is the total hamiltonian of the bulk crystal .
suppose @xmath64 denotes a certain three - fold rotation under which @xmath63 is invariant , then @xmath65 \langle \psi^{b}_{{\bf k},1 } | \mathcal{h}| \psi^{b}_{{\bf k},2 } \rangle \label{eq : hopping } .\end{aligned}\ ] ] here , @xmath66@xmath14@xmath67 and @xmath68 denotes the phase summarized in table [ tab : phases ] for the @xmath62th layer .
note that @xmath53 and @xmath69 in the second and last lines depend on the stacking pattern and rotational axis of @xmath64 . in the 2h
stacking , without loss of generality , we can assume the first and second layers to be @xmath32 and @xmath59 , respectively . in this case , the possible combinations @xmath70 are @xmath71 , @xmath72 and @xmath73 [ fig .
[ fig : mos2-struct ] ( d ) ] , all of which yield @xmath66@xmath14@xmath74@xmath14@xmath75 and @xmath76@xmath14@xmath77 , respectively . in the 3r
stacking , where only layers of @xmath32 type are stacked , @xmath70@xmath14@xmath78 , @xmath79 , @xmath80 [ fig .
[ fig : mos2-struct ] ( e ) ] , and therefore @xmath66@xmath14@xmath81 . according to eq .
( [ eq : hopping ] ) , the nonzero @xmath66 forces @xmath82 to be zero , which consistently explains the dispersionless bands along the k h direction in the 3r polytype .
the difference in the band structures of 2h- and 3r - mos@xmath12 should cause different optical properties of their valley excitations . to verify this experimentally , we measured the normal incidence reflectivity spectra on the cleaved ( 001 ) surfaces of both polytypes at @xmath83=20k ( see fig .
[ fig : reflectivity ] ) .
the details of the crystal growth were reported in ref . .
since the reflectivity were measured in the limited photon energy range of @xmath84=1.53ev , the spectra were extended up to 30ev using the data published for 2h - mos@xmath12 @xcite to facilitate proper kramers - kronig transformation .
measured at @xmath83=20k .
the imaginary part of the dielectric function was obtained from the reflectivity spectra by kramers - kronig transformation . for the @xmath32 exciton branch , @xmath85 and
@xmath86 excitons are labeled as @xmath87 and @xmath88 , while the fine structure of the @xmath89 exciton can not be resolved . ] the imaginary part of the dielectric function calculated from the reflectivity is shown in fig .
[ fig : reflectivity ] .
although the two main features in the @xmath90 spectra , the @xmath32 and @xmath89 exciton peaks corresponding to the optical excitations from the spin - orbit split vt to the cb , look similar , there are several important differences between the two spectra .
first , the energy separation between the @xmath87 and the @xmath89 peaks is larger for the 2h compound ( 197mev ) compared with the 3r polytype ( 148mev ) , which agrees with the theoretical prediction for the splitting at the vt ( see fig .
[ fig : bulk - band ] and ref . ) . beside the main excitonic resonances ,
smaller peaks were observed on the high energy shoulder of the @xmath32 exciton , labeled as @xmath88 . in a hydrogen - atom - like model of excitons ,
@xmath87 and @xmath88 are assigned to the 1@xmath91 and 2@xmath91 bound states of the electron - hole pair . in the 3r case ,
@xmath87 and @xmath88 peaks are more separated and their intensity ratio is larger than in the 2h structure . [
tab : parameters - in - the - model ] to understand these differences , we employed an anisotropic hydrogen atom model @xcite : @xmath92 \varphi({\bf r } ) \nonumber \\ & & \hspace{150pt } = e\varphi({\bf r } ) , \label{eq : exciton - model}\end{aligned}\ ] ] where @xmath93 and @xmath94 are the components of the dielectric constant tensor perpendicular and parallel to the @xmath95-direction respectively , and @xmath96 and @xmath97 are those of the reduced effective - mass tensor of excitons at the k - point .
note that in a monolayer system , the screened electron - hole interaction is rather different from that in eq .
( [ eq : exciton - model ] ) owing to the surrounding vacuum @xcite .
bulk environment allows us to use the present model as in early studies @xcite .
the values of @xmath93 , @xmath94 , @xmath96 and @xmath97 were extracted from _ ab initio _ calculations using the wien2k code .
dielectric constants @xmath98 were calculated using the random - phase approximation @xcite , @xmath99 where @xmath100 denotes the volume of the simulation cell , @xmath101 and @xmath53 are indices of the occupied and unoccupied bands , respectively , @xmath102 is the wave vector , @xmath103 denotes the unit vectors , @xmath104 denotes the spin index , @xmath105 denotes the kohn - sham wave functions , and @xmath106 denotes their orbital energies .
the effective masses @xmath107 were estimated from a parabolic curve fitting of the band dispersions of electrons and holes around the k point within the range of @xmath108 ( @xmath109 for @xmath97 of the 3r polytype ) . to obtain the effective masses accurately , @xmath110 , @xmath27 , and @xmath27@xmath29@xmath28 were set to 2.20a.u .
, 1.89a.u . , and 9.00 , respectively .
calculated dielectric constants and reduced effective masses are presented in table [ tab : parameters - in - the - model ] , which agree well with previous calculations @xcite . since both vt and cb are flat along the k
h line , @xmath97 is quite large in the 3r stacking , whereas it is small in the 2h stacking in which only cb is flat .
note that the effective masses were determined in the close vicinity of the k point , and the valence and conduction bands in 3r - mos@xmath12 have less than 2mev dispersion along the k
h line . since this range is much smaller than the binding energy of @xmath85 or @xmath86 exciton , in the following analyses , we regard the 3r bands completely flat , i.e. @xmath111@xmath14@xmath112 . [
tab : exciton - spectr - calc ] within the model of eq .
( [ eq : exciton - model ] ) , the binding energies and oscillator strengths of the @xmath85 and @xmath86 excitons were estimated .
for the 2h case , we referred to a previous report presenting a numerical solution of the model for a range of parameters @xcite . for the 3r case
, we instead took the exact solution of the 2d hydrogen atom according to the above consideration .
the ratio of the oscillator strengths was derived with the fermi golden rule @xcite using the present energy eigenvalues and the reduced effective masses .
the obtained values are presented in table [ tab : exciton - spectr - calc ] together with our experimental ones extracted from fits with lorentzian peaks ( fig .
[ fig : reflectivity ] ) .
the substantial differences in the energy levels and in the oscillator strengths between the exciton spectra of the 2h and 3r structures were captured by the model calculation .
since all the parameters in the 2h and 3r cases are similar to each other except @xmath97 , we attribute the differences solely to the 2d confinement of the electrons and holes in the 3r compound . to see another aspect of the 2h and 3r excitons with respect to their dimensionality , we examined the exciton radii by using a simple trial wave function @xcite for the @xmath85 eigenstate of eq .
( [ eq : exciton - model ] ) , @xmath113 , \ ] ] with the parameters @xmath114 and @xmath115 , which represent exciton in - plane and out - of - plane radii , respectively , optimized in a variational manner . for the 3r exciton
, we used a finite @xmath111 listed in table [ tab : parameters - in - the - model ] to estimate an upper bound of @xmath115 .
the obtained values of @xmath114 are almost the same values for the 2h and 3r structures : 19 and 15 , respectively . in contrast , the values of @xmath115 are 8.9 and 2.7 for the 2h and 3r structures , respectively . comparing these values with the interlayer distance ( 6.1 for both structures ) , we can see that the 3r exciton is tightly confined in a single layer while the 2h exciton extends to the neighboring layers .
the present results establish the 2d nature of the valley exciton in bulk mos@xmath1 with the 3r stacking . note that an apparently similar confinement effect has also been discussed for the 2h stacking as the spin - layer locking effect @xcite . since the magnitude of confinement is governed by the factor @xmath116 with @xmath117 being the spin - orbit coupling coefficient @xcite , its strength essentially relies on atomic number of @xmath3 .
distinctively , the layer - confinement effect in the 3r stacking is a quantum interference effect caused by the symmetry of the crystal and therefore is guaranteed to be relevant in general @xmath2 systems with the same structure .
the complete confinement of the 3r valley states should also provide us with an arena of exploring novel interplay of low dimensionality and valley - physics .
compared to the 2h or monolayer mos@xmath1 , the 3r polytype has a significant merit that valley polarized carriers are more robust in the latter form .
the carriers are protected from relaxation processes resulting from interlayer transitions , contrary to the 2h systems . in the previous work @xcite , we have found an indication that interlayer transitions are much suppressed and the valley excitons have much longer lifetime in 3r - mos@xmath1 .
now we have clarified that the interlayer hopping amplitude is _ exactly _ zero and thus the protection is robust ; also , it is common to all 3r-@xmath2 . note that the 2d states in the 3r systems are different from those in the monolayer systems . in the former , the thickness of the sample allows us to generate valley polarized carriers away from the substrate , which will reduce impacts of the substrate potential .
hence , for example , we can use the 3r crystals as a medium ( or lead wires , e.g. ) for the valley device , through which the valley - polarized carriers are robustly transmitted .
the unique property of the 3r stacking could thus bring about more effective valley - information transmission and generation toward dissipation - less low - energy optoelectronic devices .
to summarize , we showed rigorous absence of the interlayer hopping for the valley electrons in 3r - mos@xmath1 , which was substantiated with a group theoretical argument . through the measurement of the reflectivity spectra and analysis with an anisotropic hydrogen atom model
, the absence of the hopping has been shown to make the exciton absorption spectrum 2d hydrogen - like and the exciton wave function confined in a single layer .
our results exemplify the control of spatial dimensionality of the valley excitations in 2h- and 3r-@xmath2 , which may help to develop new valleytronics devices .
this work was supported by the strategic international collaborative research program ( sicorp - lemsuper ) , japan science and technology agency , funding program for world - leading innovative r & d on science and technology ( first program ) and grant - in - aid for scientific for specially promoted research ( no .
25000003 ) from jsps .
r. s. was supported by leading graduate program ( merit ) from mext .
s. b. was supported by hungarian research funds otka pd111756 and k108918 .
the imaginary part of the dielectric function deduced from the measured reflectivity is fitted by a sum of lorentzian peaks .
the fits shown in fig . [ fig : suppfit ] describe the experimental spectra well . the contributions from the higher energy transitions are also included .
999 k. s. novoselov , d. jiang , f. schedin , t. j. booth , v. v. khotkevich , s. v. morozov , and a. k. geim , two - dimensional atomic crystals , proc .
u. s. a. * 102 * , 10451 ( 2005 ) q. h. wang , k. k. zadeh , a. kis , j. n. coleman , and m. s. strano , electronics and optoelectronics of two - dimensional transition metal dichalcogenides , nat .
* 7 * , 699 ( 2012 ) .
k. f. mak , c. lee , j. hone , j. shan , and t. f. heinz , atomically thin mos@xmath1 : a new direct - gap semiconductor , phys .
rev . lett . * 105 * , 136805 ( 2010 ) a. splendiani , l. sun , y. zhang , t. li , j. kim , c. y. chim , g. galli , and f. wang , emerging photoluminescence in monolayer mos@xmath1 , nano lett . * 10 * , 1271 ( 2011 ) .
t. li and g. galli , electronic properties of mos@xmath1 nanoparticles , j. phys .
c * 111 * , 16192 ( 2007 ) .
s. lebgue and o. eriksson , electronic structure of two - dimensional crystals from _ ab initio _ theory , phys .
b * 79 * , 115409 ( 2009 ) .
y. ding , y. wang , j. ni , l. shi , s. shi , w. tang , first principles study of structural , vibrational and electronic properties of graphene - like mx2 ( m = mo , nb , w , ta ; x = s , se , te ) monolayers , physica b * 406 * , 2254 ( 2011 ) .
a. kuc , n. zibouche , and t. heine , influence of quantum confinement on the electronic structure of the transition metal sulfide @xmath83s@xmath1 , phys .
b * 83 * , 245213 ( 2011 ) .
l. liu , s. b. kumar , y. ouyang , and j. guo , performance limits of monolayer transition metal dichalcogenide transistors , ieee trans .
electron devices * 58 * , 3042 ( 2011 ) .
z. y. zhu , y. c. cheng , and u. schwingenschlgl , giant spin - orbit - induced spin splitting in two - dimensional transition - metal dichalcogenide semiconductors , phys .
b * 84 * , 153402 ( 2011 ) . e. s. kadantsev and p. hawrylak , electronic structure of a single mos@xmath1 monolayer , solid state commun . *
152 * , 909 ( 2012 ) c. ataca , h. ahin , and s. ciraci , stable , single - layer mx@xmath1 transition - metal oxides and dichalcogenides in a honeycomb - like structure , j. phys . chem . * 116 * , 8983 ( 2012 ) .
b. radisavlijevic , a. radenovic , j. brivio , v. guacometti , and a. kis , single - layer mos@xmath1 transistors , nat . nanotechnol .
* 6 * , 147 ( 2011 ) .
y. yoon , k. ganapathi , and s. salahuddin , how good can monolayer mos@xmath1 transistors be ? nano lett . *
11 * , 3768 ( 2011 ) .
k. taniguchi , a. matsumoto , h. shimotani , and h. takagi , electric - field - induced superconductivity at 9.4 k in a layered transition metal disulphide mos@xmath1 , appl .
* 101 * , 042603 ( 2012 ) .
j. t. ye , y. j. zhang , r. akashi , m. s. bahramy , r. arita , y. iwasa , superconducting dome in a gate - tuned band insulator , science * 338 * , 1193 ( 2012 ) .
d. xiao , g. b. liu , w. feng , x. xu , and wang yao , coupled spin and valley physics in monolayers of mos@xmath1 and other group - vi dichalcogenides , phys .
108 * , 196802 ( 2012 ) .
a. rycerz , j. tworzydlo , and w. j. beenakker , valley filter and valley valve in graphene , nat . phys . * 3 * , 172 ( 2007 ) .
d. xiao , w. yao , and q. niu , valley - contrasting physics in graphene : magnetic moment and topological transport , phys .
* 99 * , 236809 ( 2007 ) . w. yao , d. xiao , and q. niu , valley - dependent optoelectronics from inversion symmetry breaking , phys .
b * 77 * , 235406 ( 2008 ) .
h. zeng , j. dai , w. yao , d. xiao , and x. cui , valley polarization in mos@xmath1 monolayers by optical pumping , nat .
* 7 * , 490 ( 2012 ) .
k. f. mak , k. he , j. shan , and t. f. heinz , control of valley polarization in monolayer mos@xmath1 by optical helicity , nat . nanotechnol . * 7 * , 494 ( 2012 )
. t. cao , g. wang , w. han , h. ye , c. zhu , j. shi , q. niu , p. tan , e. wang , b. liu , and j. feng , valley - selective circular dichroism of monolayer molybdenum disulphide , nat .
commun . * 3 * , 887 ( 2012 ) .
a. m. jones , h. yu , n. j. ghimire , s. wu , g. aivazian , j. s. ross , b. zhao , j. yan , d. g. mandrus , d. xiao , w. yao , and x. xu , optical generation of excitonic valley coherence in monolayer wse@xmath1 , nat .
* 8 * , 634 ( 2013 ) .
k. f. mak , k. l. mcgill , j. park , and p. l. mceuen , the valley hall effect in mos@xmath1 transistors , science * 344 * , 1489 ( 2014 ) .
x. xu , w. yao , d. xiao , and t. f. heinz , spin and pseudospins in layered transition metal dichalcogenides , nat .
* 10 * , 343 ( 2014 ) .
j. a. wilson and a. d. yoffe , the transition metal dichalcogenides discussion and interpretation of the observed optical , electrical and structural properties , adv . phys . * 18 * , 193 ( 1969 ) .
s. wu , j. s. ross , g .- b .
liu , g. aivazian , a. jones , z. fei , w. zhu , d. xiao , w. yao , d. cobden , and x. xu , electrical tuning of valley magnetic moment through symmetry control in bilayer mos@xmath1 , nat .
* 9 * , 149 ( 2012 ) .
h. yuan , m. s. bahramy , k. morimoto , s. wu , k. nomura , b .- j .
yang , h. shimotani , r. suzuki , m. toh , c. kloc , x. xu , r. arita , n. nagaosa , and y. iwasa , zeeman - type spin splitting controlled by an electric field , nat . phys .
* 9 * , 563 ( 2013 ) .
p. koskinen , i. fampiou , and a. ramasubramaniam , density - functional tight - binding simulations of curvature - controlled layer decoupling and band - gap tuning in bilayer mos@xmath1 phys .
lett . * 112 * , 186802 ( 2014 ) . z. gong , g .- b .
liu , h. yu , d. xiao , x. cui , x. xu , and w. yao , magnetoelectric effects and valley - controlled spin quantum gates in transition metal dichalcogenide bilayers , nat .
* 4 * , 2053 ( 2013 ) .
a. m. jones , h. yu , j. s. ross , p. klement , n. j. ghimire , j. yan , d. g. mandrus , w. yao , and x. xu , spin - layer locking effects in optical orientation of exciton spin in bilayer wse@xmath1 , nat .
* 10 * , 130 ( 2014 ) .
r. suzuki , m. sakano , y. j. zhang , r. akashi , d. morikawa , a. harasawa , k. yaji , k. kuroda , k. miyamoto , t. okuda , k. ishizaka , r. arita , and y. iwasa , valley - dependent spin polarization in bulk mos@xmath1 with broken inversion symmetry , nat .
* 9 * , 611 ( 2014 ) .
p. blaha , k. schwarz , g. k. h. madsen , d. kvasnicka and j. luitz , wien2k , _ an augmented plane wave + local orbitals program for calculating crystal properties _
( karlheinz schwarz , techn .
universitt wien , austria ) , 2001 .
isbn 3 - 9501031 - 1 - 2 ; http://www.wien2k.at .
j. p. perdew , k. burke , and m. ernzerhof , generalized gradient approximation made simple , phys .
77 * , 3865 ( 1996 ) .
d. d. koelling and b. n. harmon , a technique for relativistic spin - polarised calculations , j. phys .
c * 10 * , 3107 ( 1977 ) .
j. kune , p. novak , r. schmid , p. blaha , and k. schwarz , electronic structure of fcc th : spin - orbit calculation with 6@xmath118 local orbital extension , phys .
b * 64 * , 153102 ( 2001 ) .
k. d. bronsema , j. l. de boer , and f. jellinek , on the structure of molybdenum diselenide and disulfide , z. anorg .
allg . chem . *
540 * , 15 ( 1986 ) .
a. a. golubichanaya , a. g. duksina , and v. l. kalikhman , variation of the antifriction and electrical properties of a number of solid lubricants with the oxidation temperature , inorg .
mater . , * 15 * , 1156 ( 1979 ) .
note that there are also recent publications @xcite utilizing the difference of the phase with @xmath119 and @xmath120 ( according to a different notation ) .
the former have addressed coupling between the k and @xmath0 points in the monolayer quantum dot .
the latter have discussed the interlayer hopping for the 2h stacking , giving the result consistent with ours .
liu , h. pang , y. yao , and w. yao , intervalley coupling by quantum dot confinement potentials in monolayer transition metal dichalcogenides , new j. phys .
* 16 * , 105011 ( 2014 ) .
liu , d. xiao , y. yao , x. xu , and w. yao , electronic structures and theoretical modelling of two - dimensional group - vib transition metal dichalcogenides , chem .
soc . rev . *
44 * , 2643 ( 2015 ) .
a. r. beal , and h. p. hughes , kramers - kronig analysis of the reflectivity spectra of 2h - mos@xmath1 , 2h - mose@xmath1 and 2h - mote@xmath1 j. phys . c. : solid state phys .
, * 12 * , 881 ( 1979 ) .
g. dresselhaus , effective mass approximation for excitons , j. phys .
solids * 1 * , 14 ( 1956 ) .
l. v. keldysh , coulomb interaction in thin semiconductor and semimetal films , jetp lett .
* 29 * , 658 ( 1979 ) .
a. chernikov , t. c. berkelbach , h. m. hill , a. rigosi , y. li , o. b. aslan , d. r. reichman , m. s. hybertsen , and t. f. heinz , exciton binding energy and nonhydrogenic rydberg series in monolayer ws@xmath1 , phys .
lett . * 113 * , 076802 ( 2014 ) . b. l. evans and p.
a. young , exciton spectra in thin crystals , proc .
a * 298 * , 74 ( 1967 ) .
a. baldereschi and m. g. diaz , anisotropy of excitons in semiconductors , nuovo cimento * 68b * , 217 ( 1970 ) .
h. ehrenreich and m. h. cohen , self - consistent field approach to the many - electron problem , phys .
115 * , 786 ( 1959 ) .
s. l. adler , quantum theory of the dielectric constant in real solids , phys . rev . * 126 * , 413 ( 1962 ) .
n. wiser , dielectric constant with local field effects included , phys .
129 * , 62 ( 1963 ) .
t. cheiwchanchamnangij and w. r. l. lambrecht , quasiparticle band structure calculation of monolayer , bilayer , and bulk mos@xmath1 , phys .
b * 85 * , 205302 ( 2012 ) . a. h. reshak and s. auluck , calculated optical properties of 2h - mos@xmath1 intercalated with lithium phys . rev .
b * 68 * , 125101 ( 2003 ) .
h. peelaers and c. g. van de walle , effects of strain on band structure and effective masses in mos@xmath1 , phys
b * 86 * , 241401(r ) ( 2012 ) . r. j. elliott , intensity of optical absorption by excitons , phys .
rev . * 108 * , 1384 ( 1957 ) .
w. kohn and j. m. luttinger , theory of donor states in silicon , phys .
98 * , 915 ( 1955 ) . | we find that the motion of the valley electrons
electronic states close to the k and @xmath0 points of the brillouin zone is confined into two dimensions when the layers of mos@xmath1 form the 3r stacking , while in the 2h polytype the bands have dispersion in all the three dimensions .
according to our first - principles band structure calculations , the valley states have no interlayer hopping in 3r - mos@xmath1 , which is proved to be the consequence of the rotational symmetry of the bloch functions . by measuring the reflectivity spectra and analyzing an anisotropic hydrogen atom model
, we confirm that the valley excitons in 3r - mos@xmath1 have two - dimensional hydrogen - like spectral series , and the spreads of the wave function are smaller than the interlayer distance .
in contrast , the valley excitons in 2h - mos@xmath1 are well described by the three - dimensional model and thus not confined in a single layer .
our results indicate that the dimensionality of the valley degree of freedom can be controlled simply by the stacking geometry , which can be utilized in future valleytronics . |
the goal of the paper is to intiate an approach to the hyperbolic volume conjecture , via asymptotics of solutions of difference equations with a small parameter .
the generalized volume conjecture links ( conjecturally ) the ( colored ) jones polynomial of a knot to hyperbolic geometry of its complement .
since the colored jones polynomial is a specific solution to a linear @xmath0-difference equation , it follows that the generalized volume conjecture is the wkb limit of a specific solution of a linear difference equation with a small parameter . motivated by this
, we study wkb asymptotics of formal and actual solutions of difference equations with a small parameter , under certain regularity asymptions . a knot in 3-space is a smooth embedding of a circle , considered up to isotopy .
two of the simplest knots , the trefoil ( @xmath1 ) and the figure eight ( @xmath2 ) are shown here : @xmath3 by the very definition , knots are flexible objects defined up to isotopy , which allows the embedding to move in a smooth and arbitrary way as long as it does not cross itself . in algebraic topology , a common way of studying knots ( and more generally , spaces ) is to associate computable numerical invariants ( such as euler characteristic , or homology ) .
invariants are useful in deciding whether two knots are not the same .
it is a much harder problem to construct computable invariants that separate knots .
the invariant that we will consider in this paper is the jones polynomial of a knot ; @xcite , which is a laurent polynomial with integer coefficients , associated to each knot . the quantum nature of the jones polynomial is apparent both in the original definition of jones ( using temperley - lieb algebras ) and in the reformulation , due to witten , in terms of the expectation value of a quantum field theory ; see @xcite .
the combinatorics associated to a planar projection of a knot show that the jones polynomial is a computable invariant .
however , it is hard to see from this point of view the relation between the jones polynomial and geometry . in quantum field theory ,
one often reproduces geometry by moving carefully chosen parameters of the theory to an appropriate limit . in our case
, we will introduce a new parameter , a natural number which roughly speaking corresponds to taking a connected @xmath4-parallel of a knot .
the resulting invariant is no longer a laurent polynomial , but rather a sequence of laurent polynomials .
the _ colored jones function _ of a knot @xmath5 in 3-space is a sequence of laurent polynomials @xmath6.\ ] ] the first term in the above sequence , @xmath7 is the jones polynomial of @xmath5 ; see @xcite .
although knots are flexible objects , thurston had the idea that their complements have a unique decomposition in pieces of unique `` crystaline '' shape .
the shapes in question are the @xmath8 different geometries in dimension @xmath9 , and the idea in question was termed the `` geometrization conjecture '' .
the most common of the @xmath8 geometries is hyperbolic geometry , that is the existence of a complete , finite volume , constant curvature @xmath10 riemannian metric on knot complements .
thurston proved that unless the knot is torus or a satellite , then it carries a unique such metric ; see @xcite . the _ hyperbolic volume conjecture _
( hvc , in short ) connects two very different views of knot : namely quantum field theory and riemannian geometry .
the hvc states for every hyperbolic knot @xmath5 @xmath11 where @xmath12 is the _ volume _ of a complete hyperbolic metric in the knot complement @xmath13 . the conjecture was formulated in this form by murakami - murakami @xcite following an earlier version due to kashaev , @xcite .
more generally , gukov ( see @xcite ) formulated a generalized hyperbolic volume conjecture that identifies the limit @xmath14 of a hyperbolic knot with known hyperbolic invariants ( such as the volume of cone manifolds obtained by hyperbolic dehn filling ) , for @xmath15-\bq$ ] or @xmath16 .
actually , the ghvc is stated for _ complex _
numbers @xmath17 . for simplicity
, we will study asympotics for real @xmath18 $ ] . at present
, it is not known whether the limit in the hvc exists , let alone that it can be computed .
explicit finite multisum formulas for the colored jones function of a knot exist ; see for example @xcite . from these formulas alone , it is difficult to study the above limit . in a sense , the question is to understand the sequence of laurent polynomials that appears in the hvc . if the sequence is in some sense random , then it is hard to expect that the limit exists , or that it can be computed . since the first term of this sequence is the jones polynomial , and
since we know little about the possible values of the jones polynomial , one would expect that there is even less to be said about the colored jones function .
luckily , the colored jones function behaves in a better way than its first term , namely the jones polynomial .
this can be quantified by recent work of ttq le and the first author , who proved that the colored jones function of a knot satisfies a @xmath0-difference equation . in other words , for every knot @xmath5 there exist rational functions @xmath19 ( which of course depend on @xmath5 ) such that for all @xmath20 we have : @xmath21 this opens the possibility of studying the @xmath0-difference equation rather than one of its solutions , namely the colored jones function .
although the @xmath0-difference equation is not unique , it was shown by the first author in @xcite that one can choose a unique @xmath0-difference equation , which is a knot invariant .
moreover , it was conjectured in @xcite that the characteristic polynomial of this @xmath0-difference equation determines the characters of @xmath22 representations of the knot complement , viewed from the boundary .
as was explained by the first author on several occasions , asymptotics of solutions of @xmath0-difference equations would have consequences on the hvc . in this introductory article
we review the history of asymptotics of solutions of @xmath0-difference equations .
excellent references for differential equations with a parameter are olver s and wasau s books ; @xcite and @xcite . in 1837 ,
liouville and green independently studied systematically existence of formal ( i.e. , perturbative ) and actual solutions for second order differential equations with a parameter ; see @xcite .
second order equations are very important for classical and quantum physics . in 1908
birkhoff had the insight to introduce and study _ arbitrary _ order differential equation with a parameter ( see @xcite ) : @xmath23 where @xmath24 , @xmath25 means @xmath4-th derivative with respect to @xmath26 ( assumed to be restricted to a real interval ) , and @xmath27 is a large complex parameter , and where the coefficient @xmath28 are complex @xmath29 functions with an expansion @xmath30 birkhoff s working assumption was that the eigenvalues @xmath31 of the characteristic equation @xmath32 were distinct but not necessarily nowhere vanishing . in 1926 ,
three theoretical physicists , wentzel - krammer - brillouin studied the second order differential equation under the assumption that its eigenvalues do not collide , and developed connection formulas linking solutions in the exponential region with those in the oscillatory region . their method is often referred to as the wkb method . as a motivation for our results ,
let us recall some fundamental results of birkhoff and trjitzinsky from 1930 on difference equations without a parameter ; see @xcite and @xcite . a _ difference equation _ for a discrete function @xmath33 has the form : @xmath34 where @xmath35 are discrete functions so that @xmath36 for all @xmath20 .
we will assume the existence of asymptotic expansions of @xmath37 around @xmath38 for all @xmath39 : @xmath40 where @xmath41 .
this certainly holds for @xmath42 if @xmath43 are rational functions of @xmath4 , as is often the case in combinatorial problems . due to the nowhere vanishing of @xmath44
, it follows that the set of solutions of is a vector space of dimension @xmath45 .
there are two main problems of difference equations : * existence of _ formal series solutions _
@xmath46 to . *
existence of a basis @xmath47 of solutions so that @xmath48 is asymptotic to @xmath49 for large @xmath4 . in @xcite
, birkhoff solved the existence of formal solutions in complete generality ( that is , without any assumptions on the eigenvalues of the characteristic equation ) . in a sequel paper @xcite ,
birkhoff - trjitzinsky solved the second problem in complete generality . among other things ,
the formal solutions of birkhoff lead to the development of differential galois theory , see @xcite .
decades later , the results of birkhoff and trjitzinsky on difference equations have found applications to enumerative combinatorics and numerical analysis ; see for example wimp and zeilberger in @xcite and references therein .
it is not surprising that difference equations are used in numerical analysis , since difference equations are numerical schemes of approximating differential equations . in enumerative combinatorics and complexity
theory , difference equations appear in recursive computation .
for example , the number @xmath50 of involutions of @xmath51 ( that is , permutations which are a product of 1 and 2-cycles ) is given by @xmath52 with @xmath53 , @xmath54 .
using the results of birkhoff - trjitzinsky and the fact that @xmath50 is monotone , it follows that @xmath55 for nonzero constants @xmath56 and some @xmath57 .
actually , the @xmath58 can be computed recursively from the difference equation ; see @xcite . by some historical accident , asymptotics of solutions of difference equations with a parameter
was not discussed a century ago .
the first paper that discusses second order difference equation with a parameter appears to be due to deift - mclaughlin ( see @xcite ) which was generalized by costin - costin to arbitrary order difference equations , @xcite .
the purpose in this paper is to show that for regular @xmath0-difference equations , a regular solution has a well - defined and computable exponential growth rate in terms of a relative entropy of the characteristic polynomial of the @xmath0-difference equation ; see theorem [ thm.1 ] below .
this subject is classical and has been reinvented over the past hundred years by several groups , often unaware of each others results . in a sense ,
the problem of formal solutions of @xmath0-difference equations is a problem in differential galois theory ; @xcite , and a problem in numerical analysis ; see for example @xcite . our results are hardly new and are contained or can be obtained by minor modifications from results of costin - costin or from work of birkhoff and collaborators , @xcite .
since the presentation in the above papers varies by time and taste , we have decided to give a self - contained account of the theory with complete proofs .
hopefully , this will benefit the researchers in quantum topology and in analysis . in this paper
, we will describe asymptotics of solutions of @xmath0-difference equations . a _ @xmath0-difference equation _ for a sequence @xmath59 of smooth functions of @xmath0 has the form : @xmath60 where @xmath61 are smooth functions and @xmath62 .
before we proceed further , let us remark that @xmath0 is a _
variable _ in and not a complex number of absolute value less ( or more ) than @xmath63 . in the usual analytic theory of @xmath0-difference equations
, @xmath0 is a complex number inside or outside the unit disk .
moreover , in the ghvc , we need to compute the @xmath4th term @xmath64 in the above @xmath0-difference equation , and then evaluate it at @xmath65 , for @xmath17 fixed .
in other words , in the ghvc , @xmath66 is a complex number that varies with @xmath4 in such a way that it stays in the unit circle and approaches @xmath63 as @xmath38 . with this in mind
, @xmath67-difference equations ( defined below ) are obtained from @xmath0-difference equations by the substitution @xmath68 where @xmath67 is a small nonnegative real number , that plays the role of planck s constant .
the _ characteristic polynomial _ of the @xmath0-difference equation is @xmath69 we will say that is _ regular _ if @xmath70 for all @xmath71 , where @xmath72 is the _ discriminant _ of @xmath73 , which is a polynomial in the coefficients of @xmath73 .
let @xmath74 denote the roots of the characteristic polynomial , which we call the _ eigenvalues _ of .
it turns out that is regular iff the eigenvalues @xmath74 never collide and never vanish , for every @xmath71 .
moreover , it follows by the implicit function theorem that the roots are smooth functions of @xmath71 . since we are interested in asymptotics of solutions of @xmath0-difference equations which , as we shall see , are governed by the magnitude of the eigenvalues , we need to partition the circle according to the magnitudes of the eigenvalues .
let @xmath75 denote a partition of @xmath76 into a finite union of closed arcs ( with nonoverlapping interiors ) , such that the magnitude of the eigenvalues does not change in each arc .
in other words , for each @xmath77 , there is a permutation @xmath78 of the set @xmath79 such that latexmath:[\[|\l_{\s_p(1)}(v)| \geq |\l_{\s_p(2)}(v)| \geq \dots \geq the following definition introduces a locally fundamental set of solutions of @xmath0-difference equations . fix a partition of @xmath81 as above .
a set @xmath82 is a _ locally fundamental set of solutions _ of iff for every solution @xmath83 for every @xmath77 and for every @xmath84 there exist smooth functions @xmath85 such that @xmath86 assume that is regular .
then , there exists a locally fundamental set of solutions @xmath82 such that * for every @xmath84 and @xmath87 such that @xmath88 we have @xmath89 * for some smooth functions @xmath90 with uniform ( with respect to @xmath91 $ ] ) _ asymptotic expansion _
@xmath92 where @xmath93 for all @xmath94 * and leading term @xmath95 where we have chosen a branch for the logarithm of @xmath96 . for every @xmath39
the smooth functions @xmath97 for positive @xmath94 are uniquely determined from the coefficients @xmath98 of by a hierarchy of first - order differential equations along with specified initial conditions . on the other hand ,
the smooth functions @xmath90 are not uniquely determined , since they are obtained by a smooth interpolation .
thus , the locally fundamental set of solutions is not uniquely determined from the @xmath0-difference equation , although its asymptotic behavior is .
it follows from theorem [ thm.1 ] that each locally fundamental solution @xmath99 of the @xmath0-difference equation satisfies the ghvc in the sense that for every @xmath18 $ ] we have : @xmath100 fix a solution @xmath83 of .
theorem [ thm.1 ] expresses @xmath83 as a linear combination of @xmath101 s in each arc @xmath102 . for every @xmath103 ,
let @xmath104 later ( in section [ sub.assregularq ] ) we will define the notion of a regular solution to a @xmath0-difference equation . as a prototypical example
, consider an @xmath0-difference equation that satisfies @xmath105 for all @xmath106 and all @xmath71 . then
, any solution that satisfies @xmath107 ( or more generally , @xmath108 has a nonvanishing derivative at @xmath109 ) is regular .
it is possible that @xmath110 .
in other words , the restriction of a fixed solution @xmath83 to different intervals @xmath102 may be a linear combination of different @xmath111s .
this is an important phenomenon , referred by the name of _ stokes phenomenon _ ; see @xcite .
our next definition captures the growth rate of regular solutions to regular @xmath0-difference equations .
fix a collection @xmath112 of subsets of @xmath113 . the @xmath114-_entropy _
@xmath115 \to \br\ ] ] of the @xmath0-difference equation is defined by @xmath116 where @xmath117 \to \br$ ] is defined by @xmath118 the _ entropy _ of is the set of functions @xmath119\to\br
\quad | \quad s \subset \{1,\dots , d\ } \}.\ ] ] notice that the entropy of a @xmath0-difference equation is not a real number , but rather a finite collection of functions .
the main result is the following if @xmath120 is an @xmath114-regular solution of the regular @xmath0-difference equation , then for every @xmath18 $ ] we have : @xmath121 finally , let us define the @xmath122-entropy of a knot . in @xcite
the first author showed that to every knot @xmath5 one can associate a canonical @xmath0-difference equation of degree @xmath45 , and a specific solution of it , namely the colored jones function of @xmath5 .
the @xmath0-difference equation itself is an invariant of a knot , which ( by definition ) is determined by the colored jones function of the knot .
thus , any invariant of the @xmath0-difference equation is also an invariant of a knot , which is determined by the colored jones function of the knot .
the @xmath122-entropy of a knot is the entropy of its associated @xmath0-difference equation .
we denote the @xmath122-entropy of a knot @xmath5 by @xmath123\to\br \quad |
\quad s \subset \{1,\dots , d\ } \}.\ ] ] the paper was written in the spring of 2004 . since then , a number of papers that discuss the asymptotics of the colored jones function have appeared ; see @xcite .
the results of this paper were announced in the jami 2003 meeting in johns hopkins . the first author wishes to thank j. morava for the invitation , and p. deligne who suggested the asymptotic behavior of solutions of @xmath0-difference equations .
the first author wishes to thank d. boyd for sharing and explaining his unpublished work and also a. riese , t. morley , and d. zeilberger .
in this section , we will translate asymptotics of solutions of @xmath0-difference equation in terms of asymptotics of solutions of _
@xmath67-difference equations_. the latter are defined as follows .
fix a positive number @xmath124 , a compact interval @xmath81 of @xmath125 and a natural number @xmath45 .
we will consider functions @xmath126 with domain @xmath127 , \quad k \e \in i \}.\ ] ] consider the @xmath67-difference equation for a function @xmath128 @xmath129 where @xmath130)$ ] .
we will assume that for all @xmath131 , @xmath132 has a uniformly ( with respect to @xmath26 ) asymptotic expansion @xmath133 where @xmath134 .
as we mentioned before , @xmath67-difference equations are obtained from @xmath0-difference equations by the substitution @xmath68 where @xmath67 is a small nonnegative real number , that plays the role of planck s constant .
the _ characteristic polynomial _ of is @xmath135 we will say that if @xmath136 for all @xmath137 . let @xmath138 denote the roots of the characteristic polynomial , which we call the _ eigenvalues _ of .
it turns out that is regular iff the eigenvalues @xmath138 never collide and never vanish , for every @xmath137 .
moreover , it follows by the implicit function theorem that the roots are smooth functions of @xmath137 . since we are interested in asymptotics of solutions of @xmath67-difference equations which , as we shall see , are governed by the magnitude of the eigenvalues , we need to partition the interval @xmath81 according to the magnitudes of the eigenvalues
. let @xmath139 denote a partition of @xmath81 into a finite union of closed intervals ( with nonoverlapping interiors ) , such that the magnitude of the eigenvalues does not change in each interval .
in other words , for each @xmath77 , there is a permutation @xmath78 of the set @xmath79 such that latexmath:[\[|\l_{\s_p(1)}(x)| \geq |\l_{\s_p(2)}(x)| \geq \dots \geq the following definition introduces a locally fundamental set of solutions of @xmath67-difference equations . fix a partition of @xmath81 as above .
a set @xmath82 is a _ locally fundamental set of solutions _ of iff for every solution @xmath141 , for every @xmath77 and for every @xmath84 there exist smooth functions @xmath142 $ ] such that @xmath143 here , the notation @xmath85 does not indicate the @xmath144th power of @xmath145 .
the next theorem summarizes the results of costin - costin .
( @xcite ) assume that is regular .
then , there exists a positive @xmath146 and a locally fundamental set of solutions @xmath82 of such that * for every @xmath84 and @xmath147 we have @xmath148 * for some smooth functions @xmath149)$ ] with uniform ( with respect to @xmath137 ) _ asymptotic expansion _ @xmath150 where @xmath93 for all @xmath94 * and leading term @xmath151 fix a solution @xmath83 of .
theorem [ thm.2 ] expresses @xmath83 as a linear combination of the @xmath111 s in each interval @xmath102 . for every @xmath103 ,
let @xmath152 later ( in section [ sub.assregulare ] ) we will define the notion of a regular solution to an @xmath67-difference equation . as a prototypical example
, consider an @xmath67-difference equation that satisfies @xmath153 for all @xmath106 and all @xmath154 $ ] .
then , any solution that satisfies @xmath155 ( or more generally , @xmath108 has a nonvanishing derivative at @xmath156 ) is regular .
our next definition captures the growth rate of regular solutions to regular @xmath67-difference equations .
fix a collection @xmath112 of subsets of @xmath113 .
the @xmath114-_entropy _ @xmath157 of the @xmath67-difference equation is defined by @xmath158 where @xmath159 is defined by @xmath160 if @xmath83 is an @xmath114-regular solution to a regular @xmath67-difference equation , and @xmath137 , we have : @xmath161 the next remarks concern the uniqueness of a set of locally fundamental solutions to . for every @xmath84
the smooth functions @xmath162 for positive @xmath94 are uniquely determined by and the initial condition @xmath163 .
indeed , applying taylor series ( with respect to @xmath67 ) in and collecting terms , we get for example : @xmath164 where @xmath165 and @xmath166 . similarly , for @xmath167 we have : @xmath168 where @xmath169 is a function of derivatives of @xmath170 and @xmath171 for @xmath172 .
notice that the denominator vanishes nowhere since the roots do not collide and do not vanish for every @xmath137
. if the coefficients @xmath132 of the regular @xmath67-difference equation are analytic functions , then the functions @xmath162 of theorem [ thm.2 ] are also analytic , for every @xmath173 and @xmath94 .
this follows by induction from the differential hierarchy which these functions satisfy , and from the fact that the eigenvalues are analytic functions .
even though @xmath162 is analytic for every @xmath173 and @xmath94 , the series @xmath174 is in general divergent , and the functions @xmath175 of theorem [ thm.2 ] are not analytic . even though the functions @xmath162 are uniquely determined by the @xmath67-difference equation , the smooth functions @xmath176 ( and thus the locally fundamental set of solutions @xmath101 ) are not uniquely determined by the @xmath67-difference equation .
the problem is that smooth interpolation is not unique .
recently developed ideas of exponentially small corrections might construct a unique set of locally fundamental solutions when the coefficients of are analytic functions .
we will elaborate on this in a separate occasion .
the translation of @xmath0-difference equations to @xmath67-difference equations is as follows . if @xmath120 satisfies the @xmath0-difference equation @xmath177 then set @xmath178 and consider the @xmath67-difference equation for a function @xmath179 ( with domain @xmath180 for some @xmath181 and @xmath182 $ ] ) : @xmath183 the following lemma , although elementary , is the key to translating @xmath0-difference equations to @xmath67-difference equations . for every @xmath184 $ ] we have : @xmath185 consequently , for every @xmath186 $ ]
, we have : @xmath187 thus , theorem [ thm.2 ] implies theorem [ thm.1 ] .
observe that @xmath188 , thus @xmath189 satisfies the equation @xmath190 and so does @xmath191 . since solutions with matching initial conditions are unique , follows .
equation follows from equation by the substitution @xmath192 : @xmath193
in this section we will review some linear algebra .
it is obvious that the complex roots of a monic polynomial uniquely determine it .
it is also known @xcite that the eigenvalues of a companion matrix uniquely determine it , in case they are distinct .
consider a _ companion _
@xmath45 by @xmath45 matrix @xmath194 the characteristic polynomial of @xmath195 is @xmath196 with roots @xmath197 .
let @xmath198 be the vandermonde matrix , and @xmath199 be the diagonal matrix with diagonal entries @xmath200 .
if a companion matrix has distinct eigenvalues , then with the above notation we have : @xmath201 observe that @xmath202 is an eigenvector of @xmath195 with eigenvalue @xmath203 .
thus , @xmath204 and @xmath205 .
the result follows .
now , consider a companion matrix @xmath206 whose entries in the bottom row are smooth functions in a variable @xmath207 , with roots @xmath208 which we assume are distinct for all @xmath207 .
the next lemma is a key estimate for the norm of long products of slowly varying matrices . in the language of physics
, @xmath206 is the _ transfer matrix _ and @xmath209 is the _ transition matrix_. assume that the eigenvalues @xmath208 of @xmath206 are distinct for all @xmath207 and @xmath210 then for @xmath211 , @xmath212 , we have @xmath213 by lemma [ lem.0 ] , we have : @xmath214 if @xmath211 , it follows that @xmath215 which implies that @xmath216 now , we have @xmath217 if @xmath218 $ ] , using the fact that @xmath219 , we obtain : @xmath220 if @xmath221 and @xmath222 are vandermonde matrices , such that @xmath223 is nonsingular , then @xmath224
in this section we will prove that has a unique set of formal solutions .
let us define those first .
a _ formal series _
@xmath225 is one of the form @xmath226 where @xmath227 are smooth functions for all @xmath94 .
note that @xmath228 lies in the ring @xmath229 $ ] of formal power series with coefficients smooth functions on @xmath81 .
note further that if @xmath225 is a formal series , so is @xmath230 for every @xmath231 , where the latter may be defined using the taylor series of @xmath232 : @xmath233 it follows that if @xmath225 is a formal series , then the ratio @xmath234 lies in the ring @xmath229 $ ] . using the language of _ difference galois theory _
( see @xcite ) this implies that @xmath229 $ ] is a finite difference ring , under the map @xmath235 .
we say that a formal series @xmath236 of is a _ formal series solution _ to iff @xmath237.\ ] ] it is easy to see that if @xmath236 is a formal solution to , then the leading term @xmath238 satisfies the equation @xmath239 where @xmath240 is an eigenvalue of .
assume that is regular .
then , has @xmath45 unique formal series solutions @xmath241 with leading terms corresponding to the eigenvalues of .
first we need to show that is indeed an equation in the power series ring @xmath229 $ ] , i.e. , that the terms involving negative powers of @xmath67 cancel .
suppose that @xmath236 is given by .
it follows from the calculation preceding lemma [ lem.galois ] that for every @xmath242 , we have : @xmath243 where @xmath244 denotes the coefficient of @xmath245 in a power series @xmath246 , and where @xmath247 is a polynomial in the derivatives of @xmath248 for @xmath172 .
expand the terms of equation into power series in @xmath67 using the above equation and , and collect terms of powers of @xmath67 .
it follows that is equivalent to a hierarchy of first order differential equations : @xmath249 where for positive @xmath94 , @xmath250 is a polynomial in the derivatives of @xmath248 and @xmath170 for @xmath172 .
now fix an @xmath251 , and choose @xmath252 such that @xmath253 , where @xmath254 are the eigenvalues of .
since is regular , it follows that the roots @xmath138 of the characteristic polynomial @xmath255 never collide , and never vanish for @xmath137 .
thus , @xmath256 for all @xmath137 .
thus , after we choose @xmath252 , it follows that we can find functions @xmath162 for @xmath257 that satisfy the above hierarchy .
moreover , for every @xmath173 , the sequence of functions @xmath162 is uniquely determined by the above hierarchy and the initial conditions @xmath163 . in this section
we present an alternative , and slightly more general form , of formal series . in case of regular @xmath67-difference equations
this alternative form will not be needed . however ,
when eigenvalues collide or vanish , one must use this alternative form of formal series .
thus , in the present paper we will not use this alternative form of formal series , and the reader may skip this section . an _ alternative formal series _
@xmath225 is one of the form @xmath258 where @xmath259 are smooth functions for all @xmath94 , and @xmath260 for all @xmath137 . in the remainder of this subsection
, we will refer to alternative formal series simply by formal series .
note that if @xmath225 is a formal series , so is @xmath230 for any @xmath231 , where the latter may be defined using the taylor series of @xmath261 and @xmath262 around @xmath26 .
it follows that @xmath263 moreover , if @xmath264 is a formal series , then the ratio @xmath234 lies in the ring @xmath229 $ ] .
this follows from and the following computation , valid for every @xmath265 : @xmath266 in analogy with lemma [ lem.galois ] , this implies that @xmath229 $ ] is a finite difference ring , under the map @xmath235 .
we say that a formal series @xmath236 of is a _ formal series solution _ to iff @xmath267.\ ] ] it is easy to see that if @xmath236 is a formal solution to , then the leading term @xmath179 satisfies the equation @xmath268 where @xmath240 is an eigenvalue of . in analogy with proposition [ prop.formal ]
, we have the following : assume that is regular .
then , has @xmath45 unique formal series solutions @xmath241 with leading terms corresponding to the eigenvalues of .
in this section we prove theorem [ thm.2 ] . the strategy is to * prove that there exists a solution @xmath269 with the stated properties where @xmath270 is an eigenvalue with maximum magnitude .
* use this solution @xmath269 to reduce theorem [ thm.2 ] to the case of a @xmath67-difference equation of degree one less than the original one .
* prove that the constructed set of solutions is a locally fundamental set .
without loss of generality , we will assume that the eigenvalues of satisfy the inequality : @xmath271 for all @xmath137 .
otherwise , we can partition @xmath81 into subintervals where this is true .
consider first a formal solution @xmath272 of given in proposition [ prop.formal ] , which satisfies and . consider the smooth functions @xmath273 of .
the proof of the following lemma ( due to borel in case @xmath274 are constant functions , for all @xmath94 ) can be found in ( * ? ? ? * lemma 2.5 ) : there exists a smooth function @xmath275)$ ] such that we have ( uniformly in @xmath137 ) : @xmath276 now , consider the unique solution @xmath269 of with initial conditions @xmath277 and for small enough @xmath67 , where without loss of generality , we assume that @xmath278 $ ] . of course , for large @xmath279 it may not be true that @xmath280 .
the next proposition estimates the error , uniformly with respect to @xmath279 : there exists an @xmath281 and constants @xmath282 such that for all @xmath283 and all @xmath242 , we have ( uniformly in @xmath279 ) : @xmath284 let us make a change of variables : @xmath285 where @xmath286 we will show that for a fixed @xmath287 , and for every @xmath94 there exists a constant @xmath282 such that for all @xmath288 we have : @xmath289 since @xmath269 satisfies , it follows that @xmath290 satisfies @xmath291 where @xmath292 it is easy to see that * @xmath293)$ ] , has uniform ( with respect to @xmath26 ) @xmath67-asymptotic expansion as in , * @xmath294 , * and since @xmath236 is a formal series solution to and @xmath295 is given by lemma [ lem.borel ] , if follows that for every @xmath94 we have : @xmath296 the characteristic polynomial of has roots @xmath297 for @xmath39 .
if is regular , so is .
notice that @xmath270 vanishes nowhere since is regular .
we now show .
let us write the difference equation in matrix form @xmath298 where @xmath299 and @xmath300 .
iterating , we obtain that @xmath301 for all @xmath302 , where @xmath303 , a column vector with all entries equal to @xmath63 .
equation gives : @xmath304 where @xmath305 uniformly in @xmath279 and @xmath67 .
feeding in the above equation , we obtain : @xmath306 now , let us look at the roots @xmath307 of @xmath308 since @xmath309 , it follows that @xmath310 since @xmath311 lies in @xmath81 , a compact set , lemma [ lem.1 ] and equation imply that @xmath312 for all @xmath313 , where @xmath314 .
this completes the proof of .
equations and differ in the presence of @xmath287 .
it is easy to see that if @xmath120 is a function such that for a fixed @xmath287 and any @xmath242 we have : @xmath315 then @xmath316 this observation shows that implies and concludes the proof of the proposition .
@xmath317 there exists a smooth function @xmath318)$ ] such that * for all @xmath283 , we have : @xmath319 * @xmath320 has an asymptotic expansion ( uniform with respect to @xmath26 ) : @xmath321 where @xmath274 are as in lemma [ lem.borel ] . as a result
, we have an asymptotic expansion ( uniform with respect to @xmath279 ) : @xmath322 consider the change of variables @xmath290 as in .
due to our choice of initial conditions it follows that for every fixed @xmath323 , the function @xmath324 is smooth . using this and the smoothness of the coefficients of
, it follows that for every fixed @xmath279 , the function @xmath324 is smooth .
so far , the function @xmath290 is defined on @xmath325 : @xmath326 which is a union of line segments in a rectangle @xmath327 $ ] , and satisfies .
the complement of these line segments in @xmath328 $ ] consists of an infinite union of open triangles , together with the horizontal segment @xmath329
. we can smoothly interpolate @xmath290 inside these open triangles so that it is defined on @xmath330 $ ] and @xmath331 $ ] and all @xmath242 .
let us extend @xmath290 to @xmath327 $ ] by defining @xmath332 for all @xmath137 .
we claim that @xmath290 is smooth on @xmath327 $ ] .
we need only to check this at the points @xmath333 for @xmath137 .
this follows easily from .
for example , to check continuity at @xmath333 , consider a sequence @xmath334 such that @xmath335 . then , for @xmath336 implies that @xmath337 , thus @xmath290 is continuous at @xmath333 . using for @xmath338
it follows that @xmath339 for all @xmath340 , and we find that @xmath290 has an @xmath67-asymptotic expansion ( uniform with respect to @xmath26 ) : @xmath341 restricting further @xmath342 if needed , we may assume that @xmath343 for all @xmath344 $ ] ; in other words @xmath345 makes sense for all @xmath344 $ ] .
now , we can finish the proof of proposition [ prop.step1 ] .
let us define @xmath346 then , implies ( a ) .
since @xmath347 is asymptotic to @xmath63 ( uniformly on @xmath26 ) , it follows that @xmath348 is asymptotic to @xmath349 . using the asymptotic of @xmath295 given by lemma [ lem.borel ] ,
( b ) follows
. we will now prove theorem [ thm.2 ] by induction on the degree @xmath45 of the @xmath67-difference equation .
for @xmath350 , it follows from proposition [ prop.step1 ] .
the inductive step is the next proposition .
assume that theorem [ thm.2 ] holds for regular @xmath67-difference equations of degree less than @xmath45 .
then it holds for regular @xmath67-difference equations of degree @xmath45 . _ proof .
_ consider a @xmath67-difference equation of degree @xmath45 .
we will use the solution @xmath269 of it constructed in proposition [ prop.step1 ] to reduce it to an equivalent equation of degree @xmath351 , and an inhomogeneous @xmath67-difference equation of degree @xmath63 . consider the dependent change of variables @xmath352 this is well - defined since @xmath269 is nowhere zero .
then , @xmath179 satisfies iff @xmath290 satisfies @xmath353 where @xmath354 the characteristic polynomials of and are related by @xmath355 as in the proof of proposition [ prop.step1 ] , it is easy to see that is a regular @xmath67-difference equation .
moreover , it is easy to see that theorem [ thm.2 ] holds for iff it holds for .
indeed , check that the change of variables given by preserves the asymptotics of the solutions of and .
thus , it suffices to work with . in that case
, @xmath356 is a solution of , since @xmath83 is a solution of .
it follows that @xmath357 ( compare this with ) .
let us define @xmath358 then , we get that @xmath359 is a solution of the @xmath67-difference equation @xmath360 where @xmath361 the characteristic equations of and are related by @xmath362 since @xmath363 ( by ) and @xmath364 , the same arguments of proposition [ prop.step1 ] imply that is regular , assuming that is regular . by the induction hypothesis
, it follows that satisfies theorem [ thm.2 ] . for the remainder of this section ,
fix a solution @xmath359 of which satisfies the properties of theorem [ thm.2 ] . in other words , @xmath359 satisfies and @xmath365 where @xmath366 is a smooth function with uniform ( with respect to @xmath26 ) asymptotics : @xmath367 there exists a formal solution @xmath368 of such that @xmath369 we need to solve the formal power series equation @xmath370 for @xmath371 in terms of @xmath366 . using the taylor expansion @xmath372 it is easy to see that the above equation equals to @xmath373
from which follows that @xmath374 . dividing the equation by @xmath375 we get an equation in formal power series with nonnegative powers of @xmath67 .
moreover , the coefficient of @xmath245 in that power series ( for @xmath257 ) equals to @xmath376 where @xmath169 is a function of @xmath377 and @xmath378 for @xmath379 . since @xmath380 is an eigenvalue of
, it is never equal to @xmath63 .
this and induction prove the lemma .
\(a ) there exists a solution to the equation @xmath381 for @xmath382 in terms of @xmath359 with appropriate initial condition .
( b ) for all @xmath383 we have : @xmath384 where @xmath385 fix @xmath386 and let @xmath218 $ ] . consider a natural number @xmath279 such that @xmath313 and @xmath387 .
this is equivalent to @xmath388 where @xmath389 and @xmath390 are natural numbers that depend on @xmath67 and @xmath81 , although we do not indicate this in our notation .
then equation implies that @xmath391 summing up , we obtain that @xmath392 choose initial conditions so that @xmath393 .
this completes part ( a ) .
part ( b ) follows by a telescoping calculation .
to finish the proof of proposition [ prop.step2 ] , it suffices to show that the solution @xmath359 of lemma [ lem.stepb ] is asymptotic to the formal solution @xmath394 of lemma [ lem.stepa ] .
since @xmath384 and @xmath395 it follows by the definition of @xmath290 given in lemma [ lem.stepb ] and by a telescopic sum , that : @xmath396 this concludes the proof of proposition [ prop.step2 ] .
let us summarize what we have obtained so far .
consider a partition @xmath397 of @xmath398 $ ] given by @xmath399 $ ] for @xmath400 , and consider a permutation @xmath78 of @xmath401 such that latexmath:[\[|\l_{\s_p(1)}(x)| \geq |\l_{\s_p(2)}(x)| \geq \dots \geq constructed solutions smooth functions @xmath90 for @xmath84 with asymptotic expansion given by and .
let us define @xmath403 where @xmath90 are smooth functions with asymptotic expansions as in equations and .
moreover , we have shown that for every interval @xmath102 ( as in the discussion prior to theorem [ thm.2 ] ) , @xmath404 is a set of solutions of when @xmath405 .
fix a solution @xmath406 of and an interval @xmath102 .
the following lemma certainly implies that @xmath407 is a locally fundamental set of solutions of .
this concludes the proof of theorem [ thm.2 ] .
in addition , the next lemma motivates the definition of a regular solution , given in the following section .
\(a ) fix @xmath83 and @xmath102 as above . then
, there exist smooth functions @xmath85 such that @xmath408 for all @xmath405 .
( b ) moreover , for every @xmath144 and @xmath84 we have @xmath409 for some smooth functions @xmath410 , with the understanding that @xmath411 . here
$ ] is the _
largest integer smaller than _ @xmath26 , and @xmath413 + 1.\ ] ] without loss of generality , let us assume that @xmath414 for @xmath39 .
equation is a linear equation in @xmath85 , with solutions @xmath415 where @xmath416 $ ] , @xmath417 and @xmath418 where the @xmath83 s are in the @xmath173th column of @xmath419 .
we will show that for small enough @xmath67 , @xmath420 is nonsingular . using equations , and
, it follows that @xmath421 uniformly in @xmath26 , where @xmath96 are the eigenvalues of . since is regular
, its eigenvalues never collide .
similarly , using equation , we have : @xmath422 where @xmath423 where the @xmath83 s are in the @xmath173th column of @xmath424 .
thus , @xmath425 the idea now is to move the recursion relation backwards @xmath45 times .
using the solution @xmath406 for @xmath426 will allow us to compute the smooth functions @xmath85 .
in detail , consider the matrix @xmath427 and move the recursion relation backwards once . using equation and the fact that the the @xmath428th column of @xmath429 for @xmath430 is an eigenvector of @xmath431 ( up to @xmath432 ) with eigenvalue @xmath433
, it follows that @xmath434 iterating @xmath351 more times , it follows that @xmath435 since @xmath436 , it follows ( as in the computation of @xmath437 above ) that : @xmath438 this , together with equation proves .
in this section we discuss regular solutions of @xmath0 and @xmath67-difference equations and their asymptotics . according to lemma [ lem.certainly ] , a solution @xmath83 to determines a collection @xmath112 of subsets of @xmath113 , where @xmath439 fix a collection @xmath112 of subsets of @xmath113 .
we say that a solution @xmath83 of a regular equation is @xmath114-_regular _ iff for every @xmath77 we have : * @xmath440 if @xmath441 .
* @xmath410 are _ not flat _ at @xmath109 for all @xmath442 .
that is , some derivative of @xmath410 at @xmath443 does not vanish . * for every @xmath77 there exists an element @xmath444 such that @xmath445 in other words , in the interior of the interval @xmath102 , and among the eigenvalues @xmath446 for @xmath447 , there is a unique eigenvalue of strictly maximum magnitude .
we will say that a solution to is _ regular _ if it is @xmath114-regular for some @xmath114 .
( of theorem [ thm.ass ] ) let @xmath83 be an @xmath114-regular solution to .
let us assume that @xmath448 , and @xmath449 of the closed interval @xmath278 $ ] .
fix an @xmath137 .
then , we have : @xmath450 for @xmath451 , where @xmath452 , and @xmath453 . then , we have : @xmath454 recall from theorem [ thm.2 ] that @xmath455 thus , @xmath456 combined with @xmath457 for @xmath458 , and @xmath459
, it follows that @xmath460 for all @xmath137 and @xmath461 therefore , @xmath462 and @xmath463 thus , equation implies that @xmath464 the result follows . in the general case , we partition the interval @xmath102 as in the discussion prior to theorem [ thm.2 ] and
repeat the above proof using equation .
the result follows .
first , we need to define what is a regular solution to a @xmath0-difference equation .
consider a solution @xmath83 of a @xmath0-difference equation and a partition of @xmath76 as in section [ sub.qdiff ] .
then , at each interval @xmath102 , we can write the solution as a linear combination of fundamental solutions , as in equation . let @xmath465 be the indexing set of the fundamental solutions that we use in each interval @xmath102 ; see .
suppose that the partition of @xmath76 is given by @xmath466 $ ] for @xmath467 , with @xmath468 .
then , with @xmath469 it turns out that for every @xmath144 and @xmath84 we have @xmath470 for some smooth functions @xmath410 , with the understanding that @xmath411 .
fix a collection @xmath112 of subsets of @xmath113 .
we say that a solution @xmath83 of a regular equation is @xmath114-_regular _ iff for every @xmath77 we have : * @xmath440 if @xmath441 .
* @xmath410 are _ not flat _ at @xmath109 for all @xmath442 .
that is , some derivative of @xmath410 at @xmath443 does not vanish . * for every @xmath77
there exists an element @xmath444 such that @xmath471 in other words , in the interior of the interval @xmath102 , and among the eigenvalues @xmath472 for @xmath447 , there is a unique eigenvalue of strictly maximum magnitude .
we will say that a solution to is _ regular _ if it is @xmath114-regular for some @xmath114 .
( of theorem [ thm.assq ] ) it follows from equation of lemma [ lem.translate ] and theorem [ thm.ass ] .
the coefficients of the @xmath0-difference equations are rational functions of @xmath0 and @xmath473 . in order to simplify the typesetting
, we will give the @xmath0-difference equation @xmath474 in operator form @xmath475 where @xmath476 and where the operators @xmath477 , @xmath478 and @xmath0 , act on a discrete function @xmath479 $ ] by @xmath480 note that @xmath0 commutes with @xmath478 and @xmath477 , and that @xmath481 . in @xcite
, the first author showed that the colored jones function @xmath488 of a knot @xmath5 satisfies an essentially unique smallest degree @xmath0-difference equation @xmath489 where the coefficients @xmath490 of @xmath491 are rational functions of @xmath207 and @xmath492 with rational coefficients .
the @xmath195-polynomial of @xmath5 parametrizes the moduli space of characters of @xmath495 representations of @xmath496 , restricted to the boundary torus @xmath497 .
the @xmath195-polynomial of a knot is an important ingredient to the geometrization of the knot complement and its dehn fillings .
* it has integer coefficients and even powers of @xmath223 , that is @xmath499 $ ] . *
it is reciprocal , that is , @xmath500 .
* it is tempered , that is the edge polynomials of its newton polygon are cyclotomic . *
it specializes to @xmath501 for some integers @xmath502 .
* @xmath503 is always a factor of @xmath494 , that corresponds to @xmath504 representations . if the colored jones function of a knot was an @xmath114-regular solution to a regular @xmath0-difference equation , and if the aj conjecture were true
, then it follows that for every @xmath18 $ ] we have : @xmath505 where @xmath506 is the _ @xmath195-entropy _ of a knot , defined as follows . for a knot @xmath5 in @xmath498 ,
let @xmath507 for @xmath508 denote the roots of the equation @xmath509 for @xmath510 $ ] , where @xmath45 is the @xmath511-degree of @xmath494 .
fix a partition @xmath512 of @xmath513 $ ] by closed intervals with nonoverlapping interiors and a permutation @xmath78 of the set @xmath401 such that latexmath:[\[|l_{\s_p(1)}(t)|
\geq |l_{\s_p(2)}(t)| \geq \dots \geq it is natural to ask how the @xmath195-entropy of a hyperbolic knot ( evaluated at @xmath16 ) compares to the hyperbolic volume .
the answer to this question is essentially contained in work of d. boyd , @xcite , which we quote without proof here .
we urge the reader to look in @xcite for beautiful and suggestive calculations .
more generally , among the roots @xmath507 there is a distinguished one , @xmath522 , corresponding to the discrete faithful representation when @xmath523 .
let @xmath524 denote the eigenvalue corresponding to the @xmath504 representations .
boyd informs us that for @xmath525-bridge knots @xmath5 ( in particular , for the @xmath1 and @xmath2 knots ) , it is true that @xmath526 it follows by ( s4 ) that the eigenvalues collide at @xmath527 .
moreover , boyd informs us that for @xmath525-bridge knots there exists a @xmath528 such that @xmath529 for all @xmath428 and all @xmath530 . in this section
we discuss in detail @xmath0-difference equation of the the colored jones function of the two simplest knots , namely the trefoil @xmath1 and the figure eight @xmath2 .
the former is not hyperbolic , and the latter is . in @xcite
, the first author computed that the colored jones function @xmath531 ( resp .
@xmath532 ) satisfies the second ( resp .
third ) order @xmath0-difference equation @xmath533 where the noncommutative @xmath195-polynomials @xmath534 and @xmath535 are given by : for @xmath2 knot , we have 3 eigenvalues @xmath540 , @xmath541 and @xmath542 . assuming appropriate choices for the branches of the eigenvalues , the plot of @xmath543 and @xmath544 for @xmath545 $ ] is given by :
since the hvc is true for the @xmath2 knot , it suggests that the colored jones function lies in a strictly smaller subspace of the vector space of solutions to the @xmath0-difference equation @xmath489 . using work of murakami @xcite
, one can figure out exactly the selection principle ; that is which locally fundamental solutions contribute to the colored jones function . *
the eigenvalues collide at @xmath527 ( since @xmath548 ) , at @xmath549 ( since @xmath550 ) and by symmetry at @xmath551 and @xmath552 .
* there is resonance on the interval @xmath553 $ ] where all three eigenvalues have equal magnitude .
* there is vanishing of the coefficients at @xmath554 ( since the denominator @xmath555 of the coefficients is singular at @xmath554 ) . | in this paper we develop an asymptotic analysis for formal and actual solutions of @xmath0-difference equations , under a regularity assumption , namely the non - collision and non - vanishing of the eigenvalues .
in particular , evaluations of regular solutions of regular @xmath0-difference equations have an exponential growth rate which can be computed from the @xmath0-difference equation .
the motivation for the paper comes from a problem in quantum topology , the hyperbolic volume conjecture , which states that a sequence on laurent polynomials ( the so - called colored jones function of a knot ) , appropriately evaluated , becomes a sequence of complex numbers that grows exponentially .
moreover , the exponential growth rate is proportional to the volume of the knot complement .
the connection of the hyperbolic volume conjecture with the paper comes from the fact that the colored jones function of a knot is a solution of a @xmath0-difference equation , as was proven by ttq .
le and the author . |
atom chips are microfabricated devices that control electric , magnetic and optical fields in order to trap and manipulate cold atom clouds @xcite , @xcite , @xcite , @xcite and to form bose - einstein condensates @xcite , @xcite , @xcite .
potential applications include atomic clocks @xcite , atom interferometers @xcite , @xcite , and quantum information processors @xcite , @xcite .
silicon is one of several materials used as a substrate for atom chips .
it is attractive for this purpose because its properties are well - known and fabrication techniques are highly developed .
the small scale of microfabricated current - carrying wires makes it easy to generate strong magnetic field gradients near the surface of the chip , forming tight traps for paramagnetic atoms .
the loading of such magneto - static traps usually starts with a magneto - optical trap ( mot ) typically some 3 - 4 mm from the surface .
this collects atoms from a tenuous , room - temperature vapour and cools them , typically to 100@xmath0k , using circularly polarised light beams in conjunction with a spherical quadrupole magnetic field .
the atoms are sometimes further cooled to a few tens of @xmath0k using optical molasses , before being captured in a weak magnetic trap to form a large atom cloud , typically 1 mm in size . at this point
, the atoms still have to be handed over to the microscopic magnetic traps on the chip , a process that involves further compression of the cloud and very accurate positioning of the atoms .
this sequence of loading and transfer is complicated and could be largely eliminated if the mot were integrated into the chip .
moreover , integration would open up the possibility of building arrays of mots to prepare large numbers of independent cold atom clouds .
this paper describes the fabrication and initial testing of an integrated array of mots on an atom chip , as proposed by trupke _ _
et al.__@xcite .
each of these mots automatically prepares all the required light beams from a single circularly polarised input beam by reflecting the light in a concave square pyramid of mirrors @xcite .
this greatly reduces both the number of expensive optical components needed to prepare the light beams and the amount of laser power needed .
integrated wires encircling the opening at the base of the pyramid produce the required magnetic field distributions with modest electrical power consumption and accurate positioning .
the fabrication of an integrated mot array on a chip represents an important step towards a truly integrated atom chip for portable applications .
the atom chip we have fabricated has 6 rows of pyramids , ranging in size from 200@xmath0 m to 1200@xmath0 m , serviced by 12 separate wires to produce the magnetic fields . for pyramids
up to 600@xmath1 m , the encircling wires have a width of 25@xmath0 m .
the larger pyramids are serviced by wires of 50@xmath0 m width . in total
there are 48 pyramid mots .
the whole chip is packaged into a ceramic pin grid array ( cpga ) with multiple wire bonds to bring high currents in and out of the chip .
the silicon sidewalls of the pyramids are coated with gold to create micro - mirrors for reflecting the laser light .
the pyramids formed by etching silicon have a 70.5@xmath2 apex angle , rather than the ideal 90@xmath2 .
the optical properties of such a pyramid have already been investigated in @xcite . in those experiments
it was observed that light reflected near the diagonal edges can prevent the mot from working .
here we have developed the necessary fabrication steps to eliminate these reflections by removing the gold near the corners of the pyramids .
this paper is organised as follows .
section[sec : principles ] outlines the principles of atom trapping in these pyramidal micro - mirrors , section[sec : fabrication ] describes the microfabrication , section[sec : characterisation ] presents initial tests of the device and section[sec : outlook ] discusses some prospects for using the chip in applications .
circularly polarised light , incident along the axis of a square pyramid , is reflected by the four metal mirrors that form the pyramid . at each reflection the helicity of the light is reversed . if the pyramid has a 90@xmath2 angle between opposite faces , these reflections produce three counter - propagating pairs of light beams that are mutually orthogonal .
together with a magnetic quadrupole field , this configuration creates a mot , whose radiation pressure forces cool and trap atoms from a room - temperature vapour @xcite , @xcite .
we form the pyramids by etching a silicon wafer , cut on the \{100 } plane .
potassium hydroxide ( koh ) etches anisotropically through square openings to reveal the \{111 } planes , which form hollow pyramids with apex angle @xmath3 .
this departure from a right angle causes the beams to be reflected into a variety of directions .
we classify these beams as type 1 , 2 , or 3 , according to the region of the pyramid where the first reflection occurs , as shown in fig.[fig : rays]a .
type-1 rays are reflected on two opposite sides of the pyramid before leaving , as illustrated in fig.[fig : rays]b .
type-2 rays also reflect on opposite faces , but strike the original face again before leaving , as shown in fig.[fig : rays]c , making a total of three reflections .
after the first reflection in fig.[fig : rays]d , type-3 rays head towards the opposite face , but because they are incident close to the diagonal edge of the pyramid , they are intercepted on the way by the adjacent face . here
they undergo a grazing reflection , marked ( 2 ) in the figure , where the helicity of the light is reversed .
finally , the opposite face is reached for a third reflection .
the wires electroplated on our chip form a square loop of side @xmath4 around the pyramid base .
a current @xmath5 in this loop makes a magnetic field as shown in fig.[fig : mot]a . at the centre of the loop , in the plane of the wire
, the field points downward with a magnitude of @xmath6 .
we superpose a uniform vertical bias field in order to create the quadrupole field configuration required by the mot , as shown in fig.[fig : mot]b .
the strength @xmath7 of this bias is chosen to centre the quadrupole halfway between the base and the apex , at a distance @xmath8 from the surface of the chip .
this requires @xmath9 .
the light beams needed to create the magneto - optical trap inside the pyramid are formed from the incoming circularly polarised beam by the first and second reflections of the type-1 rays , as shown in fig.2c .
the type-2 rays produce a slight imbalance in the mot force , but are largely unimportant .
the type-3 rays tend to destabilise the mot because they have the wrong helicity and produce a strong force that pushes the atoms out of the pyramid instead of trapping them .
we have conducted preliminary experiments using a large , glass pyramid coated with gold . initially , this failed to produce a mot because of the presence of the type-3 rays .
atoms were successfully trapped once the gold was removed from the areas where the type-2 and type-3 rays are produced .
we were also able to make the mot operate @xcite by using a lower - reflectivity ( 78% ) coating of aluminium , which was not cut away at the edges and centre .
in such a pyramid , the intensity of the harmful type-3 light is decreased relative to the type-1 beams , because of the additional reflection by the lossy surfaces .
the current in the wire determines the vertical gradient in the mot , which is related to the corresponding bias field according to @xmath10 . under typical operating conditions used in most mot experiments ,
the gradient is approximately 0.15t / m .
this is readily achieved on the chip because of the small scale : for example in the 1 mm loop it requires 0.1a . the same wires can also be used to create a purely magnetic trap , with a depth of @xmath11 , where @xmath12 is the magnetic moment of the atom .
such a trap can hold atoms for many seconds provided its depth exceeds the temperature of the atoms by a factor of 5 or so .
ideally , we would like to be able to hold a @xmath13k cloud , which requires a 0.7mt bias field , corresponding to a current in a 1 mm loop of 1a .
the feasibility of using such currents is discussed more fully in sec.[subsec : heatingtests ] .
further details about magnetic trapping of neutral atoms can be found in ref.@xcite . in summary ,
the aims of the fabrication were to create pyramidal micro - mirrors which would not reflect the damaging type-3 beams , and to surround these mirrors with wires able to create the required magnetic field strengths and gradients .
in the following sections we describe how these aims have been achieved .
the fabrication processing sequence is shown in fig.[fig : flow ] .
we start by preparing the wafer and etching the pyramids . the whole surface is then coated with gold and patterned to remove the gold from the type-3 regions and to form the wire tracks . in these steps ,
the main fabrication challenge was to cover the wafer with a uniform resist layer , because the pyramid openings in the surface impeded the flow of the resist during spinning .
finally , the wires are electroplated to make them thick enough to carry the required current .
these processes are described in more detail in the next three sub - sections .
the process begins with a 4 inch , 1mm - thick silicon wafer cut on the \{100 } plane .
this is given a standard rca and fuming nitric acid clean .
a 170nm - thick layer of silicon dioxide is grown by wet oxidation in a furnace at 1000c ( fig.[fig : flow]a ) .
subsequently a 50 nm layer of low stress silicon nitride is deposited on both sides by lpcvd .
a reactive ion plasma etch is then used to make alignment marks on the back side of the wafer for accurate positioning of all the masks used in the fabrication process . in order to make the pyramids , a 1@xmath12 m layer of photo resist az6612
is spun onto the wafers .
an array of square openings is patterned into the resist using a photo mask on a karl suss ma8 contact mask aligner .
the silicon nitride and dioxide layers are removed through these openings using a dry plasma etch and the resist is then stripped in a plasma asher ( fig.[fig : flow]b ) . before etching the silicon , the backside alignment marks
are protected by depositing a layer of pecvd silicon nitride 1@xmath0 m thick and the wafer edges are protected by ptfe tape .
the wafer is then etched for 19 hours in koh at a concentration of 33% by volume and at a temperature of 80c .
this produces pyramidal pits bounded by the four most slowly etched surfaces @xmath14 , @xmath15 , @xmath16 and @xmath17 ( fig.[fig : flow]c ) .
the faces of the pyramids are very smooth because of the layer - by - layer etching mechanisms involved @xcite and have been shown to have rms surface roughness as low as 0.5 nm @xcite .
the wafers are once again cleaned in a fuming nitric acid bath before the remaining silicon nitride is stripped from the front by a dry plasma etch and the remaining silicon dioxide is removed in an hf dip ( fig.[fig : flow]d ) .
finally , the whole front surface is covered with a plasma - enhanced teos oxide layer , 170 nm thick .
this is to provide electrical insulation between the silicon and the metallic coating that comes next .
the metal coating consists of 50 nm of chromium and 100 nm of gold , evaporated onto the front of the wafer ( fig.[fig : flow]e ) . in order to make the flower pattern on the faces of the pyramids , we require them to be covered by a uniform layer of resist .
spinning the resist at this stage does not produce a uniform layer because of the large depth of the pyramidal pits . in order to avoid this problem we use electrophoretic deposition of eagle 2100 negative photo resist .
the wafer is placed in the resist bath and heated to 33c , where it acts as a cathode at -125v .
it remains in the bath until the current drops to zero , then it is rinsed in de - ionised water and dried in a vacuum oven at 65c . at this point ,
the resist remains tacky , so the wafer is dipped into eagle 2002 topcoat for 30 seconds and again dried in the vacuum oven .
this method leaves a highly uniform layer of resist over the complex topography ( fig.[fig : flow]f ) .
next , the wafer is exposed to ultraviolet light for 77s at 6.5mw/@xmath18 , using a mask designed to remove the gold from the type-3 regions .
this pattern is developed using eagle 2005 developer heated to 40c , which is sprayed onto the surface for 4 minutes .
the wafer is then dipped for 1 minute in water at 80c to remove the resist residues and to smooth the surface of the resist .
it is dried in a vacuum oven and de - scummed for 3 minutes in oxygen plasma at 110c ( fig.[fig : flow]f(i ) ) .
the exposed gold and chromium are removed by a 35s potassium iodide etch , followed a 5s chrome etch ( fig.[fig : flow]f(ii ) ) .
the wafers are then sprayed for 15min to strip the resist with eagle 2007 remover at 50c , and placed in an asher for 1hr at 600w and 110c ( fig.[fig : flow]f(iii ) ) .
this process leaves a flower pattern on the gold inside each pyramid , as shown in fig.[fig : sem ] .
this scanning electron microscope image shows that the gold has been removed in the diagonal and central regions corresponding to the type-2 and type-3 areas illustrated in fig.[fig : rays]a .
we found that it was difficult to achieve a uniform exposure of the resist inside the pyramid .
when the lower part of the pyramid ( near the apex ) is correctly exposed , the upper part tends to be over - exposed .
the dark lines on the diagonals and the dark spot at the apex are due to residual resist .
the colour of the silicon surface appears darker at increasing depth in the pyramid .
we attribute this to charging of the silicon dioxide surface by electron bombardment from the sem .
the final stage of fabrication is to form the wires around the base of each pyramid , as described in @xcite .
first , the tracks are patterned by optical lithography on the gold coating , then the wires are electroplated to provide the 3@xmath12 m thickness required for carrying the electrical current . m pyramid .
( b ) on the side furthest from the axis of spinning , the wire of a 1.2 mm pyramid is malformed because the resist was too thin . ] once again , spinning a uniform layer of resist is very challenging since the radial flow is interrupted by the pyramid openings . to solve this problem we use the viscous photoresist az4533 as follows .
a primer is spun onto the wafer to help adhesion .
then the wafer needs to be completely flooded with the resist before the spinning begins to ensure coverage inside pyramids and in the areas of the spin shadow just outside the pyramids .
spinning is carried out at 500rpm for 10s , followed by 30s at 2000rpm .
we create a 5@xmath12 m layer of resist , which varies in thickness by approximately @xmath193@xmath12 m where the pyramids interfere with the flow ( fig.[fig : flow]g ) .
the uniformity of the resist improves substantially around pyramids that are smaller than 1 mm , and impeccable wires are formed around the smallest pyramids , as shown in fig.[fig : tracks]a .
however , the thinning of the photo - resist in the spin shadow of the larger pyramids leads to overexposure and results in a thinning of the wire tracks there , as shown in fig.[fig : tracks]b . a 6@xmath12m - thick electroplating mould is then created by optical lithography of az9260 which is spun at 500rpm for 8s , followed by 30s at 4000rpm .
this layer is exposed at 200mw/@xmath18 for 30s to create the plating mould with a minimum thickness of 3@xmath12 m ( fig.[fig : flow]h ) .
the electrochemical deposition is controlled by an autolab pgstat30 .
we use a commercial cyanide - free gold plating solution ( metalor technologies uk , with 1 g of gold per litre of solution ) .
a standard three - electrode setup is used with a platinum counter electrode .
the deposition bath is placed in a water bath kept at 50c and the solution is agitated throughout the deposition process using a magnetic stirrer .
the current is fixed at 6ma , corresponding to a current density of 5ma/@xmath18 , and the plating is run for 15 minutes to achieve a 3@xmath12m - thick gold deposit ( fig.[fig : flow]i ) . finally , the resist is stripped to leave the free - standing gold wires .
electrophoresis could potentially provide a solution to the problem of the thinning of the wires .
however , it can only be used when pattering the wire tracks and not when creating the electroplating mould , since the eagle 2100 resist can only cover areas previously covered with gold . for this reason
the fabrication of the wires was entirely designed and performed with the use of spin - on positive photoresists .
figure[fig : mounted ] shows on the right an array of patterned pyramidal micro - mirrors with integrated current - carrying wires .
the chip is fixed to a ceramic pin grid array ( cpga ) package ( cpg18023 , spectrum semiconductor materials inc . ) using an epoxy compatible with ultra - high vacuum ( bylapox 7285 ) .
gold bond wires connect the package pins to the microfabricated chip wires , with 9 bond wires serving each end of a chip wire .
the view on the left of fig.[fig : mounted ] shows the bond wires connecting the chip to the cpga .
the ceramic package is then plugged into a pin grid array ( pga ) of sockets , which in turn is soldered into an fr4 printed circuit board .
this design permits a large number of connections to be made in the limited space available in the high - vacuum chamber .
we place the chip assembly in a stainless steel vacuum chamber where a 20l / s ion pump yields a pressure below the 10@xmath20 mbar limit of our gauge after modest baking at 100c for 2 days . at this low pressure ,
a trapped atom will remain undisturbed by the background gas for typically 10s .
a stray magnetic field moves the zeros of magnetic field that form the centre of each mot . to ensure that every pyramid encloses a field zero
, we require the stray field to be less than 10@xmath12 t over the chip .
much stronger fields were initially found in the recess of the cpga where the wafer is mounted , due to permanent magnetisation of the nickel / gold coating .
we therefore removed the coating by sandblasting . immediately above the kovar pins
, there remained a stray field that reached a maximum of 10@xmath12 t , but this was of short range and did not affect the 3 cm x 3 cm active area of the chip .
there the field was below 500nt , the gradients were similarly negligible , and remagnetising the assembly produced no perceptible change at that level .
in this section we characterise the magnetic properties of our device , demonstrating that the micro - fabricated wires can carry sufficient current for magneto - optical and purely magnetic trapping . measurements of the heating of our chip wires allow us to infer the maximum depth and gradient achievable in our trap .
finally , it is shown that the flower patterning of the pyramid coating leads to the desired suppression of type-3 beams .
the wires around the base of each pyramid almost form a square loop , as shown in fig.[fig : tracks]a . in order to calculate the magnetic field of the mot
, it is adequate to approximate this by a fully closed square loop .
the solid lines in figs.7a and b show the vertical field component calculated for a current of 480ma in wire 50@xmath12 m wide , centred on a 895@xmath12 m square .
this corresponds to the geometry of the 800@xmath12 m pyramid .
we measured the field above this pyramid using a hall probe ( lakeshore 421 gaussmeter ) , as shown by the circles in fig.[fig : hallscan ] . on
the left is a vertical scan along the axis of the pyramid , while the scan on the right is horizontal at a height of 0.7 mm .
the two are entirely consistent .
m pyramid with 480ma in the wire .
dots : hall probe measurement .
lines : theory .
( a ) vertical variation along the pyramid axis .
( b ) transverse variation parallel to the pyramid row , through the pyramid axis at a height of 0.7 mm . ]
although the gold wires are very good conductors , they do of course dissipate electrical energy and , at sufficiently high currents , they blow like a fuse . considerably below that limit , the wafer heats up to 120c . at this temperature
the epoxy used to bond the silicon to the cpga begins to decompose , losing strength and outgassing strongly , thereby compromising both the mechanical stability of the chip and the high vacuum .
this limiting temperature determines the largest current that we run through the wires , and thereby limits the field and field gradient that are available . with the chip mounted and placed under moderate vacuum ( 10@xmath21 mbar ) , we first tested how much current could be passed through individual bond wires 50@xmath12 m in diameter and 2 - 3 mm in length .
we found that these blow at approximately 1.8a , but can survive indefinitely at 1.5a . since each chip pad
is normally connected by 9 wires in parallel , failure of the bond wires is not a limiting factor . in order to determine the operating currents for the chip wires ,
we monitored the temperature of the assembly at several points using thermocouples and we monitored the temperature of the wire itself by measuring the increase of its resistance . passing a current through the 50@xmath12m - wide chip wires , we measured the time it takes for the wire to reach 120c , the results being shown by the filled circles in fig.[fig : heating ] . in all cases the chip itself was much colder than the wire .
below 1a , the temperature limit was never reached , but at 1.3a , the wire approached 120c in ten seconds . for the 25@xmath12m - wide wires ,
the operation was continuous below 0.5a and was limited to ten seconds at 0.75a , as illustrated by the open squares in fig.[fig : heating ] .
m - wide wires .
open squares : 25@xmath12m - wide wires . ] since a suitable field gradient for the mot is 0.15t / m , the normal operating current is 5ma for a 200@xmath12 m pyramid and 100ma for a 1 mm pyramid . at these low currents
there is negligible heating of the chip .
by contrast , the 1 mm loop needs to operate at 1a if it is to make a purely magnetic trap for a 100@xmath12k cloud of atoms .
this can not be sustained indefinitely , as shown in fig.[fig : heating ] , but substantially less than one second should suffice for most practical applications .
since the field scales as @xmath22 , the situation is even better for purely magnetic trapping in the smaller loops .
in the course of these measurements , we found that the resistivity of the gold wire on the chip is 3.9x10@xmath23-m , approximately 1.6 times higher than that of bulk gold .
this is typical of electro - deposited gold .
the main consequence for us is a slightly higher power dissipation than we had anticipated in our design using the book value for the resistivity .
observation of the pyramids under a microscope allowed us to check that the type-2 and type-3 reflections were indeed eliminated by removal of the gold coating to create the flower pattern .
figure[fig : heating]a shows the image using unpolarized light with the microscope focussed on the apex of an unpatterned pyramid . in this figure
most of the area is bright since the mirrors are reflecting light over the whole area of the pyramid .
by contrast , fig.[fig : heating]b shows the same pyramid illuminated with linearly polarized light and viewed through a crossed polarizer , which suppresses all but the type-3 contribution , making the corner regions bright .
figures[fig : heating]c and d show the corresponding images for a patterned pyramid . in c
we see that there are no longer reflections near the apex of the pyramid , in the region of the type-2 rays . in d
, we see that the removal of the gold at the edges has completely suppressed the type-3 reflections .
it is useful to estimate the number of cold atoms that may be captured in one of these micro - pyramids .
this will depend on the usual operating variables , such as the laser detuning from the atomic transition , the positioning of the magnetic field zero , the balance of the light beams inside the pyramid and the pressure of the atomic vapour .
in addition , these pyramids are at the extreme limit of small laser beam size .
normally the laser beams of a mot are one or two centimetres in diameter , whereas these pyramids are less than a millimetre across .
according to the well - established model first described by wieman _
_ et al.__@xcite , the number of atoms @xmath24 captured in a mot is expected to scale as @xmath25 .
here the @xmath26 factor derives from the area of the laser beam , which in our case is set by the area of the pyramid opening .
the quantity @xmath27 is the capture velocity , i.e. the speed of the fastest atoms captured by the mot from the thermal background vapour . using the model of ref.@xcite ,
we have computed @xmath28 numerically , setting the maximum allowed stopping distance equal to the vertical height of the pyramid .
the circles in fig.[fig : uc ] show the results on a log - log plot for a variety of pyramid sizes , with the laser detuning optimised separately to maximise @xmath27 for each size . for pyramids larger than 1 mm , we find the empirical scaling law @xmath29 ( the solid line in fig.[fig : uc ] ) , leading to the result @xmath30 . when the pyramids are smaller than that , the atom number drops a little more rapidly .
we have established in a previous experiment@xcite that a 16 mm pyramid made from glass blocks can capture @xmath31 atoms . with the help of this model
, we extrapolate our 16 mm result to predict that a 1 mm pyramid will capture some @xmath32 atoms .
the smallest pyramids on this chip , having @xmath33 mm , are expected by the same argument to collect approximately 25 atoms .
although this model is a reasonable estimate , it does neglect some aspects of the full 3-dimensional geometry .
for example , it assumes that the atoms of the vapour have the normal thermal distribution close to the walls of the pyramid and it neglects the polarisation gradients in the laser field , which lead to additional sisyphus cooling@xcite .
experiment will have to determine how many atoms are actually captured .
the number of atoms needed depends of course on the application . at one extreme , with an array of small clouds , each containing perhaps @xmath34 atoms ,
the relative displacement of the clouds could provide a map of local magnetic field variations or be used to sense inertial forces . at the other extreme , the pyramids could serve as single - atom sources for loading integrated optical cavities , which have recently been demonstrated @xcite .
this would permit the production of single photons on demand for applications in quantum information processing . .
, title="fig : " ] [ fig9 ]
we have fabricated a silicon chip designed to trap cold atoms in an array of integrated magneto - optical traps .
the device contains 48 micro - fabricated hollow micro - pyramids surrounded by electroplated gold wires .
this was packaged into a ceramic pin - grid array , and the chip s optical and magnetic properties were tested .
we found that the chip wires can easily sustain the currents needed to operate the mots and can even operate for many seconds at the much higher currents needed to trap atoms magnetically .
the use of electrochemical deposition of photoresist allowed us to pattern a flower design on the reflective coating , a feature necessary for achieving the proper optical conditions in the magneto - optical traps .
numerical estimates show that we should be able to collect small but useful numbers of atoms in these traps , providing a simple way to feed arrays of devices on future chips .
this represents an important integration step because atom chips currently have to be loaded from a single large cloud by a series of somewhat involved manoeuvres .
s. eriksson , m. trupke , h. f. powell , d. sahagun , c. d. j. sinclair , e. a. curtis , b. e. sauer , e. a. hinds , z. moktadir , c. o. gollasch , and m. kraft , `` integrated optical components on atom chips '' , _ eur .
phys j. d _ * 35 * , pp .
135 - 139 ( 2005 ) . c. d. j. sinclair , e. a. curtis , i. llorente - garcia , j. a. retter , b. v. hall , s. eriksson , b. e. sauer and e. a. hinds , `` bose - einstein condensation on a permanent - magnet atom chip '' , _ phys .
a _ * 72 * , 03160(r ) ( 2005 ) s. knappe , p. d. d. schwindt , v. shah , l. hollberg , j. kitching , l. liew , and j. moreland , `` a chip - scale atomic clock based on rb-87 with improved frequency stability '' , _ opt .
* 13 * , pp . 1249 - 1253 , ( 2005 ) .
y. shin , c. sanner , g .- b .
jo , t. a pasquini , m. saba , w. ketterle , d. e. pritchard , m. vengalattore and m. prentiss , `` interference of bose - einstein condensates split with an atom chip '' , _ phys . rev . a _ * 72 * , 2 ( 2005 ) .
t. schumm , s. hofferberth , l. m. andersson , s. wildermuth , s. groth , i. bar - joseph , j. schmiedmayer and p. krger , `` matter - wave interferometry in a double well on an atom chip '' , _ nat .
phys . _ * 1 * , pp 57 - 62 ( 2005 ) .
m. trupke , f. ramirez - martinez , e. a. curtis , j. p. ashmore , s. eriksson , e. a. hinds , z. moktadir , c. gollasch , m. kraft , g. v. prakash , and j. j. baumberg , `` pyramidal micromirrors for microsystems and atom chips '' , _ appl .
lett . _ * 88 * , 071116 , ( 2006 ) .
e. koukharenko , z. moktadir , m. kraft , m. e. abdelsalam , d. m. bagnall , c. vale , m. p. a. jones and e. a. hinds , `` microfabrication of gold wires for atom guides '' , _ sens .
actuators a _ * 115 * 2 - 3 , pp 600 - 607 ( 2004 ) .
m. trupke , j. goldwin , b. darquie , g. dutier , s. eriksson , j. ashmore , and e. a. hinds , `` atom detection and photon production in a scalable , open , microcavity '' , _ phys .
_ * 99 * , 063601 , ( 2007 ) . | ultra - cold atoms can be manipulated using microfabricated devices known as atom chips .
these have significant potential for applications in sensing , metrology and quantum information processing .
to date , the chips are loaded by transfer of atoms from an external , macroscopic magneto - optical trap ( mot ) into microscopic traps on the chip .
this transfer involves a series of steps , which complicate the experimental procedure and lead to atom losses . in this paper
we present a design for integrating a mot into a silicon wafer by combining a concave pyramidal mirror with a square wire loop .
we describe how an array of such traps has been fabricated and we present magnetic , thermal and optical properties of the chip .
atom chips , electrophoretic resist , magneto - optical traps , cavity patterning . |
quantum dots ( qds ) coupled to source and drain electrodes represent versatile and well - controlled systems for the study of mesoscopic transport @xcite .
the many aspects of electron tunneling through qds are typically studied by measuring either the qd _ conductance _ , or the qd _
charge occupation_. there are several techniques available for measuring the qd charge occupation , among them direct capacitance measurements @xcite , and the use of single - electron transistor @xcite and quantum point - contact @xcite ( qpc ) electrometers .
more insight can be gained when _ combining _ charge and conductance measurements , and extracting information from both of them to obtain a more complete picture of the system . in the sequential tunneling regime of the qd ,
combined charge and conductance measurements can be used to determine the system timescales @xcite , whereas in a strong - coupling regime , such measurements reveal the effect of kondo correlations on the charge of a qd @xcite . in this paper , we present combined transport and qpc charge detection measurements in the cotunneling regime of a qd @xcite .
cotunneling is a second - order transmission process through the qd dominating when first - order tunneling is energetically forbidden due to coulomb blockade .
both elastic @xcite and inelastic @xcite versions of this process have been studied experimentally . in particular , their coherence properties have been of interest @xcite . in that context , the general presumption is that environmental decoherence should be weak due to the short duration of cotunneling . the short duration is in turn explained by the large degree of energy uncertainty of the order of @xmath1 , the blockade energy of first - order tunneling .
a direct measurement of the cotunneling time is , however , not available .
the present paper is based on the idea that the cotunneling time must affect the occupation of the qd , and thus the qpc charge detection measurement should provide information about it . based on this indirect approach , we test the assumption that the cotunneling time is limited by @xmath0 .
our method is a quantitative comparison of the measured differential qd charge in the inelastic cotunneling regime with an estimate based on the measured qd current and a carrier dwell time of @xmath0 .
we find that the measured signal is of the same order of magnitude as the estimate , consistent with a cotunneling time bounded by the time @xmath0 .
we contrast these results with measurements of the qd charge in a regime where inelastic cotunneling is accompanied by sequential tunneling ( cotunneling - assisted sequential tunneling @xcite , cast ) .
the charge signal in this regime is significantly larger than what would be expected assuming a tunneling dwell time of @xmath0 .
we attribute this to the comparatively long dwell time of sequential tunneling events .
in addition to inelastic cotunneling occurring at nonzero source drain voltage , we study resonant tunneling at zero source drain voltage @xcite .
the finite nonactivated qd conductance in the tails of a resonant peak ( in the coulomb - blockade valley ) is termed elastic cotunneling . in this regime ,
the direct current through the qd vanishes and we demonstrate that the main contribution to the qd charge is due to its equilibrium occupation , unlike in the inelastic regime where the time - averaged charge is mainly due to the dwell time of transmitted carriers .
the differential qd conductance and the differential qpc signal are then both a probe of the spectral density of the qd state and are found to agree over two orders of magnitude . , [ fig : ps_figure_hicoupling ] , [ fig : ps_figure_inelcotunneling ] , [ fig : ps_figure_directcurrentandtc ] .
sample b : figure [ fig : ps_figure_inco2 ] . )
the dark parts correspond to the non - depleted parts of a 2deg buried in a ga[al]as heterostructure . in both samples ,
a qd is formed using the schottky gates g1 , g2 , and g3 , ( upper half of the image ) and is electrically separated from a qpc charge - readout circuit ( lower half ) by an oxide line . on sample
a , the readout qpc is formed between the metal gate g4 and the oxide line , whereas on sample b the qpc is formed by a second oxide line .
the gray shaded metal gates marked ` x ' have not been used . ]
the experiments were done on two different qd samples shown in fig .
[ fig : ps_figure_intro ] .
they were fabricated with a combination of electron - beam and scanning - probe lithography @xcite on a @xmath2 heterostructure containing a two - dimensional electron gas ( 2deg ) @xmath3 below the surface ( density @xmath4 , mobility @xmath5 at @xmath6 ) . in both samples ,
negative bias voltages on the schottky gates g1 , g2 , and g3 define the quantum dot with a charging energy @xmath7 of around @xmath8 and a typical single - particle level spacing @xmath9 of @xmath10 .
the measurements were done in a @xmath11he/@xmath12he dilution refrigerator with a base temperature of @xmath13 .
a bias voltage @xmath14 was applied symmectically between the source and drain leads of the qd , and the differential conductance @xmath15 was measured with standard lock - in technique .
the charge signal of the qpc was measured via the transconductance @xmath16 at a second lock - in frequency . to this end ,
the qpc was biased with a constant source drain voltage of 500 to @xmath17 , and the voltage on the qd gate g2 was modulated with small amplitude ( @xmath18 or less ) @xcite .
lock - in integration time constants ranging from 0.3 to @xmath19 have been used . in order to optimize the signal strength of the charge detector ,
the qpc conductance was tuned to a value of around @xmath20 using gate g4 .
( black ) in the regime of weak dot lead coupling along with a fit ( red ) to a fermi lorentz convolution ( cf .
main text ; fit parameters @xmath21 , @xmath22 ) .
( b ) black solid curve : same data as in ( a ) .
the blue dotted curve is the transconductance signal measured with the qpc .
both data sets have been smoothened over a range of @xmath23 ( 5 data points ) . ]
( black ) in the regime of strong dot lead coupling along with a fit ( red ) to a fermi lorentz convolution ( cf .
main text ; fit parameters @xmath24 , whereas the parameter @xmath21 was held fixed ) .
( b ) black solid curve : same data as in ( a ) .
the blue dotted curve is the transconductance signal measured with the qpc .
both data sets have been smoothened over a range of @xmath25 ( 20 data points ) . ]
the coupling strength of the qd state to the leads , @xmath26 to source , and @xmath27 to drain , is controlled with the gates g1 and g3 .
the lifetime broadening of the qd state @xmath28 can be continuously tuned from below to above the thermal energy @xmath29 corresponding to the temperature of the electrons in the leads @xcite .
the larger of the two energy scales determines the line width of the conductance resonances of the qd .
the limiting regimes are characterized as thermally activated single - level transport in case @xmath30 , and as a breit wigner transmission resonance in case @xmath31 .
we experimentally approach the two regimes in the measurements shown in figs .
[ fig : ps_figure_lowcoupling ] and [ fig : ps_figure_hicoupling ] . in the first case ,
the shape of a peak in the qd conductance at zero @xmath14 is given by @xcite @xmath32 where @xmath33 is the difference between the electrochemical potential of the nondegenerate qd level ( @xmath34 ) and that of the leads ( @xmath35 , @xmath36 ) .
in the second case , tunneling through the qd is well described as a double - barrier scattering process of independent particles , and the conductance peak takes the lorentzian form of the corresponding transmission probability , @xmath37 when increasing the dot lead coupling further , kondo correlations emerge which render the single - particle approximation and equation ( [ eq : lorentz_formula ] ) invalid . in the measurements presented here ,
the coupling strength was kept well below values at which these correlations are typically observed . in order to describe the line shape of @xmath38 in the intermediate regime , where the energy scales @xmath39 and @xmath40 are comparable
, it is often a good approach to use a convolution @xcite of equation ( [ eq : lorentz_formula ] ) with the energy derivative of a fermi dirac distribution function . in fig .
[ fig : ps_figure_lowcoupling](a ) , we plot a qd - conductance peak in the weak - coupling regime , along with a fit to such a convolution ( fitting parameters @xmath41 and @xmath42 corresponding to a temperature of about @xmath43 ) .
figure [ fig : ps_figure_hicoupling](a ) shows a measurement at stronger coupling , where the lifetime broadening of the conductance peak exceeds thermal broadening ( fitting parameter @xmath44 , while @xmath42 was fixed to the value found in the weak - coupling case ) . the small asymmetry with respect to @xmath45
is mainly caused by the overlap with the next coulomb blockade peak towards higher gate voltages . in both regimes ,
the qpc transconductance signal was measured simultaneously with the qd conductance . in figs .
[ fig : ps_figure_lowcoupling](b ) and [ fig : ps_figure_hicoupling](b ) , we plot the two signals on top of each other for comparison .
the scaling of the vertical axes is chosen such to achieve an optimal overlap of the curves .
indeed , qd conductance and qpc transconductance match well over the covered range of signal strength , about two orders of magnitude .
as the measurements are done at zero ( direct ) source drain voltage , the qd is in thermal equilibrium with its leads . in the idealized weak - coupling case ,
the qd level has negligible width compared to @xmath29 .
the time - averaged occupation number @xmath46 of the qd is then determined by the fermi
dirac distribution of the electrons in the leads , @xmath47 $ ] , up to an integer offset @xmath48 of electrons on the dot . assuming a constant gate lever arm @xmath49 , the qpc signal is then given by @xmath50 , where @xmath51 is the sensitivity of the qpc current to a qd occupation change of 1 electron .
the derivative @xmath52 , and thus the qpc signal , is proportional to equation ( [ eq : fermi_formula ] ) @xcite . in the strong - coupling case ,
the occupation is determined by the spectral density of the qd state which is lorentzian , and thus the qpc signal is also expected to exhibit the same line shape as the qd conductance @xcite .
we note that our single - level transport situation is different from that of a multi - level or even metallic dot . in that case , neither equation ( [ eq : fermi_formula ] ) , nor equation ( [ eq : lorentz_formula ] ) , nor their convolution accurately describes the measurement of the qd charge as demonstrated in ref .
we tested the validity of the single - level transport assumption by measuring the temperature dependence of our data ( not shown ) .
conduction of a qd in the tails of a lifetime - broadened peak is due to elastic cotunneling , a second - order process already present close to equilibrium .
as the qd is driven out of equilibrium by a source drain voltage exceeding the energy @xmath9 of the first excited qd state , additional processes come into play .
these are called inelastic cotunneling processes and bear a close analogy to their elastic counterparts .
during such a process , as depicted in figure [ fig : ps_figure_sketches](a c ) , an electron tunnels out to the drain from the ground state , while a second electron tunnels into the excited state from the source .
the qd spends a short time in a virtual intermediate state whose energy lies outside of the classically allowed range .
the sequence of the partial tunneling processes can be interchanged , which gives rise to two channels ( hole - like and electron - like ) which both contribute to the total cotunneling amplitude . in the electron - like sequence ,
figure [ fig : ps_figure_sketches](b ) , tunneling from the source into the dot occurs first , and the virtual intermediate state is an @xmath53-electron state . in the hole - like sequence ,
figure [ fig : ps_figure_sketches](c ) , tunneling from the dot to the drain occurs first , and the virtual intermediate state is an @xmath54-electron state .
the classical blockade energies of the two channels are different in general , as is specified in the figure caption .
the overall transmission probability from source to drain , and thus the cotunneling current , depends on the detuning of the qd potential with respect to the leads , but the minimum source drain voltage is independent of detuning and given by @xmath55 .
the experimental signature of inelastic cotunneling is hence a conductance step inside a coulomb blockade diamond parallel to the line of zero bias @xcite .
such a feature is seen in the @xmath38 data shown in figure [ fig : ps_figure_inelcotunneling](a ) at a positive source
drain voltage of around @xmath56 . in the simultaneous measurement of the qpc transconductance shown in panel ( b ) ,
no finite - bias feature is visible .
-electron ground state configuration and nomenclature .
( b , c ) if the qd source drain voltage @xmath57 exceeds @xmath55 , the qd can be brought to an excited state in a cotunneling process transferring one electron from source to drain .
two cotunneling channels contribute .
the electron - like channel ( b ) consists of tunneling from the source into the dot followed by tunneling from the dot to the drain .
the virtual intermediate state is an @xmath53-electron state .
the blockade energy for this processe is at least @xmath58 .
the hole - like channel ( c ) consists of tunneling from the dot to the drain followed by tunneling from the source into the dot .
the virtual intermediate state is an @xmath54-electron state .
the blockade energy for this process is at least @xmath59 .
( d ) after cotunneling , the dot is left in an @xmath48-electron excited state .
subsequently , the qd can relax by emission of a phonon or photon ( e ) , or in a cotunneling process involving a single lead ( f ) . in the sketch ,
the excited - state energy @xmath60 lies above the drain level @xmath36 , in which case the electron can tunnel elastically to the drain ( g ) . ]
the discrepancy between the two signals becomes more evident when looking at the measurement in figure [ fig : ps_figure_inelcotunneling](c ) taken along cuts at four source drain voltages , as indicated by the white arrows in panel ( a ) . at @xmath61 ,
the peak in the qd conductance is due to elastic cotunneling as discussed in sec .
[ sec : lohicoupling ] , with a broadening of @xmath62 . as expected ,
the two signals @xmath38 and @xmath63 can be made to fit by scaling . upon increasing @xmath14 , both the peaks in @xmath38 and in @xmath63
broaden and eventually split .
it is then not expected that the two signals match over the whole range in gate voltage .
namely , the relative height of the two sub - peaks in @xmath63 can take any value depending on the coupling symmetry @xcite @xmath64 .
in contrast , the two sub - peaks in @xmath38 are expected to be equal in height independent of symmetry in case of single - level transport with energy - independent coupling .
the difference in height of the @xmath38 peaks in our data can be explained by energy - dependent tunneling rates @xmath65 . despite the overall discrepancies between @xmath38 and @xmath63 signals , by tuning the coupling asymmetry we can achieve that they agree in the tails ( after vertical scaling ) as indicated by the dashed ellipses in figure [ fig : ps_figure_inelcotunneling](c ) .
this is expected because in the limit @xmath66 , the zero - bias results for @xmath38 and @xmath63 are valid , which implies that both quantities decay proportionally to @xmath67 . .
an onset of inelastic cotunneling is visible at @xmath68 .
( b ) simultaneous measurement of @xmath63 . in these data ,
no inealstic - cotunneling onset is visible , but instead a signature ( white arrow ) of the excited dot state responsible for inelastic cotunneling .
( c ) qpc and qd signals measured along @xmath69 at four different @xmath14 indicated by arrows in ( a ) .
the two signals at zero bias are scaled to match ; this scaling is maintained for the rest of the curves , which are also vertically offset for clarity . for finite bias voltages @xmath70 and @xmath71 , the two signals still agree in the tails as indicated by the dashed ellipses . beyond the inelastic cotunneling onset ( at @xmath68 ) ,
the signals clearly deviate . ] as the bias is increased above the energy of the excited state , @xmath68 , the signals clearly deviate in the right - hand tail , as indicated by the double arrow in figure [ fig : ps_figure_inelcotunneling](c ) . at this point ,
the qd conductance is clearly enhanced compared to the occupation signal and its maximum shifts to the right into the coulomb - blockaded region .
this indicates that the extra conductance is not due to sequential tunneling through the excited state : in that case , we would instead expect a conductance feature shifting to the left into the bias window .
we therefore assume that the extra conductance is primarily due to inelastic cotunneling .
the qualitative difference between the data at @xmath61 and @xmath68 demonstrates the contrasting character of the measured qd charge in the two cotunneling regimes . at zero qd voltage
, the charge is entirely characterized as an equilibrium quantity .
it is determined by the spectral density of the qd state and does not explicitly depend on the presence of two leads , i.e. , does not depend on transport .
the proportionality of the qpc signal to the qd conductance observed at low @xmath14 is a consequence of the fact that both are determined by the spectral density of the qd state .
this simple relation is lost when going out of equilibrium , in particular in the regime of inelastic cotunneling .
but in this regime , we can identify a different connection between the qd charge state and transport .
namely , the only property distinguishing a qd slightly below the inelastic - cotunneling onset from the same qd slightly above the onset is the fact that in the latter case , processes of the kind shown in figure [ fig : ps_figure_sketches](b , c ) are energetically allowed . because an electron is transferred through the qd in each such process , any change in the qd charge occurring at the position of the inelastic - cotunneling onset in @xmath38 is most likely linked to transport . in order to be more quantitative
, we aim at linking the direct qd current , @xmath72 , to the observed qd charge state .
we define the transport occupation of the qd as the frequency of carriers passing the qd multiplied by the average dwell time of each carrier on the qd , @xmath73 .
the sign depends on whether the current is carried by electrons or holes . in the general case
, the current @xmath72 would have to be split into an electron and a hole current , and the dwell times of the two carrier types could be different .
this corresponds to the two parallel cotunneling channels discussed in the beginning of this section .
we will only look at cases where either of the two carrier types dominates , so we can ignore this complication .
the transport occupation @xmath74 is in general not equal to the total occupation @xmath75 which is relevant for the measurement .
the two are equal under the conditions that the current is constituted of a single type of tunneling processes ( all with a similar dwell time ) , and that tunneling processes are absent which contribute to the occupation but not to the current . in case
these conditions are met , the measured qpc signal can be expressed as @xmath76 where @xmath51 is the sensitivity of the qpc current to an occupation change of one electron . measured at different qd source drain voltages @xmath77 in the configuration of figure [ fig : ps_figure_inelcotunneling ] .
the equilibrium contribution to @xmath63 at @xmath78 has been subtracted from all curves .
the dashed lines in ( a ) show the theoretical qpc transconductance signal calculated from @xmath72 and a particle dwell time @xmath79 , equation ( [ eq : qpc - tc_calculated ] ) . the dashed lines in ( b ) show the corresponding calculation when assuming an energy - independent dwell time @xmath80 .
all data have been smoothened over a range of @xmath81 in @xmath69 . ]
figure [ fig : ps_figure_directcurrentandtc ] shows charge sensing data for @xmath14 values around the inelastic - cotunneling onset .
they belong to the same measurement as those in figure [ fig : ps_figure_inelcotunneling](c ) . but these traces have a smaller @xmath69 range restricted to the region in which the inelastic - cotunneling contributions to @xmath38 and @xmath63 ( at bias voltages just above @xmath82 ) clearly dominate over the remaining elastic contributions ( just below @xmath82 ) . the solid traces show the measured @xmath63 minus a small contribution at @xmath61 . in the case of an inelastic cotunneling process , the dwell time corresponds to the lifetime of the intermediate virtual qd state .
near the cotunneling onset , its energy lies outside the classically allowed range by an amount @xmath83 .
the lifetime of a virtual state in a setup like this is often treated using an energy - time uncertainty relation in the spirit of heisenberg @xcite .
the heuristic relation @xmath84 involves the quantum uncertainty @xmath85 of the energy of a state during a process and the duration @xmath86 of that process . in the context of cotunneling , it means that if an electron resides in the dot for a short time @xmath87 , its energy is necessarily dispersed enough that the difference @xmath1 between dot and lead state energies becomes irrelevant and the amplitude for a tunneling process through the dot becomes nonzero .
the intermediate state has a dispersed energy because it is a superposition between two states with rather sharp energies 0 and @xmath1 .
a dwell time on the dot exceeding @xmath0 on the other hand would mean that the energy of the intermediate state is well - defined at the value @xmath34 , which would violate energy conservation because the initial and final state energy must lie within the bias window .
we note , however , that the above energy - time relation is not strictly a version of the usual heisenberg uncertainty principle of noncommuting operators , such as position and momentum .
this is because time , unlike energy , is not a quantum operator . to compare our data with this theoretical picture
, we insert the value @xmath88 into equation ( [ eq : qpc - tc_calculated ] ) to calculate @xmath89 which we compare with the measurement . the dashed lines in figure [ fig : ps_figure_directcurrentandtc](a )
show the result of equation ( [ eq : qpc - tc_calculated ] ) .
indeed , measured and calculated charge signals are of similar magnitude , and apart from noise the measured signal is always smaller than the calculation , indicating a dwell time bounded by @xmath0
. note that in case of a _ resonant _ process ( on - peak ) , instead one typically assumes a constant dwell time @xmath90 equal to the lifetime of the qd state @xcite .
the cotunneling picture then loses its validity , which can also be recognized from the fact that the time @xmath0 diverges when @xmath1 tend to zero .
namely , we have justified this expression for @xmath91 by recognizing that the energy uncertainty of the quantum state during tunneling is determined by the blockade energy @xmath1 .
if @xmath1 tends to zero , the energy uncertainty will eventually be limited by the intrinsic width @xmath92 of the dot states participating in tunneling , and we enter the regime of resonant tunneling .
assuming a constant dwell time , as is done in the case of resonant tunneling , is problematic in the case of cotunnrling however , since a constant dwell time violates the heisenberg relation off - peak . nevertheless , it is possible to achieve a good description of the transconductance data by assuming a constant dwell time as seen in figure [ fig : ps_figure_directcurrentandtc](b ) .
the value of @xmath93 we used there was chosen such to achieve a good fit of the data .
there is no further physical justification for it , but the plots may help to judge about the comparison of measurement and calculation in the left column .
the fact that the data can be reproduced using a constant dwell time means that the main energy dependence of @xmath63 comes from the energy dependence of @xmath72 .
it is generally not surprising that in the constant - time case , data and calculation agree better than in the heisenberg time case . when assuming a constant dwell time , there is one free parameter ( the constant value of @xmath94 ) , but there is none when assuming @xmath88 .
the comparison between the two cases tells us that the heisenberg time @xmath0 is probably not a good approximation for the actual dwell time , but it appears that its role as an upper limit to the dwell time is relevant for the measurement .
the validity of the picture leading to equation ( [ eq : qpc - tc_calculated ] ) depends , as mentioned , on the additional processes taking place after the initial inelastic cotunneling event .
due to the variety of conceivable tunneling and relaxation processes at large qd bias , it is necessary to discuss their possible influence on the data in figures [ fig : ps_figure_directcurrentandtc ] and [ fig : ps_figure_inelcotunneling ] .
figure [ fig : ps_figure_sketches ] ( e ) through ( g ) shows the three main options for the continuation of the qd evolution after an inelastic cotunneling event ( panel a ) has brought it to the excited state . all these options eventually bring the qd back to the ground state and thus close the transport cycle . in the simplest case sketched in panel ( c ) ,
the electron relaxes to the ground state through emission of a phonon @xcite , a process which neither contributes to the qd current nor changes its charge state .
a typical phonon emission rate @xcite in gaas qds is @xmath95 .
alternatively , the qd may relax through another inelastic cotunneling process involving two , or also just one lead as shown in figure [ fig : ps_figure_sketches](f ) for the case of the source lead .
all these processes contribute to the occupation of the qd .
finally , sequential tunneling to the drain lead may occur if the excited - state energy @xmath60 lies above the drain potential @xmath36 . in the absence of a pronounced feature in our data at the zero - crossing of @xmath96 ( the continuation of the dashed line in figure [ fig : ps_figure_inelcotunneling][a ] ) ,
we conclude that such processes do nt contribute significantly to either occupation or current ( possibly due to a poor coupling of the excited state to the drain ) .
what remains are thus relaxation processes as shown in panel ( d ) which compete with phonon emission .
an estimate of the inelastic cotunneling relaxation rate @xcite requires the knowledge of the coupling of both ground and excited state to the source reservoir .
while the ground state coupling of around @xmath97 can be reliably inferred from finite - bias charge sensing @xcite , determining the excited - state coupling is more difficult . from the clear charging feature at the entrance of the excited state in the bias window ( arrow in figure [ fig : ps_figure_inelcotunneling][b ] ) , we conclude that the excited - state coupling must be at least comparable to the ground - state coupling .
this would lead to an inelastic - cotunneling relaxation rate of at least 6 to @xmath97 in the gate voltage range of figure [ fig : ps_figure_directcurrentandtc ] , which thus dominates over phonon emission .
the consequence is an effective dwell time per transport cycle which includes the dwell time of the hole during the relaxation process , on top of the dwell time for the initial cotunneling process . since the blockade energy @xmath1 is the same for both processes , the bound @xmath0 on the dwell times is identical .
the character of the transport process can be changed strongly in case that not the relaxation processes of the kind shown in figure [ fig : ps_figure_sketches](e , f ) , but instead the sequential process shown in figure [ fig : ps_figure_sketches](g ) is dominant . such a sequential process leaves the qd in a state with @xmath98 electrons .
another electron can then tunnel elastically from source into the excited state , and this sequential cycle may be repeated several times , until relaxation or tunneling into the @xmath48-electron ground state takes place .
this effect is called cotunneling - assisted sequential tunneling ( cast ) @xcite . .
an onset of inelastic cotunneling is visible at @xmath99 .
in addition , a signature of cast is visible as the extension of the dashed line towards negative @xmath14 .
( b ) simultaneous measurement of @xmath63 .
( c ) solid lines : qpc signal @xmath63 measured along @xmath69 at three different @xmath14 indicated by arrows in ( a ) .
the signal was integrated longer than in the diamond measurement to reveal the faint cast signature in the measurement at @xmath100 . dashed lines : theoretical qpc transconductance signal calculated from @xmath72 and the particle
dwell time @xmath101 , equation ( [ eq : qpc - tc_calculated ] ) .
the measured signal clearly exceeds the calculated one due to the long dwell times of the sequential tunneling processes occurring in this regime .
this is in contrast to the case of pure inelastic cotunneling ( figure [ fig : ps_figure_directcurrentandtc ] ) .
smoothened over a range of @xmath102 in @xmath69.[fig : ps_figure_inco2 ] ] there is a clear experimental signature for cast : a conductance step inside a coulomb blockade diamond parallel to the diamond edge @xcite . this line borders the region in which cast is energetically allowed .
the conditions to observe such a feature are special .
the qd needs to be in a configuration featuring an excited state that is more strongly coupled to source and drain than the ground state , but weakly enough to maintain sufficient energy resolution .
furthermore , relaxation from the excited to the ground state needs to be slow . albeit only partially under experimental control , these conditions vary randomly from one coulomb - blockade diamond to the next due to the quantum nature of the dot states .
this allows for a systematic search of cast features in measurement . in the coulomb - blockade diamond shown in figure [ fig : ps_figure_inco2](a )
, such a feature is visible as the continuation of the dashed line towards negative @xmath14 .
neither inelastic - cotunneling nor cast onsets are visible in the simultaneous measurement of @xmath63 in panel ( b ) .
more accurate line cuts in figure [ fig : ps_figure_inco2](c ) reveal , however , that @xmath63 is non - zero in the cotunneling regime . above the inelastic cotunneling onset at @xmath100 , a qpc signal of the order of @xmath103 builds up inside the region where cast is allowed ( for @xmath104 ) .
the particle dwell time in a sequential process is associated with a real , rather than virtual , intermediate state and as such is not limited by the heisenberg relation as it was for a cotunneling process . to emphasize this difference , we plot as dashed lines in figure [ fig : ps_figure_inco2](c ) the result of equation ( [ eq : qpc - tc_calculated ] ) , taking into account the direct qd current and the cotunneling dwell time @xmath0 .
the calculated trace at a bias above the cotunneling onset ( @xmath105 ) exhibits a clear enhancement of the expected charge signal compared to the traces below the onset .
this is a qualitative feature shared with the measured trace .
quantitatively the two disagree like in the case of pure cotunneling , but unlike there the measured signal is stronger than the calculated one , signaling a dwell time exceeding @xmath0 .
in summary , we have presented simultaneous measurements of the conductance and charge occupation of a qd in several parameter regimes . at zero qd voltage
, the charge signal is successfully interpreted in terms of the equilibrium occupation and the equilibrium conductance of the qd .
both in the regimes of weak coupling ( dominated by thermal broadening ) and of strong coupling ( dominated by lifetime broadening ) , conductance and charge signal lineshapes are found to agree . at nonzero qd voltage ,
we study the inelastic cotunneling regime where the charge signal is generally weaker , and where an analysis in terms of equilibrium occupation is not possible .
we compare the charge signal to a theoretical signal calculated from the qd current and a charge carrier dwell time of @xmath0 , where @xmath1 is the blockade energy of first - order tunneling .
such a dwell time estimate is often derived from a cotunneling picture involving a virtual intermediate state with a lifetime bounded by an energy - time uncertainty principle . since
our measured charge occupation is smaller than the calculated one , our results support this cotunneling picture experimentally .
assuming an energy - independent dwell time allows for a good fit of the data in the accessible range of energies @xmath1 .
further measurements carried out in a regime of mixed sequential tunneling and inelastic cotunneling are consistent with the above interpretation .
the charge occupation signal observed in this regime clearly exceeds the calculation based on a dwell time of @xmath0 .
this is expected because of the longer carrier dwell time in sequential tunneling as compared to that in cotunneling .
we thank c. beenakker , k. kobayashi , l. levitov , m. sanquer , g. schn , and s.e .
ulloa for discussions .
we also thank m. bttiker for clarifying the role of the time uncertainty in a tunneling process , and y. gefen for insightful discussions from the point of view of weak measurements . sample growth and processing
was mainly carried out at first laboratory , eth zurich . financial support from the swiss national science foundation ( schweizerischer nationalfonds , national center for competence in research , quantum science and technology ) is gratefully acknowledged . | we present comparative measurements of the charge occupation and conductance of a gaas / algaas quantum dot .
the dot charge is measured with a capacitively coupled quantum point contact sensor . in the single - level coulomb blockade regime near equilibrium ,
charge and conductance signals are found to be proportional to each other .
we conclude that in this regime , the two signals give equivalent information about the quantum dot system . out of equilibrium
, we study the inelastic - cotunneling regime .
we compare the measured differential dot charge with an estimate assuming a dwell time of transmitted carriers on the dot given by @xmath0 , where @xmath1 is the blockade energy of first - order tunneling .
the measured signal is of a similar magnitude as the estimate , compatible with a picture of cotunneling as transmission through a virtual intermediate state with a short lifetime . |
laser spectroscopy of optical transitions in highly charged ions ( hcis ) is a subject of considerable interest as it provides access to relativistic effects in few - electron systems and can be used to test bound - state qed in the extremely strong electric and magnetic fields in the vicinity of the ionic nucleus @xcite .
experimentally , such magnetic dipole ( m1 ) transitions in mid-@xmath1 hcis have first been studied in electron - beam ion traps ( ebits ) by laser excitation and fluorescence detection @xcite , yielding a relative accuracy of a few ppm for the determination of the wavelength .
direct laser spectroscopy of heavy ( high-@xmath1 ) hcis has so far only been performed at the experimental storage ring esr on hydrogen - like bismuth @xmath2bi@xmath3 @xcite and lead @xmath4pb@xmath5 @xcite . in both cases , the transition between the ground state hyperfine levels was induced by pulsed lasers and resonance fluorescence was recorded .
these investigations have been extended to the ground - state hyperfine transition in lithium - like bismuth @xmath2bi@xmath6 , which has recently been observed in the experimental storage ring ( esr ) @xcite .
this measurement in combination with the measurement on hydrogen - like bismuth will allow the first determination of the so - called specific difference between the hyperfine splittings @xmath7 as suggested by shabaev and co - workers @xcite .
the first observation of the transition in @xmath2bi@xmath6 is an important step , but it will not provide sufficient accuracy for a high - precision determination of the qed effects in the specific difference , since the wavelength determination for both transitions ( h - like and li - like ) is still limited in accuracy due to the large doppler width and the uncertainty of additional doppler shifts caused by the relativistic ion motion in the storage ring .
this will be considerably improved once high-@xmath1 highly charged ions are available at rest in a clean environment allowing for high - accuracy laser spectroscopy . to this end , the spectrap experiment has been designed @xcite .
it is part of the highly charged heavy ion trap ( hitrap ) project @xcite at the gsi helmholtzzentrum darmstadt , which will provide hcis up to u@xmath8 at low energies suitable for capture into a penning trap .
the precision achieved in the laser spectroscopy of trapped ions crucially depends on the width of the optical transition of interest and the mechanisms that lead to additional broadening , e.g. doppler broadening .
the study of forbidden transitions with high accuracy requires the elimination of doppler broadening .
this can be achieved by first - order doppler - free techniques like two - photon transitions or by the trapping and cooling of atoms or ions .
there is a variety of corresponding methods for the cooling of the ion motion , for a detailed overview see e.g. @xcite .
the evaporative cooling of hcis in an ebit has been used for the laser spectroscopy of ar@xmath9 @xcite and recently in a penning trap on hcis that were produced in an ebit and then transported and re - trapped in a penning trap @xcite . at spectrap
we make use of resistive cooling @xcite and laser cooling @xcite .
the former is a very effective cooling mechanism for hcis , while the latter is most effective for ions with a level scheme suitable for laser cooling such as be@xmath10 or mg@xmath10 .
laser - cooled ions can then be used for sympathetic cooling @xcite of simultaneously trapped hcis .
such experiments have so far been performed with be@xmath10 in a penning trap @xcite and are foreseen in a paul trap @xcite . here , we present first studies with laser - cooled mg@xmath10 ions in the spectrap penning trap .
we have performed systematic measurements with externally produced mg ions which have been captured in flight and stored .
the observation of laser - induced fluorescence ( lif ) down to the single - ion level allows a determination of the ion storage time , ion number and ion temperature .
evidence for the formation of ion crystals has been observed .
these measurements represent an initial characterization and optimization of the system as an important step towards the sympathetic cooling and precision laser spectroscopy of highly charged ions .
penning traps are well - established tools for capture and confinement of externally produced ions .
a static homogeneous magnetic field ensures radial confinement , while the electrode arrangement produces an electrostatic potential well which provides axial confinement of charged particles .
ions can thus be localized , which allows laser irradiation and fluorescence detection under well - controlled conditions .
stored ions can be motionally cooled to reduce the doppler broadening of transition lines to well below the ghz level .
the achievable storage time is fundamentally limited only by the residual gas pressure inside the trap , since collisions with gas particles may lead to ion loss .
typical storage times range from seconds to minutes , but also storage times of several months have been achieved @xcite . hence , also slow transitions like magnetic dipole ( m1 ) transitions can be investigated with high resolution and statistics .
such traps have been realized in numerous variations especially concerning their geometry , for details see @xcite . for the purposes of laser spectroscopy ,
trap geometries need to be chosen such that they allow both ions and light to enter and leave the trap suitably , as well as to provide the means for observing the fluorescence .
the spectrap experiment employs a five - pole cylindrical penning trap with open endcaps @xcite , with an additional pair of capture electrodes , as described in detail in @xcite .
the geometry is chosen such that the trap is orthogonal , i.e. the trapping potential depth is independent from the choice of correction voltages used to make the trapping potential harmonic close to the trap centre .
the ion motion in such a trap has been discussed in detail in e.g. @xcite .
the open endcaps and capture electrodes yield axial access to the trap from both sides . in our case ,
the ions enter from the top and the cooling laser from below , as shown in fig .
[ fig : traplaser ] .
the capture of externally produced ions is achieved by fast switching of trap voltages .
the ring electrode is radially split into four segments to allow the use of a rotating wall @xcite for ion cloud compression and shaping .
a central hole in each ring segment enables the detection of the stored ion fluorescence on radially positioned detectors outside the magnet vessel .
the fluorescence light emerging out of the holes is collimated by plano - convex lenses . the geometrical light collection efficiency of this system is the main limiting factor of the total fluorescence detection efficiency .
also , reflection and absorption in the lens and the vacuum windows as well as misalignments of the main optical axis reduce the signal . at the wavelength used for laser cooling of mg@xmath11
, the detection efficiency was measured to be about @xmath12 .
the trap is installed in a vertical , cold - bore , superconducting magnet with helmholtz configuration , such that direct optical access to the trap centre is possible through four radial ports in the horizontal plane . before it was consigned to gsi
, the magnet was used for a similar experiment ( retrap at lawrence livermore national laboratory ) , with a slightly different penning trap configuration and radially cooled @xmath13 ions @xcite .
the magnetic field in the trap centre can be set to any value up to 6 t and provides a relative central homogeneity of @xmath14 over a region of 2.5 cm .
a liquid helium cryostat is used for cooling both the superconducting solenoids and the trap with its attached electronics .
the residual gas pressure in the vacuum system is monitored in the room temperature region at the bottom of the magnet vessel , and typically amounts to @xmath15 mbar during magnet operation .
there is no direct separation between the trap and the insulation vacuum of the cryostat , so additional cryopumping of the volume inside the trap is provided by the cold surfaces .
hence , the vacuum conditions inside the trap can be assumed to be much better than indicated by the gauge , as will be discussed below .
laser beams are guided into the trap along the central vertical axis , from a laser laboratory located under the superconducting magnet setup , as shown in fig.[fig : trapsetup ] .
the fluorescence light is detected by a channel photo multiplier attached to the outside of the magnet vessel .
it has a quantum efficiency of 18% and a very low dark count of some 20 hz .
it is well suited for detection of uv light between 200 and 400 nm . because of its sensitivity to the stray magnetic field
it was mounted in a magnetically shielded housing about 1 metre away from the main magnet chamber , as depicted in fig .
[ fig : trapsetup ] . to avoid excess heating of the cryostat and prepare the system for injection of externally produced hci , mg ions are produced by an off - line ion source .
it consists of a directly heated tungsten crucible filled with grains of mg metal .
mg atoms leaving the crucible are ionized inside a cup - formed grid by electrons emitted from a thoriated tungsten filament located outside the grid .
the potential of the grid sets the energy of the produced ions .
they are collimated with an einzel lens and enter a 90@xmath16 quadrupole deflector , which guides them into the vertical part of the beamline .
the quadrupole geometry is chosen to allow injection of ions from both sides of the beamline and to have free access along the vertical axis for the laser beam .
two additional einzel lenses in the vertical beamline prepare the ion bunch for injection into the magnetic field and guide them into the trap .
the second arm of the horizontal beamline will be connected to an ebit and later to the hitrap cooling trap in order to trap heavy hci provided by the gsi accelerator facility .
mg ions are produced in bunches of 1 - 2 @xmath17s length at a rate of a few hz .
they are transported towards the trap with a kinetic energy of 200 ev and dynamically captured into the penning trap .
one typical trapping cycle is illustrated in fig .
[ fig : trappingcycle ] .
initially , only the lower capture ( reflector ) electrode is permanently switched high ( closed ) , while the upper capture electrode is switched between a confining potential and a value just below the ion transport energy , synchronized with the arrival time of the ion bunch .
it has been experimentally observed that around 50 ev out of 200 ev axial energy are transferred into the radial motion during the ion injection into the magnetic field .
that is sufficient for the accumulation of many ion bunches , with minimal losses of ions already stored during the reopening of the capture electrode .
typically 50 - 200 such accumulation cycles are repeated before permanently closing the capture electrode .
the voltage on the endcaps and correction electrodes is then slowly ( with respect to the ion motion ) ramped up in order to compresses the ion cloud towards the trap centre .
afterwards , laser cooling by scanning the laser wavelength as well as electronic ion excitation and detection are performed .
a fast cooling method , such as laser cooling , is needed in order to rapidly decrease the energy of stored ions and reduce losses .
as previously stated , direct laser cooling is limited to ions with a favourable level structure .
such ions can then be used for sympathetic cooling of other ions of interest which are simultaneously stored .
a suitable species for laser cooling is the @xmath18mg@xmath0 ion .
it can be easily produced and the undisturbed ion provides a closed , ground - state , two - level @xmath19 - @xmath20 transition , with an excited state natural lifetime of only 4 ns . however , magnetic fields of several tesla in the penning trap lead to splitting of the mg sublevels due to the zeeman effect , as shown in fig .
[ fig : mgzeeman ] . the zeeman slope coefficient for each sublevel
is provided in table [ zeemanshift ] .
ground state transition in @xmath21 .
the necessary polarisation for driving the corresponding transition is also indicated.,scaledwidth=45.0% ] ccc * level * & * @xmath22 * & * @xmath23(ghz / t ) * + @xmath24s@xmath25 & @xmath26 & @xmath27 + @xmath24s@xmath25 & @xmath28 & @xmath29 + @xmath24p@xmath30 & @xmath31 & @xmath32 + @xmath24p@xmath30 & @xmath26 & @xmath33 + @xmath24p@xmath30 & @xmath28 & @xmath34 + @xmath24p@xmath30 & @xmath35 & @xmath36 + [ zeemanshift ] in this level scheme only the @xmath37 ( here and in further text the @xmath22 quantum numbers ) zeeman transitions remain closed systems .
we have chosen the lowest transition @xmath38 for the cooling process .
it has a zeeman shift of @xmath27 ghz / t compared to the unperturbed @xmath24s@xmath39p@xmath30 transition frequency of @xmath40 ghz @xcite . considering the required polarization for the cooling laser and the injection direction of the ions ,
the laser is polarized @xmath41 and sent along the trap axis .
a specific issue of the experiment is the relatively high kinetic energy of the captured ions , required for an efficient transport and a small ion bunch width .
frequency detuning of the laser corresponding to a typical transport energy of 200 ev together with a very fast adjustment of this detuning to match the dropping energy caused by the cooling would be a serious technical challenge .
however , a much simpler approach can be employed at the expense of the cooling speed : since the axial speed of the injected ions varies between a maximum corresponding to 200 ev at the trap centre and zero at the turning points near the endcaps , the laser can be kept fixed at a frequency corresponding to a cold ion .
it can then absorb a photon near each turning point @xcite which is still efficient provided that the laser intensity is sufficiently large .
the condition is that the rabi frequency for this transition is much larger than the axial frequency of the ion inside the trap , @xmath42 .
it should be noted that only the axial ion motion will be directly cooled this way , since there is no cooling force acting on the ion cloud in the radial direction .
an all - solid - state laser system at the required wavelength of 279 nm has been set up as depicted in fig .
[ fig : laser ] .
it has been described in detail in @xcite and comprises a single - mode fiber laser at 1118 nm as well as two cavities for second harmonic generation to obtain frequency quadrupling .
the laser is a koheras boostik fiber laser , specified to deliver 1.66 w maximum output power .
experimentally , a maximum of 1.2 w including the amplified spontaneous emission ( ase ) was obtained .
: the laser beam from a fiber laser is frequency quadrupled using two non - linear crystals in bow - tie resonators with the length stabilized using the hnsch - couillaud locking scheme .
the wavelength is controlled using a high - finesse wavemeter .
( @xmath43 and @xmath44 : quarter- and half - wave plate respectively , app : anamorphic prism pair , m1-m8 cavity mirrors , fc : fiber coupler),scaledwidth=43.0% ] since the doubling efficiency for second harmonic generation ( shg ) inside a non - linear crystal is proportional to the square of the fundamental power , bow - tie optical resonators were constructed to enhance the laser power inside the crystal .
the first doubler uses non - critical phase - matching in a lithium triborate ( lbo ) non - linear crystal , which has a phase - matching temperature at 1118 nm of @xmath45 90@xmath46c .
the second doubler employs critical phase - matching of a beta barium borate ( bbo ) non - linear crystal , using the round , gaussian output of the first doubler .
both resonators were designed using the ray transfer matrix analysis and computer simulations in order to obtain the optimal experimental parameters according to the boyd - kleinmann theory .
the length of the doublers is actively stabilized using hnsch - couillaud polarization - analysis locking @xcite .
for the first doubler , a maximum overall shg efficiency of 33% was obtained , providing 320 mw of laser power at 559.3 nm from 950 mw of the fundamental 1118.5 nm power .
for the second doubler , a maximum of 16.7 mw at 279.6 nm was achieved using 210 mw of green power in front of the resonator , equivalent to an overall shg efficiency of 8% .
all power levels of the harmonics were measured after appropriate filtering of the fundamental power leaking from the bow - tie resonators . in spite of several problems with the koheras main fiber laser , which greatly affected the available pump power ,
a sufficient amount of about 2 mw uv laser power was available for the first trapping tests .
the parameters of the frequency quadrupling system are summarized in table [ resonatorsummary ] .
ccc * parameter * & * @xmath47 resonator * & * @xmath48 resonator * + crystal type & lbo & bbo + phase matching &
ncpm type i & cpm type i + crystal length & 20 mm & 7.4 mm + input wavelength @xmath49 & 1118.54 nm & 559.27 nm + crystal cut @xmath50 & 90@xmath46 & 44.4@xmath46 + crystal cut @xmath51 & 0@xmath46 & 0@xmath46 + crystal surfaces & dual ar & brewster - cut + crystal temperature & 96@xmath46c & 50@xmath46c + total cavity length & 1158 mm & 504 mm + focusing mirrors @xmath52 & 70 mm & 50 mm + focusing arm length & 157.5 mm & 104.8 + full folding angle & 38@xmath46 & 18.8@xmath46 + enhancement factor @xmath53 & @xmath4560 & @xmath4570 + coupling efficiency & @xmath5485% & @xmath5485% + input power ( @xmath55 ) & 950 mw & 210 mw + output power ( @xmath56 ) & 320 mw & 16.7 mw + doubling efficiency & @xmath4533@xmath57 & @xmath458@xmath57 + +
the storage time constant has been determined by monitoring the laser induced fluorescence of the trapped ions as a function of time . for this measurement ,
the laser has been tuned to a frequency @xmath45 200 mhz above the resonance frequency of the @xmath58 transition in order to avoid strong laser cooling and the resulting fluctuation of the fluorescence due to the decreasing doppler width of the transition . on the other hand , although a blue - detuned laser frequency leads to heating of the ion cloud , it is important to note that it is still red - detuned for the other mg isotopes which might be present .
additionally , resistive cooling contributes to ion cooling with a time constant of around 100 s for @xmath59 stored in spectrap @xcite .
thus , an equilibrium between heating and cooling processes is established , resulting in a fluorescence signal proportional to the number of stored ions .
a typical trace resulting from this procedure is shown in fig .
[ fig : ionlifetime ] . a storage time constant of about 140 s can be extracted from a single exponential fit to the data . since in this measurement the ions were not cooled to sub - k temperature it can be regarded as a lower limit for the ion lifetime in the trap . stored in spectrap .
the @xmath58 transition was continuously excited with a 200 mhz blue - detuned laser .
the observed fluorescence rate is plotted as a function of time . at @xmath60 loading was complete and the endcaps were closed .
the data was fitted with a single exponential function for points with @xmath61 s and the storage time constant of 137(5 ) s was observed.,scaledwidth=42.0% ] after improving the laser stability and the vacuum conditions , longer storage times of up to an hour have recently been observed for a cold cloud of @xmath21 ions @xcite .
further efforts in this direction are ongoing .
unlike @xmath21 , the storage time of hci can be significantly smaller and an estimate should be made .
ion loss is mainly attributed to charge exchange with residual gas particles .
although the cross sections for electron capture in ion - neutral collisions at very low energies are largely unknown , they can be estimated using the semi - empirical mller - salzborn formula @xcite @xmath62 \label{eq : schlachter}\ ] ] where @xmath63 is the charge state of the ion and @xmath64 is the ionization potential of the residual gas particle expressed in ev , which amounts to @xmath65 ev for h@xmath66 @xcite and @xmath67 ev for he .
partial pressures of all other typical residual gases are much smaller at the cryogenic temperature around the trap , and can be safely neglected . alternatively , the cross section can also be estimated using the so - called classical barrier model @xcite which brings similar results .
the rate @xmath68 of electron capture is then calculated by multiplying the cross - section @xmath69 from eq . by the neutral particle density @xmath70 and the relative velocity @xmath71 of the two colliding particles .
the expected ion storage time is given through the reciprocal value of this rate @xmath72 where @xmath73 is the boltzmann constant , @xmath74 the pressure , @xmath75 and @xmath76 are the mass and temperature of the residual gas atoms and the ions , respectively . assuming that the pressure inside the trap volume is not worse than @xmath77 mbar , which corresponds to @xmath78 mbar measured in the 300 k region and scaled down to liquid helium temperature , the lifetime of around 160 s and 18 s is calculated using eq
. for charge states @xmath79 and @xmath80 , respectively .
these lifetime - estimates were calculated for ion temperatures around 1 k or less and rapidly decrease with increasing ion temperature , pointing towards the need for rapid ion cooling , such as sympathetic cooling with laser cooled @xmath21 .
the laser cooling time can be estimated by evaluating the cooling force exerted by the photons .
generally , for low energy ions , the scattering of the photons leads to a frictional force @xmath81 which slows the ion down .
it can be written as @xcite @xmath82 where @xmath83 is the photon momentum , @xmath84 is the transition linewidth , @xmath64 is the intensity of the laser , @xmath85 the saturation intensity and @xmath86 the detuning of the laser frequency .
this force is proportional to the laser intensity below the saturation value and it approaches its maximum value @xmath87 for intensities @xmath88 . here , the kinetic energy of the captured ions is typically 200 ev , and the laser frequency can not be scanned fast enough to maintain the cooling condition .
the laser is thus kept fixed at a small red - detuning and the scattering force in eq .
( [ fscatt0 ] ) can then be written as @xmath89 where @xmath90 is a scattering frequency inversely proportional to the ion velocity .
this equation is a good approximation for an intensity close to or above the saturation intensity .
since under our experimental conditions the laser intensity was about 1/3 of the saturation intensity , the cooling force is reduced by a factor of @xmath91 .
the deceleration can then be expressed as @xmath92 where @xmath93 is the wavelength of the cooling laser and @xmath94 the mass of @xmath18mg@xmath0 .
the stopping time can be calculated accordingly as @xmath95 where @xmath96 is the initial speed of the ions .
if @xmath90 is approximated as the maximal scattering frequency reduced by the ratio between the natural linewidth @xmath84 and the doppler width @xmath97 of the transition , in our case this stopping time approximation results in @xmath98 s. the experimental cooling time was determined using the same measurement procedure as described in the previous section for recording the lif signal shown in fig .
[ fig : ionlifetime ] .
initially , the ions have a large spatial oscillation amplitude between the endcaps while the fluorescence detection system is focused on a small volume at the centre of the trap . hence ,
the emitted photons can not be recorded and only the background signal is present . as the ions are cooled , they get localized in the centre of the trap and the fluorescence rate per ion rises as the doppler - shifted transition matches the fixed red - detuned laser frequency for an increasing amount of ions .
this results in the sharp rise in fluorescence observed in fig .
[ fig : ionlifetime ] , which appears about 10 seconds after raising the endcaps to the trapping potential .
the measured value is of the same order of magnitude , but smaller than the predicted 40 s because of uneven ion velocity distribution and a finite probability for photon absorption also outside the natural linewidth of the transition .
in contrast to trapping a large ion cloud , with the current experimental setup it was also possible to isolate and observe the fluorescence of a single trapped ion . the spectra shown in fig . [
fig : singleions ] were recorded by scanning the laser from -1 ghz with 100 mhz / s across the resonance , using a 0.9 mw laser beam with a diameter of @xmath99 mm .
taken under identical conditions and trapping times , they show quantized changes of the laser induced fluorescence , associated with single trapped ions on top of a constant background signal . in order to verify that the single - ion regime was reached the following procedure
was carried out : first an integrated number of detected photons over the resonance was determined for each of the recorded few - ion spectra .
the error was treated as the statistical uncertainty with an added offset from the average deviation of the background .
the number of photons per ion was then varied between 10 and 500 and compared to the recorded spectra , producing a deviation for each point .
these deviations were used to calculate the reduced chi - square for each assumed number of photons per ion .
the result of this procedure is plotted in fig .
[ fig : chisquare ] .
the area around @xmath100 corresponds to the most probable number of integrated photons per ion ( roughly @xmath101 ) which was used to determine the number of ions in the spectra shown in fig .
[ fig : singleions ] . represents the most probable value of roughly @xmath101 photons.,scaledwidth=42.0% ] according to it , a single stored @xmath21 ion in full resonance yields a rate of around @xmath102 fluorescence photons per second , on top of around @xmath103 background photons per second .
the uncertainty of this value is dictated by the fluctuation of the background and the frequency uncertainty of the scanning laser frequency .
having that in mind it can be concluded that under the given conditions a single ion yields a fluorescence signal of @xmath104 photons / s . in spite of the large uncertainty of around 20%
, this result shows that even a few trapped ions with a fast optical transition can be detected in spectrap via laser induced fluorescence .
if the observed number of fluorescence photons is compared to the maximum expected number of photons from a non - fully saturated @xmath21 ion , this yields a total detection efficiency of @xmath105 . disregarding the quantum efficiency of the detector and considering only geometrical factors the detection efficiency amounts to about @xmath106 .
additionally , the expected number of photons per ion can be used to quantify a signal from an ion cloud and estimate the number of ions stored under the same conditions .
a maximum of about 2000 ions were trapped and cooled using the current ion source and @xmath99 mw / mm@xmath107 of cooling laser intensity .
as an addition to lif detection , spectrap can also perform ft - icr ( fourier transform ion cyclotron resonance ) measurements , a well - established technique for non - destructive mass and charge state spectrometry in ions traps @xcite . by a combination of both , it is possible to gauge the electronic signal height obtained in ft - icr to the number of observed ions as measured by lif . hence , a stored ion cloud can be characterized by the ion number and temperature . for ft - icr
, the ion motion is excited by a fix - phase burst such that subsequent signal pickup of the ions oscillatory motions is efficient .
a transient of that signal is recorded and its fourier transform represents a spectrum of the mass - to - charge spectrum of ions present in the trap . in the present case ,
the signal for ion excitation is generated by an agilent 33250a frequency generator and processed by the spectrap rotating wall drive @xcite , which splits the input signal into two with a 180@xmath16 phase difference .
these two signals are transmitted to two opposing ring segments , while the remaining two segments are kept at dc potential .
the signal induced in the trap electrodes by the excited ions is amplified by a cryogenic amplifier mounted next to the trap and processed by a hp3589a spectrum analyser .
the dipole excitation was performed by applying 5000 cycles of a 2.555 mhz signal , where the modified cyclotron resonance @xmath108 was expected for a magnetic field of 4 t. the amplitude was set to @xmath109 . because of the short coherence time the spectrum analyser was triggered by the last excitation cycle and averaged over 10 excitation - detection rounds .
the observed resonance signal is depicted in fig .
[ fig : cycresonance ] .
the width of the resonance is @xmath45 100 hz , while the modified cyclotron frequency can be determined with an accuracy of a few hz .
the resulting mass resolving power @xmath110 is of the order of @xmath111 , while the magnetic field can be determined with a relative accuracy of @xmath112 .
both of these values exceed the requirements of the experiment and show that electronic and optical ion detection can be performed simultaneously . ; @xmath108 is the modified cyclotron frequency.,scaledwidth=42.0% ] the transfer of energy into the cyclotron motion during excitation pushes the ions into larger orbits , where they either have a smaller overlap with the laser beam , or are even lost from the trap .
this was used to perform a measurement of the modified cyclotron frequency via lif .
the ions were laser - cooled and their fluorescence recorded while applying the dipole excitation to the ring electrode .
the excitation frequency was changed stepwise across the expected cyclotron resonance , while the trap was reloaded under identical conditions for each point .
this resulted in a fluorescence dip seen in fig .
[ fig : cyclaser ] , fitted well with a gaussian function , with the central frequency marking the resonance .
it was noticed that the range of possible excitation amplitudes was rather narrow - excitation with more than 400 mv@xmath113 resulted in a total loss of fluorescence or even ion loss , i.e. the fluorescence did not return after switching off the excitation .
conversely , amplitudes smaller than 100 mv@xmath113 had little or no observable influence on the ion fluorescence .
.,scaledwidth=42.0% ] it can be seen that the central frequencies from the two measurements , shown in fig .
[ fig : cycresonance ] and fig .
[ fig : cyclaser ] , differ by 1.33 khz .
this exceeds the statistical fitting uncertainty and was found in several repeated measurements .
the systematic shift corresponds to a magnetic field difference of 2 mt .
the discrepancy is ascribed to the non - ideal magnetic field and the different spatial positions where the two measurements were performed : while the ft - icr induces a signal directly in the trap electrodes , the lif - signal depends on the optical axis of the detector system , which is not necessarily aligned exactly along the trap radial axis . adding a camera to the system will allow us to measure position , shape and radial extent of the cloud , which is of special interest when the rotating wall is applied .
nevertheless , it was demonstrated that electronic and laser induced fluorescence ion detection methods can be used simultaneously , with reasonably good agreement . by
determining the transition linewidth @xmath114 and assuming the absence of line - broadening mechanisms other than doppler broadening , the upper limit to the ion temperature @xmath115 can be calculated according to @xcite @xmath116 a series of measurements was performed in order to determine the transition linewidth of the laser - cooled @xmath21 .
the laser frequency was kept 1 ghz red - detuned during ion accumulation , as well as for another 10 seconds after closing the trap .
it was observed that due to the large initial ion energy this pre - cooling time was necessary for efficient laser cooling .
after pre - cooling , the laser frequency was scanned over the central transition frequency of mg@xmath11 and the fluorescence recorded .
a typical result is shown in fig .
[ fig : crystal ] , where the recorded fluorescence rate was plotted against the laser frequency detuning . as a function of frequency detuning .
the width of the measured transition can be used to set the upper limit to the achieved ion temperature . a precooling peak and an abrupt drop after crossing the zero is typically associated with crystalline structure of the ion cloud.,scaledwidth=42.0% ] after crossing the resonance frequency , ion cooling turns into heating and the fluorescence drops quickly to zero .
it can therefore be safely assumed that the total fwhm of the voigt profile is less than twice the observed width of 33(10 ) mhz indicated in fig .
[ fig : crystal ] .
this value is of the same order of magnitude as the natural linewidth of the transition ( 42 mhz ) , and a deconvolution of the doppler and the natural linewidth contribution to the line profile needs to be performed @xcite .
the deconvoluted value for the doppler width can be inserted into eq . and an upper limit for the ion temperature is obtained @xmath117 by experience from a similar experiment @xcite , the typical volume of such an ion cloud is of the order of 0.5 mm@xmath118 , resulting in an ion number density of around @xmath119 ions / mm@xmath118 for @xmath21 ions stored in spectrap . under such conditions
the single - particle description begins to break down and the ion cloud has to be treated as a non - neutral plasma .
the plasma coupling parameter , describing the ion coulomb coupling intensity in one - component plasmas , is given by @xcite @xmath120{\frac{3}{4\pi
n } } \label{eq : plasmacoupling}\ ] ] where @xmath121 is the wigner - seitz radius .
gilbert and co - workers have predicted that for coupling parameters @xmath122 the plasma starts gradually to exhibit liquid - like properties @xcite . according to eq . ,
for @xmath21 trapped and cooled in spectrap ( @xmath123 ions / mm@xmath118 and @xmath124 ) the plasma coupling parameter amounts to @xmath125 , such that strong ion coupling may be assumed .
after the planned introduction of hci into the trap , the temperatures of the two components are expected to roughly equalize , resulting in a much larger @xmath126 for high charge states @xcite .
studies have already shown that for sufficiently low temperatures , a trapped ion cloud exhibits a structural change and its spectrum resembles the one of single ions @xcite .
a similar behaviour was observed for the trapped @xmath21 in spectrap and is shown in fig .
[ fig : crystal ] .
a small pre - cooling peak appears at the point where the transition s doppler broadened half - width becomes smaller than the laser detuning ( here at @xmath45 400 mhz ) , after which the fluorescence disappears and can be observed again only close to the natural linewidth of the transition .
such structures were observed also in @xcite and mark the transition of the stored ion plasma from a non - correlated to a strongly coupled state . because of strong cooling and simultaneous reduction of the doppler width
, the fluorescence close to the resonant frequency is characterized by a sharp asymmetric shape , followed by an abrupt drop to zero after crossing the central frequency . the spectrum shown in fig . [ fig : crystal ] was recorded with 1.1 mw / mm@xmath107 of laser power and a 100 mhz / s frequency sweep .
it has been observed that different pre - cooling times cause different positions of the pre - cooling peak with respect to the main one , moving them closer together for shorter pre - cooling times .
this structure was , however , not observed for very short pre - cooling times below roughly 8 s , which were also typically followed by much smaller or no detectable fluorescence .
however , after allowing sufficient pre - cooling time and observing the crystalline structure , a smaller , sharp fluorescence peak was observed at resonance even when scanning the laser frequency in the opposite direction .
we have performed systematic measurements with laser - cooled @xmath127mg@xmath128 ions stored in a penning trap .
these ions were externally produced , transported , captured and stored in the trap for subsequent measurements . using both optical and electronic non - destructive detection techniques , the properties of stored ion clouds were determined . combining electronic with optical detection , it is possible to determine stored ion numbers down to the single ion level and to characterize the stored ion cloud with respect to its temperature , storage time and related properties .
laser cooling was achieved to temperatures below 0.1 k and evidence of ion crystallization was found .
such laser - cooled ions are ideal for sympathetic cooling of simultaneously trapped ion species which lack a suitable level scheme , as for example the highly charged ions that will be available in the near future .
all necessary prerequisites for precision spectroscopy of externally produced highly charged ions to be delivered by the hitrap facility at gsi were demonstrated in the present setup with singly charged magnesium ions .
further investigation of sympathetic cooling of mid - z highly charged ions in spectrap will be carried out in the near future .
finally , the measurements of forbidden transitions in heavy highly charged ions will open the way to precision tests of qed calculations in extreme fields and to the determination of fundamental constants .
this work is supported by hgf under contract number vh - ng-148 , bmbf under contract numbers 06ms7191 , no . 05p12rdfa4 and 06da9020i as well as epsrc under the grant number ep / d068509/1 .
we also thank llnl and the former retrap group for their support and loan of the equipment to gsi .
99 t. beier , phys . rep . * 339 * , 79 ( 2000 ) .
m. vogel and w. quint , phys .
rep . * 490 * , 1 ( 2010 ) .
v. mckel , r. klawitter , g. brenner , j.r .
crespo lpez - urrutia , and j. ullrich , phys .
* 107 * , 143002 ( 2011 ) .
i. klaft , et al .
lett . * 73 * , 2425 ( 1994 ) .
p. seelig , et al .
lett . * 81 * , 4824 ( 1998 ) .
w. nrtershuser et al . , proceedings of hci 2012 , heidelberg , phys .
scripta , submitted ( 2012 ) .
shabaev , a.n .
artemyev , v.a .
yerokhin , o.m .
zherebtsov , g. soff , phys .
lett . * 86 * , 3959 ( 2001 ) .
m. vogel , d.f.a .
winters , d.m . segal and r.c .
thompson , rev .
inst . * 76 * , 103102 ( 2005 ) .
z. andjelkovic , s. bharadia , b. sommer , m. vogel and w. nrtershuser , hyp . int . * 196 * , 81 ( 2010 ) .
j. kluge et al . , advances in quantum chemistry * 53 * , 83 ( 2007 ) .
major , v.n .
gheorghe and g. werth , _ charged particle traps _ , springer ( 2004 )
. m. hobein , a. solders , m. suhonen , y. liu , and r. schuch , phys .
lett . * 106 * , 013002 ( 2011 ) d.j .
wineland and h.g .
dehmelt , j. appl .
phys . * 46 * , 919 ( 1975 ) .
winters , m. vogel , d.m segal and r.c .
thompson , j. phys .
b * 39 * , 3131 ( 2006 ) .
w.m . itano and d.j .
wineland , phys .
a * 25 * , 35 ( 1982 ) .
d. leibfried , r. blatt , c. monroe and d. wineland , rev .
phys . * 75 * , 281 ( 2003 ) .
j. eschner , g. morigi , f. schmidt - kaler and r. blatt , j. opt .
b * 20 * , 1003 ( 2003 ) .
thompson , s. donnellan , d.r .
crick and d.m .
segal , j. phys .
b * 42 * , 154003 ( 2009 ) .
l. gruber , j.p .
holder and d. schneider , phys .
* 71 * , 60 ( 2005 ) .
h. hffner et al .
j. d * 22 * ( 2003 ) 163 .
ghosh , _ ion traps _ , oxford science publications ( 1995 ) .
g. gabrielse , l. haarsma and s.l .
rolston , int .
j. mass spectr .
* 88 * , 319 ( 1989 ) .
brown and g. gabrielse , rev . mod .
phys . * 58 * , 233 ( 1986 ) .
s. bharadia , m. vogel , d.m . segal and r.c .
thompson , appl .
b , * 104 * , 1105 ( 2012 ) .
g. weinberg , b.r .
beck , j. steiger , d.a .
church , j. mcdonald and d. schneider , phys .
a * 57 * , 4452 ( 1998 ) .
l. gruber , j.p .
holder , j. steiger , b.r .
beck , h.e .
dewitt , j. glassman , j.w .
mcdonald , d.a .
church , and d. schneider , phys .
lett . * 86 * , 636 ( 2001 ) .
m. herrmann , v. batteiger , s. knnz , g. saathoff , th .
udem , and t.w .
hnsch , phys .
lett . * 102 * , 013006 ( 2009 ) .
r. cazan , phd thesis , johannes gutenberg university of mainz , germany ( 2012 ) .
r. cazan , c. geppert , w. nrtershuser and r. sanchez , hyp . int . * 196 * , 177 ( 2010 ) .
hnsch and b. couillaud , opt
35 * , 441 ( 1980 ) .
z. andelkovic , phd thesis , johannes gutenberg university of mainz , germany ( 2012 ) .
t. murbck , et al . ,
proceedings of hci 2012 , heidelberg , phys .
scripta , submitted ( 2012 ) .
a. mller and e. salzborn , phys
62a * , 391 ( 1977 ) .
g. herzberg , phys .
* 23 * , 1081 ( 1969 ) .
r. mann , z. phys .
d 3 , 85 ( 1986 ) .
foot : _ atomic physics _
, oxford university press , ( 2005 ) .
k. blaum , phys . rep . * 425 * , 1 ( 2006 ) . j.j .
olivero and r.l .
longbothum , j. quant .
transfer * 17 * , 233 ( 1977 ) .
gilbert , j.j .
bollinger and d.j .
wineland , phys .
lett . * 60 * , 2022 ( 1988 ) .
f. diedrich , e. peik , j.m .
chen , w. quint and h. walther , phys .
lett . * 59 * , 2931 ( 1987 ) .
t. gudjons , f. arbes , m. benzing , f. kurth and g. werth , phys .
scripta t59 , 396 ( 1995 ) .
k. okada , t. takayanagi , m. wada , s. ohtani and h.a .
schuessler , phys .
a * 80 * , 043405 ( 2009 ) . | we have performed laser cooling of mg@xmath0 ions confined in a penning trap .
the externally produced ions were captured in flight , stored and laser cooled .
laser - induced fluorescence was observed perpendicular to the cooling laser axis
. optical detection down to the single ion level together with electronic detection of the ion oscillations inside the penning trap have been used to acquire information on the ion storage time , ion number and ion temperature .
evidence for formation of ion crystals has been observed .
these investigations are an important prerequisite for sympathetic cooling of simultaneously stored highly - charged ions and precision laser spectroscopy of forbidden transitions in these . |
in recent years , great progress has been made in quantifying the evolution of the galaxy population at the end of cosmic reionization around @xmath15 .
deep hubble space telescope ( hst ) legacy fields , such as the hubble ultra - deep field ( hudf ; * ? ? ? * ) or goods @xcite , and wide area ground - based imaging , have made it possible to study the evolution of the uv luminosity function ( lf ) across @xmath16 to great accuracy ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . over the last two years , with the installation of the wide field camera 3 ( wfc3 ) onboard the hst
, the observational frontier of galaxies has now been pushed into the reionization epoch , as deep wfc3/ir data led to the identification of more than 100 galaxy candidates at @xmath17 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this is essential for estimating the contribution of galaxies to cosmic reionization .
one of the most important conclusions from these studies is thus the realization that the uv luminosity density ( ld ) emitted by the galaxy population gradually falls towards higher redshifts .
for example , the ld of the @xmath18 galaxy population is about an order of magnitude larger than that of the @xmath12 population , about 1.5 gyr earlier .
how this evolves to even higher redshifts is still very unclear .
a sizable galaxy population at @xmath19 is expected based on the first estimates of stellar population ages of @xmath20 galaxies , indicating that these sources very likely started forming stars already at @xmath21 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this is still somewhat uncertain due to possible nebular line emission contaminating the spitzer photometry ( e.g. * ? ? ?
nonetheless , an early epoch of star - formation is also required by the mean redshift of reionization as measured by wmap ( @xmath22 ; * ? ? ?
* ) , if galaxies are assumed to be the main drivers for this process .
however , previous searches for @xmath19 sources in the pre - wfc3 era only resulted in very small samples of relatively low reliability candidates , none of which have been confirmed ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ?
this is mainly due to the extreme faintness of the @xmath0 galaxy population .
not only are these galaxies fainter due to their increased distance , but also they are expected at intrinsically lower luminosities . additionally , the detection of such high - redshift sources is further complicated by the fact that they are invisible in optical data . due to the highly neutral inter - galactic medium before the end of cosmic reionization , their uv photons
are absorbed shortward of the redshifted ly@xmath23 line , which shifts to @xmath24 at @xmath25 .
thus , these sources can only be seen in the nir , where previous detectors were significantly lagging behind optical technology . with 40@xmath26 higher efficiency relative to nicmos to detect high redshift galaxies in the nir , wfc3/ir has the potential to change this and to push galaxy studies to beyond @xmath27 . several deep and wide area wfc3/ir data sets have been taken already and several more are upcoming .
the challenge for identifying genuine @xmath28 sources in these data sets , is that these galaxies will only be visible in one band ( @xmath4 ) .
this has already led to some controversy in the first searches for @xmath0 sources in the first - epoch wfc3/ir data over the hubble ultra deep field ( see e.g. * ? ? ?
* ; * ? ? ?
in our recent analysis , which includes the full two - year wfc3/ir data over the hudf as well as the shallower , but wider early release science ( ers ) data , @xcite found only one single galaxy candidate detected at @xmath29 with an estimated redshift at @xmath30 . given that about three should have been detected , if the evolution of the lf continued as extrapolated from the trends established across @xmath31 to @xmath15 , this provided first tentative evidence for an accelerated evolution in the galaxy population from @xmath12 to @xmath0 .
in this paper we significantly expand on our first @xmath0 analysis from wfc3/ir data presented in @xcite by extending the search to all the deep wfc3/ir fields in the chandra deep - field south area that have since become available .
the inclusion of the two deep hudf09 parallel fields is especially useful , since both reach just @xmath32 mag shallower than the ultra - deep hudf field but triple the search area for @xmath33 ab mag sources .
additionally , we use different analysis tools developed by the first author that provide an independent analysis of the hudf and ers data . while the lyman - break approach is similar in principle to that of @xcite the use of independently tested software and procedures for the source detection and its analysis provides confirmation and validation .
the expanded data set also covers @xmath34 the area at moderately deep @xmath35 ab mag , thanks to the inclusion of the first epochs of candels data over these fields .
this will be used to constrain the evolution of the galaxy population over the @xmath36 myr from @xmath0 when the universe was @xmath37 myr old to @xmath12 at @xmath38 myr .
we start by describing the full data set in [ sec : data ] and present the @xmath39 candidate selection and its efficiency in [ sec : selection ] . in
[ sec : lf ] we present our new constraints on the lf at @xmath0 .
we will refer to the hst filters f435w
, f606w , f775w , f850lp , f098 m , f105w , f125w , f160w as @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45 , @xmath7 , @xmath4 , respectively
. throughout this paper , we adopt @xmath46 kms@xmath47mpc@xmath47 , i.e. @xmath48 . magnitudes are given in the ab system @xcite . and table [ tab : data ] ) .
the hudf / hudf09 - 1/hudf09 - 2 fields comprise the deepest nir data available to date , reaching to @xmath49 ab mag .
these three wfc3/ir pointings are covered in a total of 192 hst orbits , which obtained photometry in @xmath45 , @xmath7 , and @xmath4 .
the ers field consists of 60 orbits of wfc3/ir imaging in @xmath44 , @xmath7 , and @xmath4 spread over a 2@xmath265 tile .
the candels program is an on - going multi - cycle treasury survey consisting of two parts : candels - deep ( @xmath50 arcmin@xmath2 , limited at @xmath51 ) and candels - wide ( @xmath52 arcmin@xmath2 , limited at @xmath53 ) . only @xmath7 and @xmath4 data obtained before august 6 , 2011 is used in this analysis ( together with the shallow @xmath45 imaging of candels wide ) . ]
lcccccccccc hudf09 & 4.7 & 29.2 & 29.6 & 29.4 & 28.8 & 29.1 & 29.3 & 29.4 + hudf09 - 1 & 4.7 & & 28.9 & 28.7 & 28.6 & 28.6 & 28.8 & 28.6 + hudf09 - 2 & 4.7 & 28.8 & 29.3 & 28.9 & 28.7 & 28.6 & 28.9 & 28.9 + ers & 41.3 & 27.8 & 28.0 & 27.5 & 27.2 & 27.4 & 27.8 & 27.6 + candels - deep & 63.1 & 27.8 & 28.0 & 27.5 & 27.2 & & 27.7 & 27.5 + candels - wide & 41.9 & 27.8 & 28.0 & 27.5 & 27.2 & 27.1 & 27.2 & 26.9
our analysis is based on the public wfc3/ir data sets that are available over the goods south fields @xcite as a result of three different programs ( hudf09 , ers , and candels ) , which we describe below .
the outline of all fields is shown in figure [ fig : fields ] .
the key feature of these fields is that they have wfc3/ir coverage in @xmath7 and @xmath4 , which will be used to select @xmath0 sources .
additionally , they have deep optical coverage as well as deep irac imaging , which is essential to exclude low redshift contamination ( see section [ sec : selection ] ) .
all the wfc3/ir data has been reduced following standard procedures outlined , e.g. , in @xcite . in particular
, our reduction pipeline includes the subtraction of a super median image , careful image registration to the acs frames and an automatic elimination of pixels affected by persistance .
the final pixel scale of the images in our analysis is set to @xmath54 .
a summary of the hst data used in this analysis can be found in table [ tab : data ] .
the final resolution of the wfc3/ir data is @xmath55 ( fwhm ) , and @xmath56 in the optical acs data .
the acs data on the goods fields we used are the v2 reductions that are publicly available from mast ( m. giavalisco and the goods team , in preparation ) .
the spitzer data are the goods irac images @xcite as made publicly available by the simple team ( see e.g. * ? ? ?
additionally , we include the newly acquired spitzer irac [ 3.6 ] and [ 4.5 ] data over the hudf field from the iudf10 survey ( proposal 70145 , pi : labb ) .
this so far adds @xmath57h of observations , which increases the depth in both filters by an additional @xmath10.4 mag .
where available , we also matched our wfc3/ir sources with the publicly available goods - music multi - band photometry catalog of @xcite .
the hudf09 program ( pi : illingworth ; * ? ? ? * ) consists of 192 hst orbits to provide ultra - deep wfc3/ir imaging over three pointings centered on the hudf @xcite and its two parallel fields from the udf05 program ( pi : stiavelli ; * ? ? ?
the program has been completed , providing the deepest ir images ever taken .
it comprises @xmath58 arcmin@xmath2 imaging in the three filters @xmath45 , @xmath7 , and @xmath4 , reaching down to @xmath59 , and 28.9 ( @xmath60 in 05 diameter apertures ) for the hudf , hudf09 - 1 , and hudf09 - 2 , respectively ( see fig . [
fig : fields ] ) . for a more detailed description of this data
set and the data reduction see @xcite .
in addition to the ultra - deep hudf09 data , we also analyzed shallower , wider area wfc3/ir imaging from the ers and candels programs , in order to constrain the volume density of more luminous star - forming galaxies at @xmath0 .
the early release science data ( ers ) provide wfc3/ir imaging of @xmath61 arcmin@xmath2 of the northern part of the goods south field .
two orbits of wfc3/ir imaging were obtained in each of the filters @xmath44 , @xmath7 , and @xmath4 , over a @xmath62 grid of pointings ( 60 orbits in total ) .
these data are reduced in an analogous way to our hudf09 data , and are aligned and drizzled to the goods acs mosaics after rebinning to a @xmath54 pixelscale .
these data reach to @xmath63 ( see also * ? ? ?
for a more detailed description of this data set see @xcite .
the last two fields included in our analysis are obtained as part of the multi - cycle treasury program candels ( pi : faber / ferguson ; * ? ? ?
* ; * ? ? ?
in particular , we include the first six visits of the candels - deep program ( obtained until august 6 , 2011 ) , which covers the central part of goods south in @xmath64 tiles with @xmath65 s exposures in both @xmath7 and @xmath4 in a total of 92 orbits .
this data covers @xmath66 arcmin@xmath2 and reaches to @xmath51 mag .
additionally , we also included the imaging data of the supernova follow - up program of candels ( pi : riess ) , which adds imaging over two pointings over candels deep ( one of which is essentially centered on the hudf ) .
finally , we made use of the 29 orbits of wfc3/ir data of the candels - wide survey ( @xmath45 , @xmath7 and @xmath4 , obtained until march 29 , 2011 ) .
these comprise 9 wfc3/ir pointings ( @xmath67 arcmin@xmath2 ) , completing the coverage of the goods south field , and reach to @xmath68 ( @xmath69 s exposures ) . as for the ers
, the wfc3/ir data of the candels program has been aligned to the goods acs mosaics with a pixel scale of @xmath54 .
the part of the candels field overlapping with the wfc3/ir hudf has been omitted when analyzing this data set in order not to duplicate the analysis of that area .
we will subsequently refer to the combination of the ers and the two candels fields as ` wide fields ' .
source catalogs are derived with the sextractor program @xcite , which is used to detect galaxies in the @xmath4 images and perform matched aperture photometry on psf - matched images .
the colors used here are based on isophotal apertures derived from the @xmath4 images , and total magnitudes are measured in standard 2.5 kron apertures , corrected by 0.2 mag in order to account for flux loss in the psf wings .
the detection significance of sources was established in @xmath70 radius apertures .
the rms maps were scaled based on the detected variance in 1000 random apertures for each wfc3/ir frame on empty sky regions after @xmath71 clipping .
this procedure ensures that the sextractor weight maps correctly reproduce the actual noise in the images .
subsequently only sources with signal - to - noise ratios larger than 5 ( in @xmath70 radius apertures ) in @xmath4 are considered .
galaxies ( [ sec : candsel ] ) . shown are the @xmath72 colors of different types of galaxies as a function of their redshift .
star - forming galaxies are shown as solid blue lines .
the lighter blue spectral energy distribution ( sed ) is reddened by e(b - v)@xmath73 mag using a @xcite dust law . as the ly@xmath23 absorption due to inter - galactic neutral hydrogen shifts into the @xmath7 band ,
galaxies start to exhibit progressively redder colors beyond @xmath12 .
a selection with @xmath6 thus identifies galaxies at @xmath5 .
strong balmer and 4000 breaks in evolved @xmath18 galaxies , combined with some dust obscuration can also result in very red @xmath72 colors .
the dashed lines correspond to more evolved galaxies dominated by progressively older stellar populations from the library of @xcite as well as a 1 gyr old single stellar population ( ssp ) from the library of @xcite .
spitzer irac data provides a way to separate these different populations at intermediate and very high redshifts ( see text and fig .
[ fig : maglim ] ) . ] ) of some of the data used in this analysis .
the red arrows indicate the limits of the acs , wfc3/ir and irac data over hudf , while blue arrows correspond to the ers data set .
the sed of a @xmath74 star - forming galaxy is plotted in purple , showing the distinguishing feature of complete absorption shortward of the redshifted ly@xmath23 line . evolved and dusty galaxy seds can mimic the same features as a @xmath0 source , i.e. red @xmath72 and undetected in the optical data , unless this is extremely deep .
such sources are typically detected in the irac data .
alternatively , the solid and dashed green lines show possible seds of @xmath75 galaxies which could escape the irac detection at @xmath76 .
these correspond to evolved galaxies ( 500 myr ) , with moderate amounts of dust extinction , and are normalized to the 5 @xmath3 detection limits of the @xmath4 bands .
note that much larger amounts of dust can be hidden for interloper galaxies at the detection limit of the hudf ( green solid , @xmath77 mag ) than for the ers ( green dashed , @xmath78 mag ) .
thus even with irac it can be a challenge to separate real @xmath79 sources from lower redshift contaminants . ]
galaxies at @xmath80 are expected to exhibit very red @xmath72 colors since the redshifted , strong ly@xmath23 absorption ( by the predominantly neutral inter - galactic hydrogen ) cuts into the flux in the @xmath7 filter ( see figure [ fig : colsel ] ) .
this makes such high redshift galaxies completely invisible blueward of @xmath7 , and we use this fact for their identification .
as can be seen from figure [ fig : colsel ] , however , also passively evolving or dusty galaxies at intermediate redshifts ( @xmath81 ) can exhibit similarly red colors in @xmath72 . while the requirement of optical non - detections removes the bulk of lower redshift contamination , certain intermediate redshift galaxies with evolved or dusty stellar populations can still be included due to the fact that the optical data does not reach deep enough , if at similar depth as the ir ( see fig . [
fig : maglim ] ) .
deep spitzer irac data provides a way to identify contaminating galaxies .
these are expected to exhibit very red @xmath82 $ ] colors , which discriminates them from genuine @xmath0 candidates . to exclude possible low - redshift contamination , we thus use two steps to select @xmath5 galaxy candidates .
for the first step , the primary criteria are based on hst data only : @xmath83 @xmath84 additional to excluding objects that are detected in any band blueward of @xmath7 at more than @xmath76 , we include a cut in the optical @xmath85 value of a galaxy ( see e.g. * ? ? ? * ; * ? ? ? * ) .
this is computed from the @xmath70 radius aperture fluxes as @xmath86 , where the sum runs over all the bands available in the given data set blueward of @xmath7 , i.e. it includes all the available optical data as well as the nir band @xmath45 for the hudf09 data , and @xmath44 in the ers .
the relatively large apertures were chosen in order sample @xmath87 of the light of point - like sources .
the limiting @xmath88 are derived from photometric scatter simulations .
they are set to exclude the majority of interlopers which remain undetected at @xmath76 purely due to photometric noise , but not to cut a substantial fraction of galaxies with real zero flux in the optical bands .
the scatter simulations utilize all galaxies in our catalogs that are @xmath89 mag above the completeness limit , applying photometric gaussian noise from 1 mag fainter sources . from these simulations
it is clear that contamination is mainly an issue at 0.75 mag above the completeness limits , but that @xmath90% of contaminants can be eliminated by using a @xmath85 limit of @xmath91 or 2.4 , for 5 filters or 4 filters , respectively . in the hudf09 data , the resulting number of expected contaminants due to photometric scatter is thus reduced from @xmath32 source per wfc3/ir field to @xmath92 source . on the other hand ,
the adopted @xmath88 limits do remove an additional @xmath93% of sources with real zero flux , simply due to gaussian statistics .
this reduction of the real galaxy sample is reflected in our subsequent analysis in the reduction of the selection volume .
lcccccccccc hudfj-39546284 & 03:32:39.54 & -27:46:28.4 & @xmath94 & @xmath95 & 6.3 all galaxies passing the above selection criteria , using both the acs and wfc3/ir data , are retained and analyzed individually .
these total to 17 sources with @xmath96 in the range @xmath97 mag ; one source in the hudf , none in the parallel hudf09 fields , three in the ers and 8 and 5 in the candels deep and wide , respectively ( see tables [ tab : phot ] and [ tab : photcontamin ] ) .
interestingly , only one source ( the previously reported galaxy with @xmath98 mag from bouwens et al .
2011 ) did pass our selection in the three deep hudf09 fields , while the shallower candels and ers fields contribute a total of 16 sources ( all with @xmath99 mag ) .
upon inspection of their images , it turns out that all these brighter sources are very well detected in the irac data , even in the shallow 8.0@xmath100 band .
their measured @xmath101 $ ] colors are in the range @xmath102 , which , for a @xmath0 source , would correspond to a uv continuum slope @xmath103 , or a dust reddening with @xmath104 mag .
given that galaxies at the bright end of the @xmath105 population are measured to have very low extinction values and continuum slopes of @xmath106 ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , the extremely red colors of these galaxies rules out @xmath19 solutions with any sensible sed . all sources with irac detections are thus removed from our sample of potential @xmath0 galaxies , reducing the sample to one single candidate in the hudf , previously reported in ( * ? ? ?
* see figure [ fig : stampz10 ] and table [ tab : phot ] ) .
the 16 removed sources are shown in the appendix in figure [ fig : stampscontamin ] and listed in table [ tab : photcontamin ] .
images of the only possible @xmath0 galaxy candidate are shown in figure [ fig : stampz10 ] .
the source is detected at 6.3 @xmath3 in @xmath4 ( measured in circular apertures of @xmath70 radius ) .
this is higher but completely consistent with the @xcite significance estimates , which are based on smaller apertures .
as can be seen , the source is not significantly detected in any other band .
its value of @xmath107 is very close , but just below the limit of @xmath91 .
this is mainly due to a 1.5@xmath3 flux excess in @xmath42 , which appears to be due to an extended structure in the background of that image .
when adopting smaller apertures , the @xmath85 value is found to be reduced , indicating also that this excess of flux is not associated with the source itself .
the spitzer irac 3.6@xmath108 m data shows some flux from a nearby source .
however , after subtraction of all the neighboring sources in irac , the candidate is undetected ( @xmath109 ) , with a @xmath76 upper limit on its irac [ 3.6 ] magnitude of @xmath110 mag ab ( gonzalez et al .
, in prep . ) .
this thus corresponds to @xmath101<1.6 $ ] at @xmath76 , which is much bluer ( by @xmath111 mag ) than the typical low - redshift contaminants that we culled from our sample ( see also next section ) .
after adding the newly acquired irac data from the iudf10 program ( spitzer proposal 70145 , pi : labbe ) to the goods irac data and removing neighboring sources , the @xmath0 candidate is also undetected at [ 4.5 ] ( @xmath112 ) providing added weight to the likelihood of it being at high rather than low reshift .
we derive the photometric redshift of the candidate using the code zebra @xcite with synthetic stellar population models from @xcite to which we added nebular continuum and line emission following , e.g. , @xcite . using the full 11 band fluxes and flux errors
, we derive a photometric redshift for this source of @xmath113 , with a likelihood of a low redshift solution at @xmath114 of @xmath115 .
the full spectral energy distribution ( sed ) of the source and its redshift likelihood function are shown in figure [ fig : jdropsed ] .
the best - fit sed corresponds to a very young , dust - free star - burst ( see also gonzalez et al . in prep . ) .
the best low redshift solution is found at @xmath116 , for an evolved , very low mass galaxy sed ( @xmath117 @xmath118 ) with moderate extinction .
interestingly , this sed is expected to be detected only at @xmath119 in @xmath7 , but it nevertheless has a significantly higher @xmath120 value ( 13.7 compared to @xmath121 ) .
deeper hst data shortward of the break would be extremely useful in order to constrain the possible non - detection of the source shortward of 1.4 .
as with other high - redshift catalogs , the added shorter - wavelength optical / near - ir data would play a key role in helping to further tighten its photometric redshift measurement .
candidate , previously reported in @xcite .
the measured photometry is shown with the red circle with errorbars and with 2@xmath3 upper limits in case of non - detections .
all fluxes and flux errors were used in the sed fit , however , even if they were negative .
the best fit sed is found at @xmath122 and is shown as a solid line ( @xmath123 ) , corresponding to a dust - free , young galaxy . the best template for a low redshift solution is also shown as the dashed green line .
this sed has a redshift of @xmath116 , and is passive with an age of 650 myr and of rather low mass ( only @xmath124 @xmath118 ) .
additionally , the low - z sed is reddened with @xmath125 mag . based on its larger @xmath120 value ( @xmath126 ) ,
this sed is formally excluded at @xmath127 probability , however .
the redshift probability function is shown in the inset in the upper left .
the sharp decrease above @xmath128 of @xmath129 is due to the use of a lf prior at @xmath130 and due to the fact that ly@xmath23 absorption starts affecting the @xmath4 band ( which would require the source to be brighter intrinsically ) . ]
the properties of the 16 intermediate brightness sources that did not pass the irac non - detection criteria are discussed in the appendix . from fitting their seds ,
these sources are found to be mostly massive galaxies ( @xmath131 @xmath118 ) with obscured but evolved stellar populations at @xmath8 .
interestingly , all these galaxies are essentially limited to @xmath9 mag .
this is @xmath11 mag brighter than the detection limits of our bright survey fields ( see figure [ fig : contaminmz ] in the appendix ) , and thus suggests that such very red galaxies have a somewhat peaked luminosity function .
this will have to be confirmed and quantified with future wide area data .
however , at face value , it would appear that contamination from such sources with @xmath6 , and with @xmath101\gtrsim2 $ ] is less problematic at fainter magnitudes .
we note , however , that contamination from similar galaxies with less extreme colors ( which may be more abundant also at @xmath132 mag ) may still be non - negligible due to photometric scatter ( see also [ sec : additionalcontamin ] ) .
furthermore , it is interesting to note that it is very difficult to construct clean @xmath0 galaxy samples purely based on hst data alone .
even if we were to increase the color criteria to @xmath133 , there would still be two contaminating galaxies in the sample together with our only viable @xmath0 candidate .
thus , also for future constraints on the bright end of the @xmath0 lf , it will be important to perform additional follow - up studies to validate the candidates , e.g. with spitzer .
this will be less of a concern for future @xmath27 galaxy samples , which can be obtained , e.g. , based on new f140w filter data .
in such a data set intrinsically red galaxies can be sorted out by requiring a blue continuum across f140w and @xmath4 . here , we only give a brief summary of the possible sample contamination . for a thorough discussion
we refer the reader to @xcite . essentially , the only probable chance for contamination is due to photometric scatter of a red lower redshift source .
we estimate this to be a 10% chance based on our photometric scatter experiments described in section [ sec : candsel ] , including the @xmath85 cuts .
other typical contaminants to lbg selections such as very cool dwarf stars and supernovae can essentially be excluded based on the relatively blue @xmath72 ( @xmath134 ) colors of stellar seds , on the fact that the source is detected in both the first and the second year of the hudf09 wfc3/ir data , and due to the fact that the candidate shows clear signs of an extended morphology .
additionally , it is very unlikely that this source is spurious , since the flux distribution in circular apertures randomly distributed over empty regions of the hudf @xmath4 image are nearly exactly gaussian , and the source is well detected at @xmath135 .
in this section we compute the expected abundance of @xmath0 galaxies in our dataset , and derive constraints on the @xmath0 lf based on our data . in order to estimate the number of sources we expect in our data from a given lf , we have to estimate the completeness as a function of magnitude @xmath136 and selection function as a function of redshift and magnitude @xmath137 .
following @xcite , this is done by inserting artificial galaxies in the observational data and rerunning the source detection with the exact same setup as for the original catalogs .
this is done for each of our fields individually .
two sets of simulations were run . in the first set , we follow @xcite , where the artificial galaxies are ` cloned ' from the @xmath31 dropout sample of the goods and hudf fields .
the images of these sources are adjusted for surface brightness dimming , the difference in angular diameter distance , as well as a size scaling of @xmath138 as observed for the lyman break galaxy population across @xmath139 ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these are then inserted in the observed images with galaxy colors as expected for star - forming galaxies between @xmath140 and @xmath141 . when computing the galaxy colors we assume a distribution of uv continuum slopes with @xmath142 ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . from these simulations ,
we compute the completeness as a function of observed @xmath96 magnitude for each field , taking into account the scatter and offsets between input and output magnitudes .
additionally , we compute the selection probabilities as a function of redshift and magnitude by measuring the fraction of sources that meet our selection criteria . by construction , galaxies are selected at @xmath143 at redshifts @xmath5 . in the second set of simulations , we repeat the above procedure . however ,
instead of using observed galaxy images , we use theoretical galaxy profiles from a mix of exponential ( sersic @xmath144 ) and de vaucouleur ( sersic @xmath145 ) profiles .
the size distribution is chosen to be log - normal , again with the same size scaling as a function of redshift .
the completeness and selection functions of our two procedures are in excellent agreement .
this demonstrates the reliability of our approach , which appears to be essentially independent of the adopted galaxy profiles ( unlike what has been claimed elsewhere , e.g. * ? ? ?
* but see also bouwens et al .
2010b ) .
candidates in the different fields assuming the lf to be as measured at @xmath146 ( top ; * ? ? ? * ) or as expected from extrapolating the @xmath11 trends to @xmath0 ( bottom ) .
the one detected source is in stark contrast to the @xmath147 sources expected if the lf was constant across the 170 myr from @xmath0 to @xmath12 .
the lf appears to evolve at a significantly accelerated pace with respect to the empirical evolution observed across @xmath11 .
6 sources at @xmath0 should be detected when extrapolating the lower redshift trends , while only one probable candidate is found.,title="fig : " ] candidates in the different fields assuming the lf to be as measured at @xmath146 ( top ; * ? ? ? * ) or as expected from extrapolating the @xmath11 trends to @xmath0 ( bottom ) .
the one detected source is in stark contrast to the @xmath147 sources expected if the lf was constant across the 170 myr from @xmath0 to @xmath12 .
the lf appears to evolve at a significantly accelerated pace with respect to the empirical evolution observed across @xmath11 .
6 sources at @xmath0 should be detected when extrapolating the lower redshift trends , while only one probable candidate is found.,title="fig : " ] the expected number of sources in a given magnitude bin @xmath148 can be estimated for any given lf , @xmath149 , through : @xmath150)\ ] ] the result of this calculation is shown in figure [ fig : nexp ] for each individual field using the observed lfs at @xmath12 @xcite , as well as that expected at @xmath0 .
the latter is based on extrapolating the evolution of the schechter function parameters presented in @xcite , which is based on fitting the likelihood contours of lbg lfs across @xmath11 . in particular ,
significant evolution is only seen in the characteristic luminosity @xmath151 , which is found to dim by 0.33 mag per unit redshift .
the other two schechter function parameters were kept constant , in agreement with the small evolutionary trends that were found at low significance in @xcite . for reference , the final parameter evolution we use in this work is : @xmath152 @xmath153 @xmath154 as can be seen in figure [ fig : nexp ] , the wide - area , shallow data of candels and ers are very useful for constraining the evolution in the bright end of the lf .
in particular , if there was no evolution in the lf from @xmath0 to @xmath12 , the wide area data should contain 8.5 @xmath0 sources . in total
, we would expect to see @xmath147 sources if the lf was unchanged over the 170 myr from @xmath12 to @xmath0 .
this implies that the wide area data provides essentially 30% of the total search power in case of an lf evolution in @xmath155 only .
given that we only detect one probable candidate , the lf appears to drop faster than expected from the empirical lower redshift extrapolation . in particular
, we do not detect the @xmath156 galaxies that we would have expected to find at @xmath0 if the lower redshift trends remained valid . from these extrapolations we predicted to find three sources in the hudf ( consistent with the expectations from * ? ? ? * ) and about one in each of the two hudf09 parallel fields . the poissonian probability to find @xmath157 source , given that 6 are expected is @xmath158 .
this remains significant , even after including cosmic variance , which adds an additional uncertainty on these low number counts of about @xmath159 for an individual wfc3/ir pointing ( see e.g. , * ? ? ?
* ; * ? ? ?
we derive an upper limit of 6% to the probability of finding @xmath157 source in our search area , based on the cosmic variance calculator of @xcite and combining the number counts uncertainty in the different fields assuming the final distribution is gaussian ( justified by the central limit theorem ) .
therefore , the detection of an accelerated evolution relative to the low - redshift extrapolation is significant at @xmath160 .
the accelerated evolution in the uv lf can also be seen from our constraints on the @xmath0 lf .
the stepwise lf is computed using an approximation of the effective selection volume as a function of observed magnitude @xmath161 .
the lf in bins of absolute magnitude is then given by @xmath162 .
this is shown in figure [ fig : lfevol ] , where the lf was evaluated in bins of 0.5 mag , and non - detections correspond to @xmath163 upper limits including the effects of 50% cosmic variance per pointing . as can be seen in figure [ fig : lfevol ] , the current constraints on the @xmath0 lf are significant even at bright magnitudes where the wide area data are particularly valuable .
these data sets reduce the upper limits by more than an order of magnitude relative to using only the hudf09 fields , therefore indicating that the @xmath0 lf at @xmath164 drops by a factor @xmath165 with respect to the observed lf at @xmath12 .
however , from these shallow data sets , no constraints can be obtained on any accelerated evolution of the uv lf around @xmath166 .
this only becomes apparent at @xmath167 ( corresponding to @xmath51 mag ) .
such faint limits are only probed by the hudf09 data set . in particular , at @xmath168 , the upper limit on the lf is a factor @xmath169 below the expectation ( fig . [
fig : lfevol ] ) .
thus , it is clear that data reaching to deeper than @xmath170 mag will be necessary to further constrain the drop in the lf in the future ( which is beyond the reach of the current mct programs ) . in order to quantify the change in the lf from @xmath0 to @xmath12 more robustly , we consider two possible scenarios .
first , we assume the accelerated lf evolution occurs only in @xmath151 , at a constant rate since @xmath15 , and we fit the poissonian likelihood for the observed number of sources .
this can be written as @xmath171 @xmath172 , where @xmath173 runs over all fields , and @xmath174 runs over the different magnitude bins , and @xmath175 is the poissonian probability .
our extrapolation of the uv lf is a modification of the fitting formulae of @xcite .
we thus use @xmath176 mpc@xmath177const .
and @xmath178const . ,
and assume : @xmath179 .
we then fit for @xmath180 , finding @xmath181 , which results in an estimate for @xmath182 mag .
alternatively , we assume @xmath183const .
( as derived for @xmath12 from our empirical extrapolation ) , @xmath184 , and we fit only for an evolution in the normalization with redshift relative to the @xmath12 lf .
this results in @xmath185 mpc@xmath186 , with best fit @xmath187 .
thus , using this extrapolation , the normalization of the uv lf from @xmath0 to @xmath12 is expected to increase by a factor 12 .
these results are summarized in table [ tab : summary ] .
lf from our combined data set , evaluated in 0.5 mag bins .
the upper limits correspond to @xmath163 poissonian limits including the additional uncertainty of 50% cosmic variance per pointing .
it is clear that the luminosity function evolves strongly from @xmath12 to @xmath0 , as our upper limits are a factor @xmath12 - 5 below the measured @xmath12 lf of ( orange data ; * ? ? ?
the expected @xmath0 lf as extrapolated from fits to lower redshift lbg lfs is shown as a dashed red line . using this lf
, we would expect to detect six sources in the full data set ( @xmath188 in the hudf , and @xmath189 in each of the hudf09 parallels ; see figure [ fig : nexp ] ) .
for comparison also the @xmath31 and @xmath15 lfs are plotted @xcite , showing the dramatic build - up of uv luminosity across @xmath190 gyr of cosmic time .
the light gray vectors along the lower axis indicate the range of luminosities over which the different data sets dominate the @xmath0 lf constraints . ] above @xmath191 mag ( @xmath192 ) .
the filled circle at @xmath193 is the luminosity density directly measured for our only @xmath0 galaxy candidate .
the two connected dots at @xmath74 show the range of possible ld values , given the two simple , accelerated extrapolations of the uv luminosity function described in section [ sec : lfconstraints ] .
the red line corresponds to the empirical lf evolution from @xcite .
its extrapolation to @xmath130 is shown as dashed red line .
the @xmath194 data at @xmath11 is taken from @xcite .
as can be seen , @xmath194 increases by more than an order of magnitude in the 170 myr from @xmath0 to @xmath12 , indicating that the galaxy population at this luminosity range evolves by a factor @xmath195 more than expected from low redshift extrapolations .
the predicted @xmath194 evolution of the semi - analytical model of @xcite is shown as dashed blue line , and the theoretical model prediction of @xcite is shown as blue solid line . these reproduce the expected luminosity density at @xmath0 remarkably well . ]
ccccccc @xmath196 ( fixed ) & @xmath197 & @xmath198 ( fixed ) & @xmath199 & @xmath200 + @xmath201 & @xmath202 ( fixed ) & @xmath198 ( fixed ) & @xmath203 & @xmath204 + & @xmath205 & @xmath206 the quantity most easily comparable between observations and simulations is the observed luminosity density ( @xmath194 ) above a given limiting magnitude , which is shown in figure [ fig : ldevol ] .
the ld from the @xmath0 candidate alone amounts to @xmath207 erg@xmath208s@xmath47hz@xmath47mpc@xmath186 .
however , a more realistic estimate of the luminosity density can be obtained from the two possible extrapolations of the uv lf we derived in the previous section , which include the contribution from galaxies at @xmath209 mag that are currently undetected .
assuming the lf evolution only occurs in the characteristic luminosity , we find for the @xmath0 luminosity density @xmath210 erg@xmath208s@xmath47hz@xmath47mpc@xmath186 , while assuming the evolution to be driven by a normalization of the schechter function only , we find @xmath211 erg@xmath208s@xmath47hz@xmath47mpc@xmath186 .
these different estimates are also summarized in table [ tab : summary ] .
given that the observed luminosity density at @xmath12 is @xmath212 erg@xmath208s@xmath47hz@xmath47mpc@xmath186 @xcite , the inferred increase in luminosity density in the 170 myr from @xmath0 to @xmath12 amounts to more than an order of magnitude .
this is a factor @xmath195 higher than what would have been inferred from the empirical relation for the uv lf evolution , which predicts an increase by only a factor @xmath169 .
note however , that such a rapid increase in luminosity density is actually predicted by many theoretical models . in figure
[ fig : ldevol ] we also show the luminosity densities at @xmath213 mag as derived from the semi - analytical model ( sam ) of @xcite , and from the theoretical model of @xcite .
although there is still some discrepancy on the exact shape of the @xmath214 uv lf between the @xcite sam and the observations , the integrated luminosity density and its evolution is remarkably well reproduced across the full redshift range @xmath11 ( see also discussion in * ? ? ?
the semi - analytic model predicts a constant growth in @xmath194 with cosmic time at somewhat faster pace than observed , leading to some discrepancy between the observations and the model at @xmath105 and @xmath12 , where the observational data shows higher luminosity densities .
however , the model reproduces our estimates of the @xmath0 ld remarkably well .
similar conclusions are reached for the purely theoretical model of @xcite . in particular , this model predicts the uv ld to evolve at an accelerated rate at @xmath130 , being only @xmath215 erg@xmath208s@xmath47hz@xmath47mpc@xmath186 at @xmath0 , in excellent agreement with our observed estimates . since the model is only based on the evolution of the underlying dark matter mass function , this indicates that an accelerated evolution in the galaxy population can be explained even without the need for a change in the physical mechanisms of galaxy formation .
for further theoretical model predictions , see also e.g. @xcite , or @xcite .
it is also interesting to note that the star - formation rate densities ( see table [ tab : summary ] ) inferred from our data , are more than an order of magnitude too low to account for the stellar mass densities observed at @xmath105 in systems of similar brightness . with constant star - formation over @xmath216 myr from @xmath0 to @xmath105 the observed galaxy population would only produce a stellar mass density of @xmath217mpc@xmath186 , compared to @xmath218mpc@xmath186 as estimated by , e.g. , @xcite .
if the inferred mass densities and sfrs are correct , this suggests that the majority of the stars found in @xmath105 galaxies down to @xmath219 mag have to be formed in systems below our detection limit at @xmath0 or are younger than 300 myr .
in this paper , we have extended our search for @xmath0 galaxies to @xmath220 arcmin@xmath2 of public wfc3/ir data obtained around the goods south field ( see figure [ fig : fields ] ) .
these data sets have been acquired through the three surveys hudf09 , ers , and candels , and reach to varying depths , from @xmath68 to @xmath221 .
based on strict optical non - detection requirements and a color cut of @xmath6 , we search these fields for lyman break @xmath7-dropout galaxies , which are expected to lie at @xmath80 .
a total of 17 sources satisfy these criteria .
however , 16 out of these sources show strong irac detections which rule out their being at such very high redshifts .
rather , these galaxies are found to have best - fit photometric redshifts in the range @xmath222 ( see section [ sec : contaminants ] and appendix ) .
they remain undetected in the optical due to their evolved stellar populations with non - negligible dust obscuration .
this shows how important spitzer irac data is for removing contaminating lower redshift galaxies .
interestingly , these contaminants are essentially only detected with magnitudes in the range @xmath223 mag .
this is @xmath190 mag brighter than the detection limits of our bright surveys , which suggests that such evolved and dusty galaxies follow a peaked luminosity function at these wavelengths .
if confirmed by future wide - area wfc3/ir data sets , this would indicate that such extremely red systems are not as much of a problem for @xmath28 searches at fainter levels as has been expected to date .
however , the existence of such galaxy populations will make it challenging to use large - area wfc3 surveys such as pure - parallel fields ( e.g. * ? ? ? * ) for constraining the bright end of the @xmath0 uv lf without additional follow - up observations to validate the candidates .
even with our expanded search area , the only @xmath224 detected galaxy with a color limit of @xmath101<1.6 $ ] ( @xmath76 ) and thus the only possible @xmath0 galaxy candidate is the same source that we reported already in @xcite . in appendix
[ sec : possiblez10 ] we additionally note one other lower s / n candidate that is suggestive of being at comparable redshift , but requires confirmation from further hst wfc3/ir data . interestingly , we would have expected to detect six @xmath0 galaxies in our data , if the uv lf evolved to @xmath0 as expected from lower redshift trends ( see [ sec : nexp ] ) .
thus , the galaxy population appears to evolve at an accelerated rate beyond @xmath130 .
we infer that the uv luminosity density increases by more than an order of magnitude in only 170 myr from @xmath0 to @xmath12 , and we are thus likely witnessing the first massive build - up of the galaxy population at these early epochs in the reionization era . the fact that theoretical models based on the evolution of dark matter halos do , in fact , predict such an accelerated increase in the ld , indicates that these rapid changes are mainly driven by an accelerated evolution of the underlying dark matter mass function , rather than due to a change in star - formation properties of these early galaxies ( see e.g. , * ? ? ?
* ; * ? ? ?
the accelerated evolution of the galaxy population also has interesting consequences for cosmic reionization by galaxies brighter than @xmath225 mag , the current detection limits .
with such a sharp decrease in the luminosity density above @xmath12 , it is impossible for such bright galaxies alone to create a reionization history in agreement with the high optical depth measurement of wmap and the vast majority of the ionizing flux has to be created by fainter galaxies ( see also * ? ? ? * ) .
unfortunately , our knowledge of galaxies at @xmath130 still remains very uncertain .
however , wfc3/ir offers unique opportunities to make significant progress in expanding the number of galaxies at @xmath79 and addressing some of the key issues related to early galaxy formation and its impact on reionization , even before the advent of jwst . in figure
[ fig : nrequired ] , we show the depth required to detect ten @xmath27 and ten @xmath0 galaxies for a given survey area .
such future data sets will have to reach significantly fainter than @xmath226 mag to accomplish this goal , even for large surveyed areas of 50 wfc3/ir fields .
this is deeper than currently planned wide - area wfc3/ir data ( including mct programs ) .
in fact , to characterize the luminosity function at luminosities below l@xmath227 and to set better constraints for reionization , surveys to fainter than 29 ab mag are really needed .
note that with only one wfc3/ir field , a magnitude of @xmath228 has to be reached to detect a significant @xmath0 population , which is fainter than is practical with hst .
based on the current lf constraints , we find that imaging a few fields provides the best chance to improve on current @xmath0 constraints .
furthermore , in order to further constrain the possible accelerated evolution of the uv lf , the @xmath27 regime offers the best opportunity .
a sizable population of @xmath27 galaxies is expected to be seen already down to @xmath229 mag over multiple wfc3/ir fields , which can be achieved with the efficient f140w filter .
thus , it is likely that already with wfc3/ir , we can soon push the frontier of statistical galaxy samples with wfc3/ir from @xmath12 another @xmath230 myr back out to @xmath79 .
galaxies ( blue shaded region and dashed line ) and ten @xmath0 galaxies ( red shaded region and dashed line ) as a function of survey area .
these regions show the expected number of sources calculated using our lf extrapolations to @xmath0 from section [ sec : lfconstraints ] , i.e. assuming only evolution in @xmath151 or only in @xmath155 to reproduce the detection of our one @xmath0 candidate .
the dashed lines show the same , but assuming the standard lower - redshift extrapolation of the uv lf @xcite . in particular , at @xmath0
the estimated depths differ significantly .
the range of @xmath231 values of these extrapolations is indicated with an errorbar in the lower left .
as can be seen , surveys reaching significantly deeper than @xmath226 mag will be required to detect a significant population of @xmath130 galaxies .
the vertical dashed lines indicate the area of one and 50 wfc3/ir fields , respectively .
a survey with only one pointing would need to reach to @xmath228 mag in @xmath96 to significantly constrain the @xmath0 galaxy population , which is out of reach with hst wfc3/ir .
therefore , multiple fields are favorable for searching for @xmath0 galaxies . additionally , comparably deep optical data ( @xmath232 mag ) and deep irac imaging would be required over such fields in order to robustly exclude low - redshift contaminants .
the identification of @xmath27 galaxies would benefit from imaging in different filters ( e.g. , f140w ) than adopted in current deep wfc3/ir fields . ]
facilities : . , s. v. w. , et al .
2006 , , 132 , 1729 , e. , & arnouts , s. 1996 , , 117 , 393 , r. , broadhurst , t. , & illingworth , g. 2003 , , 593 , 640 , r. j. , illingworth , g. d. , blakeslee , j. p. , broadhurst , t. j. , & franx , m. 2004 , , 611 , l1 , r. j. , illingworth , g. d. , franx , m. , & ford , h. 2007 , , 670 , 928 , r. j. , illingworth , g. d. , thompson , r. i. , & franx , m. 2005 , , 624 , l5 , r. j. , et al .
2009 , , 705 , 936 . 2010 , , 709 , l133 . 2010 , , 708 , l69 . 2011a , , 469 , 504 . 2011b , , 737 , 90 . 2011c , arxiv e - prints , 1105.2038 , g. , & charlot , s. 2003 , , 344 , 1000 , a. , et al .
2010 , , 409 , 855 , d. , armus , l. , bohlin , r. c. , kinney , a. l. , koornneef , j. , & storchi - bergmann , t. 2000 , , 533 , 682 , g. d. , wu , c. , & weedman , d. w. 1980 , , 43 , 393 , m. , labb , i. , franx , m. , van dokkum , p. g. , taylor , e. n. , & gawiser , e. j. 2009 , , 690 , 937 , m. , giavalisco , m. , & goods team . 2003 , in the mass of galaxies at low and high redshift , ed .
r. bender & a. renzini , 324+ , j. s. , mclure , r. j. , robertson , b. e. , ellis , r. s. , stark , d. p. , cirasuolo , m. , & de ravel , l. 2011 , arxiv e - prints , 1102.5005 , r. , et al . 2006 , , 372 , 565 , h. c. , et al .
2004 , , 600 , l107 , s. l. , papovich , c. , giavalisco , m. , reddy , n. a. , ferguson , h. c. , koekemoer , a. m. , & dickinson , m. 2010 , , 719 , 1250 , m. , et al .
2004 , , 600 , l93 , v. , labbe , i. , bouwens , r. , illingworth , g. , franx , m. , & kriek , m. 2010 , arxiv e - prints , 1008.3901 , a. , et al .
2010 , arxiv e - prints , 1011.6569 , n. a. , et al .
2011 , arxiv e - prints , 1105.3753 , a. l. , malkan , m. a. , colbert , j. w. , siana , b. , teplitz , h. i. , & mccarthy , p. 2008
, , 680 , l97 , a. l. , et al .
2009 , , 697 , 1128 , i. , ohta , k. , tamura , n. , akiyama , m. , aoki , k. , ando , m. , kiuchi , g. , & sawicki , m. 2007 , , 376 , 1557 , a. m. , et al .
2004 , , 600 , l123 , a. m. , et al .
2011 , arxiv e - prints , 1105.3754 , e. , et al .
2011 , , 192 , 18 , i. , et al .
2010 , , 716 , l103 . 2010 , , 708 , l26 , c. g. , baugh , c. m. , frenk , c. s. , & benson , a. j. 2011 , , 45 lorenzoni , s. , bunker , a. j. , wilkins , s. m. , stanway , e. r. , jarvis , m. j. , & caruana , j. 2011 , , 414 , 1455 mao , j. , lapi , a. , granato , g. l. , de zotti , g. , & danese , l. 2007 , , 667 , 655 , r. j. , cirasuolo , m. , dunlop , j. s. , foucaud , s. , & almaini , o. 2009 , , 395 , 2196 , r. j. , dunlop , j. s. , cirasuolo , m. , koekemoer , a. m. , sabbi , e. , stark , d. p. , targett , t. a. , & ellis , r. s. 2010 , , 403 , 960 , b. , et al .
2005 , , 635 , 832 muoz , j. a. 2011 , arxiv:1104.0927 , p. a. , et al .
2007 , , 671 , 1212 . 2009 , , 690 , 1350 . 2010 , , 709 , l21 . 2010 , , 714 , l47 . 2010 , , 709 , l16 , j. b. , & gunn , j. e. 1983 , , 266 , 713 , m. , et al .
2004 , , 611 , 660 , m. , theuns , t. , & lacey , c. 2011 , , 410 , 775 , j. , stark , d. p. , ellis , r. s. , george , m. r. , egami , e. , kneib , j. , & smith , g. p. 2008
, , 685 , 705 , b. e. 2010 , , 713 , 1266 , g. , cimatti , a. , franceschini , a. , brusa , m. , fritz , j. , & bolzonella , m. 2007 , , 470 , 21 salvaterra , r. , ferrara , a. , & dayal , p. 2011
, , 414 , 847 , p. , et al .
2009 , , 504 , 751 , m. , & thompson , d. 2006 , , 642 , 653 , d. , & de barros , s. 2009 , , 502 , 423 . 2010 , , 515 , a73 + , e. r. , mcmahon , r. g. , & bunker , a. j. 2005 , , 359 , 1184 , d. p. , ellis , r. s. , richard , j. , kneib , j. , smith , g. p. , & santos , m. r. 2007 , , 663 , 10 , a. m. , papovich , c. , & eisenstein , d. j. 2008 , , 677 , 828 , m. , & stiavelli , m. 2008 , , 676 , 767 , m. , stiavelli , m. , bouwens , r. j. , oesch , p. , shull , j. m. , illingworth , g. d. , bradley , l. d. , & carollo , c. m. 2010 , , 714 , l202 , m. , et al .
2011 , , 727 , l39 + , t. , dickinson , m. , ferguson , h. c. , giavalisco , m. , mobasher , b. , grogin , n. a. , & panagia , n. 2008 , , 676 , 781 , s. m. , bunker , a. j. , ellis , r. s. , stark , d. , stanway , e. r. , chiu , k. , lorenzoni , s. , & jarvis , m. j. 2010 , , 403 , 938 wilkins , s. m. , bunker , a. j. , lorenzoni , s. , & caruana , j. 2011 , , 411 , 23 wilkins , s. m. , bunker , a. j. , stanway , e. , lorenzoni , s. , & caruana , j. 2011 , arxiv:1106.5977 , r. a. , et al .
2011 , , 193 , 27 , h. , windhorst , r. a. , hathi , n. p. , cohen , s. h. , ryan , r. e. , oconnell , r. w. , & mccarthy , p. j. 2010 , research in astronomy and astrophysics , 10 , 867
here we briefly summarize the properties of the galaxies that formally met the @xmath7-dropout criteria based on the hst data alone , but which were found to exhibit very red @xmath4 to irac colors .
this makes it extremely unlikely that these galaxies are at @xmath19 .
a summary of these sources is listed in table [ tab : photcontamin ] .
note that all these galaxies are obviously clearly detected in the irac images , with the exception of one source ( jd-2080646581 ) , which is close to a quintet of irac bright galaxies .
however , after subtraction of the contaminating flux from its neighbors based on a careful convolution of the @xmath4 image to the irac psfs ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) it also shows a clear detection ( see fig .
[ fig : stampscontamin ] ) .
the best estimate for its color is @xmath101=1.6\pm0.5 $ ] , which is too red for a likely @xmath28 source .
interestingly , six of these 16 sources are present in catalogs of potential passive high redshift galaxies of @xcite and @xcite
. one of these is also detected in x - ray emission and classified as a so - called exo source hosting a potentially highly obscured agn , particularly if it is an evolved galaxy at intermediate redshift , as pointed out by @xcite .
we have derived photometric redshifts and mass estimates for these galaxies by complementing our hst photometry with the irac fluxes from the goods music catalog @xcite or from the simple images . based on sed fits with @xcite models ,
these galaxies are confirmed to be evolved , reddened systems with stellar masses around @xmath233 @xmath118 at redshifts @xmath234 .
note that with the exception of four sources these are all detected individually in mips 24@xmath100 data . .
the surface densities are extremely low .
only @xmath10.4 sources are expected per wfc3/ir pointing .
the rapid disappearance of such sources beyond @xmath235 , almost one mag brighter than the detection limits of our shallow fields from ers and candels , suggests that such sources may have a peaked lf at the wavelengths of interest in these surveys and so could be less of a problem for contamination for fainter @xmath0 galaxy samples . ] in figure [ fig : contaminmz ] we additionally show the surface density of these evolved intermediate redshift sources . even at their peak surface density ,
they are only found at 0.06 sources per magnitude per arcmin@xmath2 .
therefore , only @xmath236 sources are expected per wfc3/ir pointing , providing an explanation for why none of these bright sources are found in the three deep hudf09 fields .
the dearth of such galaxies faintward of @xmath237 , at @xmath190 mag brighter than the completeness limit of our bright surveys , suggests that such extremely red sources become even less frequent ( i.e. , that their lf may have peaked at these magnitudes ) and so they could be less problematic for contamination of fainter @xmath19 dropout searches .
this will have to be confirmed , however , with future wide - area wfc3/ir data , searching explicitly for this type of galaxies .
note that our finding of a disappearance of passive intermediate redshift galaxies to fainter magnitudes is in agreement with @xcite , who found an absence of passive , red galaxies at @xmath238 at masses below @xmath239 @xmath118 .
lcccccccccc + jd-2162843432 & 03:32:16.28 & -27:43:43.2 & @xmath240 & @xmath241 & 14.6 & 0.65 & 70081 & + jd-2188742241 & 03:32:18.87 & -27:42:24.1 & @xmath242 & @xmath243 & 12.2 & 0.29 & 70040 & + jd-2226644214 & 03:32:22.66 & -27:44:21.4 & @xmath242 & @xmath244 & 10.9 & 0.26 & 70104 &
+ + jd-2487849357 & 03:32:48.78 & -27:49:35.7 & @xmath245 & @xmath246 & 11.4 & 0.33 & 70314 & + jd-2532547516 & 03:32:53.25 & -27:47:51.6 & @xmath247 & @xmath248 & 11.2 & 0.26&70236 & 4 + jd-2304648166 & 03:32:30.46 & -27:48:16.6 & @xmath249 & @xmath250 & 29.4 & 0.23 & 70258 & 1 + jd-2387448399 & 03:32:38.74 & -27:48:39.9 & @xmath251 & @xmath252 & 14.9 & 0.33 & 70273 & 1,2,4 + jd-2412344008 & 03:32:41.23 & -27:44:00.8 & @xmath253 & @xmath254 & 11.0 & 0.22 & 70092 & + jd-2158349541 & 03:32:15.83 & -27:49:54.1 & @xmath255 & @xmath256 & 19.1 & 0.38 & 70316 & 1 + jd-2249748085 & 03:32:24.97 & -27:48:08.5 & @xmath257 & @xmath258 & 7.5 & 0.33 & 70252 & + jd-2080646581 & 03:32:08.06 & -27:46:58.1 & @xmath259 & @xmath260 & 6.1 & 0.16 & &
+ + jd-2489152264 & 03:32:48.91 & -27:52:26.4 & @xmath245 & @xmath261 & 12.3 & 0.26 & 70442 &
+ jd-2358952367 & 03:32:35.89 & -27:52:36.7 & @xmath262 & @xmath263 & 7.2 & 0.35 & 70455 & 1 + jd-2351353198 & 03:32:35.13 & -27:53:19.8 & @xmath264 & @xmath258 & 5.8 & 0.35 & 70484 &
+ jd-2331152057 & 03:32:33.11 & -27:52:05.7 & @xmath265 & @xmath266 & 5.8 & 0.20 & 70429 & 1,3 + jd-2211356269 & 03:32:21.13 & -27:56:26.9 & @xmath255 & @xmath267 & 7.9 & 0.44 & & mag , but with strong irac detections .
this rules out the possibility that these sources are at @xmath28 .
the images are 7.5 arcsec on a side , and show from left to right ( 1 ) a stack of @xmath40 , @xmath41 , @xmath42 , @xmath43 , ( 2 ) @xmath7 , ( 3 ) @xmath4 , ( 4 ) spitzer irac [ 3.6 ] , and ( 5 ) irac [ 8.0 ] .
all these sources are clearly detected in all irac bands , including [ 8.0 ] .
thus , their @xmath268 $ ] colors are greater than @xmath269 , which is very different from the relatively flat color expected for a real @xmath0 source .
note that irac flux of the source jd-2080646581 is heavily contaminated by neighboring sources . in the [ 3.6 ] band ,
we therefore show the cleaned image , after subtraction of the contaminating flux , which reveals its clear detection . ]
as pointed out in section [ sec : discussion2 ] , due to the apparent paucity of such sources , it will be difficult to extend the @xmath0 candidate lists to large numbers of reliable sources in the near future with hst . in cycle 18 ,
an additional 128 orbit imaging survey was granted ( pi : ellis ) to further study the @xmath25 galaxy population over the hudf data .
however , given that only one single field will be targeted and that only a relatively small amount of @xmath4 data is currently planned to be taken ( 22 new orbits , relative to the @xmath158 already available from the hudf09 and candels programs ) , we can forecast what this program will bring in terms of @xmath0 galaxy science .
essentially , every 5@xmath3 source that will be obtained as a result of this new program should already be present as a 4.2@xmath3 source in the current @xmath4 band data .
we therefore systematically searched for lower significance sources ( down to 4@xmath3 detections ) that satisfy our @xmath0 dropout criteria .
essentially , the only additional candidate that we found is already detected at 4.8@xmath3 in the current data , and is located at ra@xmath27003:32:43.01 , dec@xmath270 - 27:46:53.3 ( see also * ? ? ?
note , however , that the @xmath42-band data shows very faint positive flux ( @xmath119 ) exactly at the location of this source , which increases the chance that it is at lower redshift .
this demonstrates the value of even deeper optical data over these fields for more robust high redshift galaxy selections .
given the scarcity of low significance candidates in the hudf , it is clear , however , that it will be very important to image at least one additional field to similar depth in order to further constrain the accelerated evolution in the cosmic star - formation rate density ( see fig .
[ fig : nrequired ] ) . | we search for @xmath0 galaxies over @xmath1160 arcmin@xmath2 of wfc3/ir data in the chandra deep field south , using the public hudf09 , ers , and candels surveys , that reach to 5@xmath3 depths ranging from 26.9 to 29.4 in @xmath4 ab mag .
@xmath5 galaxy candidates are identified via @xmath6 colors and non - detections in any band blueward of @xmath7 .
spitzer irac photometry is key for separating the genuine high - z candidates from intermediate redshift ( @xmath8 ) galaxies with evolved or heavily dust obscured stellar populations . after removing 16 sources of intermediate brightness ( @xmath9 mag ) with strong irac detections ,
we only find one plausible @xmath0 galaxy candidate in the whole data set , previously reported in bouwens et al .
( 2011 ) .
the newer data cover a @xmath10 larger area and provide much stronger constraints on the evolution of the uv luminosity function ( lf ) .
if the evolution of the @xmath11 lfs is extrapolated to @xmath0 , six @xmath0 galaxies are expected in our data .
the detection of only one source suggests that the uv lf evolves at an accelerated rate before @xmath12 .
the luminosity density is found to increase by more than an order of magnitude in only @xmath13 myr from @xmath0 to @xmath12 .
this increase is @xmath14 larger than expected from the lower redshift extrapolation of the uv lf .
we are thus likely witnessing the first rapid build - up of galaxies in the heart of cosmic reionization .
future deep hst wfc3/ir data , reaching to well beyond 29 mag , can enable a more robust quantification of the accelerated evolution around @xmath0 . |
it is well established that the measurement of the galaxy luminosity function ( lf ) is sensitive to the type of galaxy being sampled .
morphologically early - type galaxies tend to be systematically brighter than their late - type counterparts , resulting in lf estimates with significantly brighter @xmath9 and shallower faint end slope @xmath2 ( e.g. efstathiou , ellis & peterson 1988 ; loveday et al .
1992 ; blanton et al .
similar trends are also present if one selects galaxies based on h@xmath2 equivalent widths ( loveday , tresse & maddox 1999 ) , [ oii ] equivalent widths ( ellis et al .
1996 ) or colour ( lin et al . 1996 ; marzke & da costa 1997 ) .
understanding and being able to quantify this variation in the lf is of great importance to a full understanding of galaxy formation and evolution . up until now the greatest obstacle faced when attempting to determine how the lf varied with different galaxy properties was the fact that one was required to divide already small data sets , thereby losing much of the statistical significance in the lf estimations . in this paper
we address this issue by making use of a subset of the galaxies observed to date in the 2df galaxy redshift survey ( 2dfgrs maddox 1998 ; colless 2001 ) .
this subset is the largest single data set used in the calculation of lfs and can easily be divided several times whilst still maintaining very precise statistics .
the 2dfgrs is a joint uk - australian effort to map the distribution of galaxies down to an extinction corrected @xmath0 magnitude of @xmath10 ( median redshift @xmath11 ) . in
so doing we expect to obtain 250,000 galaxy spectra from which redshifts can be determined .
this is a factor of 10 more than any previous redshift survey . in order to divide our data set in a meaningful way we develop in this paper a classification of the galaxies based upon their observed spectra . as with other astronomical data , there are different approaches for analysing galaxy spectra ( see e.g. lahav 2000 and references therein ) .
if one has a well defined physical model for galaxy spectra , then it is appropriate to estimate parameters of interest ( such as age and star - formation rate ) e.g. by maximum likelihood , directly using all the spectral bins , or via a compressed version of the data designed to give maximum information on the physical parameters of interest ( e.g. heavens , jimenez & lahav 2000 ) .
if , on the other hand , one prefers to let the data ` speak for themselves ' , in a model - independent way , then it is more useful to look at the distribution of the galaxies in the high - dimensional space defined by the spectral bins .
it is then possible to either look for distinct groups , e.g. early and late types , or more refined classes ( e.g. slonim et al .
an alternative , which we present here , is to find a continuous ( sequence - like ) parameterisation of the spectral features , which can later be divided into subsets .
the parameterisation we develop is denoted by @xmath1 and essentially represents a measure of the average absorption or emission line strength present in each galaxy s spectrum .
this classification is robust to the known instrumental uncertainties and also has the advantage of being easily interpreted in terms of the current star - formation present in each galaxy .
the set of galaxies we use in this paper comprises a sample of 75,000 galaxy spectra .
this is more than 12 times larger than the one used in our previous analysis ( folkes et al .
1999 , hereafter f99 ) and is by far the largest sample used to date . as well as showing the latest lf determinations we will also present a new procedure for calculating self - consistent @xmath8-corrections and investigate possible fibre - aperture biases .
the outline of this paper is as follows .
section [ section : data ] briefly summarises the 2dfgrs and describes the data set we are using , section [ section : spectypes ] outlines our method for dividing the data set based on our parameterisation of the galaxy spectral type , @xmath1 .
section [ section : lumfunctions ] gives a detailed description of the calculations involved in estimating the lf and in section [ section : discussion ] we discuss our results and future work .
the 2dfgrs has already observed approximately 200,000 unique galaxies for which it has obtained redshifts .
the survey , once complete , will cover approximately 2000@xmath12 on the sky , split between two independent strips : one in the northern galactic hemisphere and the other centred roughly on the south galactic pole .
in addition to this there are 99 random fields in the southern galactic cap .
the 2df instrument itself is capable of observing up to 400 galaxy spectra simultaneously ( lewis et al . , 2001 ) .
each galaxy has been selected from a revised and extended version of the automated plate measuring ( apm ) galaxy catalogue ( maddox et al .
1990 ) in order to determine its position and magnitude .
it is then observed as part of the survey through a 140@xmath13 m ( @xmath14 arcsec ) diameter optical fibre .
the @xmath15 magnitudes we use in this analysis are total magnitudes derived from updated versions of the original apm scans .
these magnitudes have been updated to take into account new ccd calibration data ( see colless et al .
2001 and norberg et al . , 2001 ) and are believed to have an rms error of approximately @xmath16 magnitudes .
each of the spectra observed using the 2df instrument spans 1024 channels with a spectral scale of 4.3@xmath17 per pixel ; the fwhm is measured by arc lines to be of the order of 1.8 - 2.5 pixels .
typically at the survey limit the observed spectra have an average signal - to - noise ratio of @xmath18 per pixel , sufficient for determining redshifts and performing spectral analyses . for the purposes of our spectral analysis
we have restricted ourselves to the redshift range of @xmath19 so that we are left with a uniform sample of galaxy spectra in the rest - frame wavelength range spanning [ oii ] to h@xmath2 .
a detailed outline of the spectral reduction pipeline is given in colless et al .
( 2001 ) and f99 which we follow with only minor modifications .
the data that we have available for this analysis include all observations up until january 2001 . at this stage a total of 111404 and 60062 galaxy spectra had been obtained in the sgp and ngp regions respectively . from these objects we remove those which did not receive accurate redshifts ( @xmath20 , colless et al .
2001 ) , repeated observations and spectra with particularly low signal - to - noise ratio ( @xmath21 ) .
this leaves us with 78994 ( sgp ) and 44937 ( ngp ) galaxies for use in our spectral analysis .
because of interference from sky emission and atmospheric absorption we have made a further restriction to our redshift range such that @xmath22 .
this cut ensures that the h@xmath2 line in our spectra is not corrupted .
note that this is only a problem for the small redshift range @xmath23 , however , we have also excluded the galaxies with @xmath24 in order to simplify our analysis . imposing this cut and removing spectra observed in poor quality fields ( section 4.1 ) leaves us with 43449 ( sgp ) and 32140 ( ngp ) galaxies .
a total of 75589 unique galaxies will therefore be used in our subsequent analysis and measurement of the lf .
the spectral classification presented here is based upon a principal component analysis ( pca ) of the galaxy spectra .
pca is a statistical technique which has been used with considerable success by multiple authors in the past ( e.g. connolly et al .
1995 ; folkes , lahav & maddox 1996 ; glazebrook , offer & deeley 1998 ; bromley et al . 1998 ; f99 ) to deal with large multi - dimensional data sets .
a detailed mathematical formulation of the pca adopted here in is given in f99 .
note that the most significant difference between this formulation and that used by other authors ( e.g. connolly 1995 ) is that our spectra have been mean - subtracted ( fig . [ aver ] ) before constructing the covariance matrix .
this makes no substantial difference to the analysis since using other methods simply yields the mean spectrum as the first component .
we note that throughout the remainder of this paper we will denote the eigenvectors ( herein eigenspectra ) as @xmath25,@xmath26 etc . and the projections onto these axes by @xmath27 , @xmath28 etc .
pca is a useful technique in that it allows us to easily visualise a multi - dimensional population in terms of just a handful of significant components .
it does this by identifying the components of the data ( in this case the galaxy spectra ) which are the most discriminatory between each galaxy .
the significance of each component is measured in terms of its contribution to the variance over the sample and is determined in the pca .
this allows us to identify just the most significant components for future use .
it is clear from such a formalism that any clustering in a space defined by the pca is indicative of distinct sub - populations within the sample in terms of reduced dimensionality we find , after applying the pca to our galaxy spectra , that rather than using the original 738 spectral channels to describe each spectrum we can use only two projections and still retain two thirds of the total variance within the population .
indeed we can retain 50% of the total variance just from the first component ( the total variance of the population includes noise and hence this is an underestimate of the true fraction of variance in these projections ) .
the significance of each successive principal component drops off very sharply so these first few components are by far the most important ( table [ tabpc ] ) .
.the relative importance ( measured as variance ) of the first 8 principal components . [ cols="^,^,^,^",options="header " , ] [ tabres ] f99 has performed similar calculations of the lf per spectral type using a subset of the data presented here . in fig .
[ cpsf ] we compare their results with those presented here .
note that in order to make this comparison we have recalculated the lfs assuming ( for consistency ) an einstein de - sitter universe ( @xmath29 , @xmath30 ) , and we have reduced the 5 classes of f99 to our 4 types by taking linear combinations of their lfs which give the same fractions of galaxies per type as our own .
it can be seen from the figure that our results are in good agreement with those of f99 except for some small systematic shifts .
these differences are due to four factors which we believe have significantly improved the accuracy of our analysis . 1 .
we have used a considerably larger sample of galaxies in our calculations .
we have adopted self consistent @xmath8-corrections , resulting in a small change in the predicted @xmath9 and @xmath31 .
3 . taking into account recent advances in our ccd calibrations we have been able to update the catalogue @xmath0 magnitudes ( colless et al .
2001 ; norberg et al . , 2001 ) .
this has also resulted in a small change in the number counts and hence @xmath31 .
we have corrected for saturation effects in high surface brightness objects ( section 4.3 ) .
this has a particularly large impact on our most late - type ( type 4 ) galaxies .
the most similar independent analysis of the lf per spectral type has been conducted by bromley et al .
( 1998 ) , using the las campanas redshift survey .
we also compare these results with our own in fig .
[ cpsf ] . again , in order to compare the 6 clans with our 4 types we have had to combine their lf determinations in a way which yields the same fractions of galaxies per type as our own .
in addition we have applied the transformation @xmath32 ( lin et al .
1996 ) to convert the lcrs gunn-@xmath33 magnitudes to @xmath0 .
the trend of the galaxy lfs to become progressively fainter as one moves from passive to active star - forming galaxies is clear for both sets of results .
however there appears to be some significant disagreement particularly in the faint end slope between the two lf determinations .
given the poor fit of the schechter function to the faint end of the lf we would not expect these two sets of results to necessarily agree within their stated errors .
the large differences observed are most probably due to the different selection effects ( e.g. surface brightness and @xmath0 versus @xmath33 selection ) which can be particularly prominent for fainter galaxies , hence adding more uncertainty to @xmath2 than can be estimated from a standard error analysis .
our total luminosity density ( again assuming an einstein de - sitter universe ) is within 4% of the figure obtained in f99 and is 30% lower than the figure quoted by blanton et al .
( 2001 ) for the sloan digital sky survey early data release .
the reasons for this discrepancy are discussed in norberg et al .
( 2001 ) , but in summary we believe that the blanton et al .
figure is too high , partly owing to large - scale structure in their small initial sample , and also through the use of an inappropriate colour equation to predict 2dfgrs magnitudes .
the spectra observed as part of the 2dfgrs are acquired through optical fibres with a diameter of 140@xmath13 m , corresponding to between @xmath34 and @xmath35 on the sky ( depending on plate - position , see lewis et al . , 2001 ) .
in our assumed cosmology ( @xmath36=0.3 , @xmath37=0.7 ) this translates to an aperture width of approximately @xmath38 kpc at the survey median redshift of @xmath39 .
the fact that fitted galaxy brightness profiles have scale lengths of the order of 3 - 4 kpc ( binney & merrifield 1998 ) raises the question of how representative the observed spectra are of the galaxies as a whole .
however , with typical seeing ( @xmath40 ) and random fibre positioning errors ( rms @xmath41 ) the 2df fibre - aperture will in fact sample a much larger portion of each galaxy .
if the effective fibre apertures were significantly smaller than the galaxies over a range of redshifts then the variation in this aperture with redshift would result in systematic biases in the determination of the lfs .
these so - called ` aperture effects ' are difficult to quantify without an in depth knowledge of the spatially resolved spectral properties of the targeted galaxies .
kochanek , pahre & falco ( 2000 ) recently suggested a test to estimate the influence of this fibre - aperture bias on the determination of type dependent lfs @xmath42 .
the test involves comparing the observed fractional number counts of galaxies in each class to the fractional number counts predicted by our derived lfs .
kochanek , pahre & falco found that for redshift surveys classified morphologically the two were generally consistent but that for surveys classified spectrally using fibre - aperture spectra the two were very inconsistent .
we have performed this test on our data and found no substantial inconsistencies ( fig .
[ aper ] ) . with the exception of small systematic offsets
the observed type fractions were found to agree with those predicted by the lf to within 2@xmath43 over the entire range in @xmath0 magnitude .
we note that as the number counts predicted by the lf are normalised by construction to the observed number counts this test is only necessary and not sufficient . for this reason
more sophisticated tests such as comparisons with detailed models are necessary in order to isolate and analyse the competing effects on the data ( e.g. positional errors , seeing and evolution ) .
we show the observed and predicted @xmath44 distributions for each spectral type out to @xmath45 in fig .
[ nofzfig ] .
it can be seen that the observed @xmath44 is well recovered by the lf for the first two types .
there appears to be a significant over abundance of high-@xmath46 galaxies for the two highest star - forming types as well as a slight under abundance of low-@xmath46 galaxies .
similar deviations in abundance could be explained by aperture effects . however , if this were the only explanation then we would also observe a significant over abundance of low-@xmath46 passive star - forming galaxies , which is not the case .
a much more reasonable explanation for these deviations between the predicted and observed @xmath44 distributions is the presence of evolution ( see e.g. lin et al .
1999 ) . because these two effects affect the observed galaxy population in similar ways they will be difficult to dis - entangle in more detailed analyses .
for the present we note that as far as the determination of the lf in broad bins of spectral type is concerned both of these biases have only a small effect on our calculations .
in this paper we have applied a method of spectral classification , @xmath1 , based on the the relative emission / absorption line strength present in a galaxy s optical spectrum .
this classification correlates well with morphology and we interpret it as a measure of the relative current star - formation present in each galaxy .
based on this classification we have divided the galaxies observed to date in the 2dfgrs into 4 spectral types and calculated the lf for each .
our results are consistent with previous determinations in that the faint end slope systematically steepens and the characteristic magnitude becomes fainter as we move from passive to active star - forming galaxies . in addition
we have been able to show that the schechter function is a poor fit to the lf over a large range of @xmath47 magnitudes , perhaps as a result of sub - populations within the data .
the calculations presented in this letter represent the most accurate determinations of the optical @xmath0 galaxy lf to date . as @xmath1 is a continuous variable the next step in quantifying
the variation of the lf with spectral type is to develop a bivariate formalism @xmath48 .
such a result would be extremely valuable in that it would quantify the variation in the schechter parameters with galaxy type , @xmath49 and @xmath50 , allowing the results to be applied much more objectively to such things as semi - analytic models from which spectra as well as other galaxy properties can be derived ( e.g. slonim et al .
2001 and references therein ) .
we can then also extend this approach of comparing the observed lfs with model lfs by mimicking the observational selection effects ( e.g. aperture bias ) .
one of the key advantages of our spectral type , @xmath1 , is that it can easily be applied to any other set of galaxy spectra which occupy the relevant wavelength range .
this will be invaluable when comparing results from the many different galaxy surveys currently underway ( e.g. sdss , virmos , 6df etc . ) .
the 2df galaxy redshift survey has made available to the community 100,000 galaxy spectra and redshifts through their website , .
additional information regarding the spectral classification and lfs presented here can be obtained from http://www.ast.cam.ac.uk/@xmath51twodfgrs/.
dsm was supported by an isaac newton studentship from the university of cambridge and trinity college .
we thank r. kaldare , s. ronen and r. somerville for many helpful discussions .
the 2df galaxy redshift survey was made possible through the dedicated efforts of the staff at the anglo australian observatory , both in creating the two - degree field instrument and supporting it on the telescope .
baldry i.k . ,
et al . ( the 2dfgrs team ) 2001 , mnras , submitted ( astro - ph/0110676 ) binney j. , merrifield m. , 1998 , ` galactic astronomy ' , princeton university press blanton m. , et al . ( the sdss collaboration ) , 2001 , aj , 121 , 2358 bromley b.c . ,
press w.h .
, lin h. , kirshner r.p . , 1998 , apj , 505 , 25 colless , m. m. , et al . ( the 2dfgrs team ) , 2001 , mnras , 328 , 1039 connolly a.j . , szalay a.s .
, bershady m.a . ,
kinney a.l .
, calzetti d. , 1995 , aj , 110 , 1071 cross n.j.g , 2001 , phd thesis , st andrews university cross n.j.g .
, et al . ( the 2dfgrs team ) , 2001 , mnras , 324 , 825 efstathiou g. , ellis r.s . , peterson b.a . , 1988 , mnras , 232 , 431 ellis r.s . , colless m. , broadhurst t. , heyl j. , glazebrook k. , 1996 , mnras , 280 , 235 folkes s.r . , lahav o. , maddox s.j . , 1996 , mnras , 283 , 651 folkes s. , et al . (
the 2dfgrs team ) , 1999 , mnras , 308 , 459 glazebrook k. , offer a.r . , deeley k. , 1998 , apj , 492 , 98 heavens a.f . , jimenez r. , lahav o. , 2000 , mnras , 317 , 965 kennicutt r.c . , jr . , 1992 ,
apjs , 79 , 255 kochanek c.s . ,
pahre m.a . ,
falco e.e .
, 2000 , apj submitted ( astro - ph/0011458 ) lahav o. , 2000 , to appear in eso astrophysics symposia , springer - verlag ( astro - ph/0012407 ) lewis i. , et al . ,
2001 , mnras , submitted lin h. , kirshner r.p .
, schectman s.a . , landy s.d .
, oemler a. , tucker d. , schechter p.l . , 1996 ,
apj , 464 , 60 lin h. , yee h.k.c . , carlberg r.g .
, morris s.l . , sawicki m. , patton d.r . , wirth g. , shepherd c.w .
, 1999 , apj , 518 , 533 loveday j. , tresse l. , maddox s. , 1999 , mnras , 310 , 281 maddox s.j . , efstathiou g. , sutherland w.j . , 1990 , mnras , 246 , 433 maddox s.j . , 1998 , in dodorico s. , fontana a. , gialongo e. , eds , asp .
ser . , the young universe : galaxy formation and evolution at intermediate and high redshift .
, san fransisco , p. 198
marzke r.o . ,
da costa l.n . , 1997 , aj , 113 , 185 norberg p. , et al .
2001 , mnras , submitted ( astro - ph/0111011 ) pence w. , 1976 , apj , 203 , 39 ronen s. , aragn - salamanca a. , lahav o. , 1999 , mnras , 303 , 197 sandage a. , tammann g.a .
, yahil a. , 1979 , apj , 232 , 352 slonim n. , somerville r. , tishby n. , lahav o. , 2001 , mnras , 323 , 270 wilmer c.n.a , 1997 , aj , 114 , 898 | we calculate the optical @xmath0 luminosity function of the 2df galaxy redshift survey ( 2dfgrs ) for different subsets defined by their spectral properties . these spectrally selected subsets
are defined using a new parameter , @xmath1 , which is a linear combination of the first two projections derived from a principal component analysis .
this parameter @xmath1 identifies the average emission and absorption line strength in the galaxy rest - frame spectrum and hence is a useful indicator of the present star formation .
we use a total of 75,000 galaxies in our calculations , chosen from a sample of high signal - to - noise ratio , low redshift galaxies observed before january 2001 .
we find that there is a systematic steepening of the faint end slope ( @xmath2 ) as one moves from passive ( @xmath3 ) to active ( @xmath4 ) star - forming galaxies , and that there is also a corresponding faintening of the rest - frame characteristic magnitude @xmath5 ( from @xmath619.6 to @xmath619.2 ) .
we also show that the schechter function provides a poor fit to the quiescent ( type 1 ) lf for very faint galaxies ( @xmath7 fainter than @xmath616.0 ) , perhaps suggesting the presence of a significant dwarf population .
the luminosity functions presented here give a precise confirmation of the trends seen previously in a much smaller preliminary 2dfgrs sample , and in other surveys .
we also present a new procedure for determining self - consistent @xmath8-corrections and investigate possible fibre - aperture biases .
[ firstpage ] galaxies : distances and redshifts galaxies : elliptical and lenticular , cd galaxies : stellar content galaxies : formation galaxies : evolution |
a great deal of effort has recently gone into the study of higher - dimensional models as unified theories of gravity and fundamental matter fields .
the original interest in such models in kaluza
klein ( kk ) theory @xcite was revived by the important role of dimension 11 in supergravity @xcite and more recently by superstring theory @xcite and m - theory @xcite , which favour spacetimes of dimension 10 and 11 respectively .
the usual ( kk ) argument for the apparent 4-dimensionality of spacetime is that the extra dimensions are compactified i.e curled up sufficiently small so as not to conflict with observation .
however , recently much attention has been placed instead on string - theory - inspired models with large extra dimensions @xcite in which most of the physics is confined or closely - bound to a lower - dimensional braneworld surrounded by a higher - dimensional bulk . though motivated by superstring and m - theories , many of these models @xcite are in fact formulated within the framework of ( higher - dimensional ) general relativity ( gr ) . among these are : 1 ) the second randall
sundrum scenario @xcite in which the graviton is tightly - bound to the brane by the curvature due to the warping of the bulk metric , which is pure anti - desitter ( ads ) .
2 ) a more general scheme of shiromizu , maeda and sasaki ( sms ) @xcite , in which the 4-dimensional einstein field equations ( efe s ) are replaced by 4-d `` braneworld efe s '' ( befe s ) , which are not closed since there is a ` dark energy ' weyl tensor term , knowledge of which requires solving also for the bulk .
the interpretational difficulties due to this constitute the ` weyl problem ' .
the other main distinctive feature of sms s befe s is the presence of a term quadratic in the braneworld energy - momentum , which arises from the junction condition used @xcite .
two important questions arise in consideration of such models .
* q1 : * how should such models be built and interpreted consistently within the framework of gr ?
this would require a careful underlying choice of conceptually - clear general framework , in the sense we discuss below .
* q2 : * what exactly is the connection between such models and the underlying string / m theory ?
more precisely , to what extent can the agreements or otherwise of predictions of such models with observations be taken as support or disagreement with such theories ? here
we concentrate on the first question and make a comparative study of the general schemes that have been employed in the literature in order to construct the bulks which surround branes .
such a comparative study requires a sufficiently general common language .
we use that of _ p.d.e problems , which consist of both the p.d.e system to be solved in some region of a manifold , and data prescribed on ( portions of ) the boundary of this region . for ( @xmath1 , @xmath2 ) a spacetime of dimension @xmath3 with @xmath2 time dimensions , we denote the problem involving the addition of an extra dimension by ( @xmath1 , @xmath2 ; @xmath4 ) , where @xmath5 if the new dimension is spacelike or @xmath6 if it is timelike .
this is the generalization of the gr cauchy problem ( cp ) @xcite based on the arbitrary ( @xmath1 , @xmath2 ; @xmath4 ) generalization of the arnowitt
misner @xcite split of the metric ( sec 2.1 ) and hypersurface geometry ( sec 2.2 ) .
its simple signature - independent features are pointed out in sec 2.3 and the crucial dependence of many of the harder features on the usual cp signatures @xmath7 , @xmath8 is discussed in sec 2.4 . as is well - known , the gr cp presupposes the existence of the data .
thus one is in fact considering a two - step process , the other step being the construction of the data on the ( @xmath1 , @xmath2 ) manifold i.e the generalization of the gr initial value problem ( ivp ) @xcite .
this is discussed in sec 3 . _
our framework permits a profitable look at a number of recent topics .
the aim is to consider gr - based models containing thin matter sheets such as branes or domain walls .
we shall compare two broad schemes that have been proposed to construct bulks : the ( 3 , 1 ; 1 ) construction @xcite starting from information on a ( 3 , 1 ) spacetime hypersurface ( usually the brane ) and the ( 4 , 0 ; 1 ) construction @xcite starting by the construction of data on a ( 4 , 0 ) spatial hypersurface .
however , first we emphasize that one should grasp the fundamental arguments and results which before the specializing to the thin matter sheet models . a few ideas about the general unspecialized framework recently arose in the literature on ( generalizations of ) the campbell magaard arbitrary - signature embedding theorem @xcite ( see sec 2.3 ) .
however , we find it far more profitable instead to adopt the generalized gr cp
ivp point of view since this literature is by far more developed and thus a far greater source of well - documented pitfalls and carefully thought - out techniques which avoid them .
we identify the embedding step of the campbell magaard theorem with the well - known signature - independent parts of the gr cp in sec 2.3 .
but the harder @xmath7 , @xmath8 specific parts of the gr cp strongly suggest that ( 4 , 0 ; 1 ) schemes should be favoured over ( 3 , 1 ; 1 ) ones on very general grounds : well - posedness and causality ( sec 2.4 ) .
furthermore , magaard s data construction method @xcite ( sec 3.1 ) does not compare favourably with york s data construction @xcite ( sec 3.2 ) , and its application to @xmath9 , @xmath5 has further conceptual difficulties . in this light
we look at the extent to which york s method is adaptable to @xmath9 , @xmath5 , and also consider the thin sandwich method @xcite in this context ( sec 3.3 ) .
we then introduce thin matter sheets in sec 4 , and study the ( 3 , 1 ; 1 ) schemes with thin matter sheets , recollecting the derivation of the junction conditions ( sec 4.1 ) , and showing how the sms formulation ( sec 4.2 ) may be reformulated in a large number of ways using geometrical identities that interchange which terms are present in the braneworld efe s .
this is illustrated by the formulation in sec 4.3 which directly parallels the gr cp formulation and thus makes no explicit use of the weyl term , and by formulations in sec 4.4 in which the quadratic term has been re - expressed entirely in terms of derivatives off the brane .
these formulations are used to clarify a number of aspects of the ` weyl problem ' , in particular to argue for the study of the full brane - bulk system .
further aspects of such formulations are discussed in sec 4.5 .
the implications of sec 2.4 for thin matter sheets , the heuristic use and limitations of the ( 3 , 1 ) york method with thin matter sheets , and formulations of thin matter sheet thin sandwich conjectures are discussed in sec 4.6 .
having gathered arguments against using ( 3 , 1 ; 1 ) schemes in secs 2.4 and 3 , we further consolidate this viewpoint by arguing in sec 5 against a suggested virtue of ( 3 , 1 ; 1 ) schemes @xcite : their use to remove singularities .
in addition to using the framework of sec 2 , the corresponding study of geodesics is also used here , which presents further conceptual complications relevant to thin matter sheet models .
these follow from the ( 3 , 1 ) geodesics not usually being among the ( 4 , 1 ) geodesics . in sec 6 , we continue the study of the ( @xmath10 , 0 ; 1 ) scheme favoured by our arguments , in the presence of both thin and thick ( i.e finitely thin ) matter sheets .
we begin in sec 6.1 by providing a hierarchy of very difficult general thin matter sheet problems in which the main difficulties stem from low differentiability and details about the asymptotics . within this class of problems
we identify how the far more specific scenarios currently studied emerge as more tractable cases , and thus identify which as - yet unjustified assumptions such studies entail .
the ivp step , of use in the study of how braneworld black holes extend into the bulk ( the ` pancake ' versus ` cigar ' debate @xcite ) , involves fewer of such assumptions .
thus in this paper we restrict attention to the ( @xmath10 , 0 ) data construction problem ( sec 6.2 ) , which we apply to thin matter sheets in sec 6.3 and more straightforwardly to thick matter sheets in sec 6.4 .. sec 7 discusses * q2 and contains the conclusions of this paper as regards * q1 . * *
in this section , we give splittings of efe s with respect to a hypersurface of codimension 1 .
this results in 2 systems of equations that are treated as constituting two separate steps .
the first , embedding or evolution step , will be dealt with in this section and the second , data construction step , will be presented in sec 3 .
the embedding step includes the gr cauchy problem ( cp ) for @xmath7 , @xmath8 .
we work with general ( @xmath1 , @xmath2 ; @xmath4 ) since many applications of interest have @xmath9 and @xmath5 .
our approach is to treat the general case as far as possible , while taking careful note of the many important roles played by the signatures @xmath4 and especially @xmath2 ( whereas a change in the dimension @xmath10 plays only a mild role ) .
an important point regarding the well - known analytic p.d.e system approach employed here is the identification of the input to be prescribed and the output to expect . with the advent of schemes to construct bulks @xcite , the old questions regarding these prescriptions have become relevant in a new context .
we begin by recalling the arbitrary ( @xmath1 , @xmath2 ; @xmath4 ) arnowitt - deser misner ( adm ) @xcite type split w.r.t the ( @xmath1 , @xmath2 ) ` initial ' hypersurface @xmath11 of the higher - dimensional metric @xmath12 along with the split of its inverse @xmath13 , ll g_cd = ( ll _ i^i + ^2 & _ d + _ c & h_cd ) , g^cd = ( ll & - + - & h^cd + + ) .
[ adms ] the geometric interpretation of ( and further notation for ) the ( @xmath10 , 0 , 1 ) and ( @xmath1 , 1 , 1 ) splits discussed in this paper are exhibited in fig .
1 . for the specific adm case , the added dimension ( of signature @xmath8 ) is time , and the lower - dimensional hypersurface is 3-dimensional and entirely spatial : ( @xmath1 , @xmath2 ) = ( 3 , 0 ) .
the efe s are then decomposed into 4 constraints and 6 evolution equations ( sec 2.3 ) .
the gr ivp and cp are defined w.r.t this split .
the gr ivp is the solution of the 4 constraints [ on some region @xmath14 of some ( 3 , 0 ) initial hypersurface @xmath15 for initial data @xmath16 , where @xmath17 is the extrinsic curvature k_ab = -_h_ab , [ ecd ] which is a measure of how much the hypersurface is bent in the surrounding spacetime .
this data , together with the 6 evolution equations ( for @xmath18 ) and the 6 equations for @xmath19 which follow from ( [ ecd ] ) constitute the gr cp .
this is a ( 3 , 0 ; 1 ) problem for the construction of the spacetime region in which @xmath14 is embedded .
it is useful to juxtapose here another metric split , the arbitrary ( @xmath1 , @xmath2 ; @xmath4 ) kk type split ll g_cd = ( ll ^2 & ^2a_d + ^2a_c & h_cd + ^2a_ca_d ) , g^cd = ( ll a_ma^m + ^2 & - a^d + - a^c & h^cd ) , [ kks ] so as to discuss some of the literature @xcite .
note first that for kk theory itself , this split is ( 3 , 1 ; 1 ) , with @xmath20 , @xmath21 , @xmath22 held to be independent of the added dimension s coordinate @xmath23 .
this so - called cylindricity condition implicitly permits @xmath21 to be interpreted as the classical electromagnetic potential .
more recent generalizations of this scheme , called ` noncompact kk ( nkk ) theory ' @xcite , have involved large extra dimensions in place of cylindricity , by permitting @xmath20 , @xmath21 , @xmath22 to depend on @xmath23 . in some simple instances , @xmath21
is held to be 0 .
but clearly for @xmath24 and @xmath25 , the two splits ( [ adms ] ) and ( [ kks ] ) are identical .
so there is already a vast literature on embeddings with a large extra dimension since this is what is usually treated in the gr cp . in sec 4 , we reinterpret ( 3 , 1 ; 1 ) methods such as that of sms as _ z - dynamics i.e dynamics " with the coordinate of the new dimension , @xmath23 , as dynamical variable " .
note that such a dynamical " interpretation was of no avail in kk theory proper , for there @xmath20 was taken to be independent of @xmath23 .
note also that although a way of relating sms s befe s and nkk has been pointed out @xcite , we emphasize that it is far more useful as here to relate the steps leading to sms s befe s to the gr cp and ivp literature , since this is by far more developed and thus a good source of ideas and theorems , on which the critical sec 2.4 is based . _
gauss _ outstanding theorem _
( @xmath26 for a 2-sphere embedded in flat 3-space ) already provides a good intuitive feel both of why matter squared terms arise in brane cosmology and of why intrinsic curvature singularities can in principle be removed upon applying an embedding by being balanced out by extrinsic - curvature - squared terms .
we work with the suitable generalization of this theorem : the ( @xmath10 + 1)-dimensional geometry of projections w.r.t a ( @xmath1 , @xmath2 ) hypersurface .
we use @xmath27 for projections onto the ( @xmath1 , @xmath2 ) hypersurface @xmath28 , and @xmath29 for projections onto the normal .
the three projections of the riemann tensor @xcite are respectively the gauss , the codazzi and the ricci equations : _ abcd = r_abcd - 2k_a[ck_d]b , [ gaussful ] _ abc = -2d_[ck_b]a , [ codful ] _ ab = ( _ k_ab - d_bd_a ) + k_a^ck_cb .
[ thirdproj ] the ricci equation ( [ thirdproj ] ) is particularly important in sec 4.3 - 4 .
contracting these equations gives _ bd - _ bd = r_bd - ( kk_bd - k_b^ck_cd ) .
[ contg ] _ a = r_a = -(d_bk^b_a - d_ak ) , [ gpa ] _ = - kk .
[ tpcont ] the gauss equation may be contracted a second time to give ( using @xmath30 ) 2 g _
- = - r + k^2 - kk [ gpp ] .
the ricci equation can be used in the contracted gauss equation to obtain ( _ k_ab - d_bd_a ) - kk_ab + 2k_a^ck_bc + r_ab = _ ab . [ protoev ] the trace of this [ a linear combination of ( [ gpp ] ) and ( [ tpcont ] ) ] is : ( _ k - d^2 )
- k^2 + r = - r _ [ prototrev ] . another linear combination of ( [ gpp ] ) and ( [ tpcont ] ) is - = . from this and ( [ protoev ] )
it follows that + ( 2k_a^ck_bc - kk_ab + h_ab ) + g_ab = g_ab [ gid ] it is well - known that the general ( @xmath1 , @xmath2 ; @xmath4 ) adm - type split of the efe s @xmath31 gives rise via ( [ gpp ] ) and ( [ gpa ] ) to the ( @xmath10 + 1 ) gauss and codazzi constraints , 0 ; 1 ) these are known as the hamiltonian and momentum constraints respectively . ]
k^2 - kk - r = 2 g _
= 2 t _ 2 , [ gauss ] -(d_bk^b_a - d_ak ) = g_a = t_a j_a , [ cod ] together with @xmath32 evolution equations " w.r.t the dynamical variable " @xmath33 ( _ k_ab - d_bd_a ) - kk_ab + 2k_a^ck_bc + r_ab = _ ab = ( s_ab - h_ab ) - h_ab , [ evk ] obtained via ( [ protoev ] ) . here
@xmath34 , @xmath35 and @xmath36 ( n.b @xmath37 ) are general matter terms which are usually prescribed as functions of matter fields that are governed by usually - separate field equations .
a useful equation is the trace of ( [ evk ] ) _ k = ( k^2 - r - ) + d^2= + d^2 , [ trevk ] the second equality of which follows from the gauss constraint ( [ gauss ] ) . also applying to @xmath17 the decomposition of symmetric second - rank tensors @xmath38 into their trace @xmath39 and
tracefree part , @xmath40 , the equations ( [ gauss ] ) and ( [ cod ] ) and ( [ trevk ] ) take on the useful forms k^k^ - k^2 + r + 2= 0 , [ agauss ] d_b k^b_a - d_ak + j_a = 0 , [ acod ] ( _ k - d^2 ) - = k^k^ + .
[ protoray ] we now consider the simpler results of the ( 3 , 0 ; 1 ) gr cp , most of which hold irrespective of both @xmath4 and @xmath2 .
first , there is the two - step nature of the problem for the whole efe system .
* evolution " or embedding step : given as data @xmath16 together with coordinate functions @xmath34 and @xmath35 , all on some portion @xmath41 of @xmath11 , such that the constraints are satisfied , one is to solve the evolution equations ( [ ecd ] ) and ( [ evk ] ) as a prescription for evaluating @xmath16 on some portion @xmath42 of a nearby @xmath43 i.e for a small increment of the dynamical variable @xmath33 .
* * initial value " or data construction step : to construct such data by solving the constraints ( see sec 3 ) .
* second , provided that the constraints hold on some `` initial '' @xmath41 , they are guaranteed to hold throughout the embedding spacetime region .
this is most quickly seen by virtue of energy - momentum conservation and the bianchi identities @xcite , making use of the einstein tensor projection form of the constraints .
third , constraints come hand - in - hand with gauge freedoms .
thus @xmath44 of the metric components are freely - specifiable . for many applications a choice of gauge
is required . to begin with we mostly use the _ normal coordinate gauge ( @xmath45 , @xmath46 ) , and
the _ harmonic gauge @xcite ( coordinates @xmath47 such that @xmath48 ) , though we use more elaborate gauge choices later .
we stress that careful consideration of gauge choice can be an important ingredient toward producing prescriptions that are generally robust and physically clear @xcite . _ _
fourth , by choice of a suitable gauge it can be ascertained by the cauchy
kovalevskaya ( ck ) theorem @xcite for admissible analytic data that there locally exists a unique solution to the evolution equations " ( [ evk ] ) . for this and further applications in the gr cp , the harmonic gauge is usually chosen . in the ( @xmath10 , 0 ; 1 )
gr cp , the use instead of the normal gauge is discouraged because in practice it typically breaks down quickly . to see this recall that in normal gauge equation ( [ protoray ] ) may be written as the normal raychaudhuri equation - = k^ k^ + r _ 0 , [ normray ] where the inequality follows from the definition of the strong energy condition ( sec ) : _ abu^au^b 0 u^a , [ protosec ] by picking @xmath49 . to make ( [ protosec ] ) manifestly an energy condition , one uses the efe s to obtain
_ abu^au^b - u^a .
[ sec ] integrating the differential inequality in ( [ normray ] ) shows that if @xmath50 , @xmath51 within a finite range of the parameter @xmath52 along @xmath53 .
this blowup is by definition a _ caustic and signifies a breakdown of the normal coordinates .
however , for the ( @xmath1 , 1 ; 1 ) embedding @xmath54 is spacelike .
there is then no good reason for @xmath55 to be always positive , so caustics may be less likely to form .
the issue of the longevity of normal coordinates is of interest because they have hitherto been used for some of the applications below .
other applications below fix @xmath46 alone .
_ an embedding result we discuss below is the campbell magaard ( cm ) theorem , according to which any @xmath10-space can be surrounded by @xmath10 + 1 dimensional vacuum space , where the spaces in question are analytic .
this is closely related to the above split of the efe s ; in particular it has an embedding step which we consider in this section and a data construction step , ` magaard s method ' , which we consider in sec 3.1 .
the embedding step of the cm theorem is simply the result above based on the ck theorem , which is well - known for the gr cp , and follows for the other ( @xmath1 , @xmath2 ; @xmath4 ) embedding cases from the well - known insensitivity of the ck theorem to ( @xmath1 , @xmath2 ; @xmath4 ) .
strictly , the cm result is the vacuum case ( @xmath56 ) but the proof may easily be extended to any functional form of @xmath57 @xcite since the use throughout of the ck theorem is totally insensitive to changes in such subleading order terms ( contributions to @xmath58 ) .
there have been suggestions that such @xcite or similar @xcite ( 3 , 1 ; 1 ) methods could be used for the construction of bulks .
however , even though the above results in the ( @xmath10 , 0 ; 1 ) case of the gr cp are protected and extended by a host of further theorems , they do not carry over to the ( 3 , 1 ; 1 ) case , as we argue in the next section .
there are also problems with ( 3 , 1 ) data construction , as we shall see in sec 3 . in this section
we argue by further comparison with the gr cp that the popular practise of building bulks by @xmath23-dynamics is in general unsatisfactory .
for , in the gr cp a large number of important results @xcite for the gr cp turn out to be entirely dependent on the lower - dimensional signature @xmath7 .
in other words , the choice of methods which properly respect the difference between space and time is absolutely crucial . to have a sensible problem in mathematical physics ,
one requires not just local existence and uniqueness of solutions but rather _ well - posedness .
whereas well - posedness always includes local existence and uniqueness , it also always includes continuous dependence of the data , without which an arbitrarily small change in the data could cause an arbitrarily large immediate change in the evolution @xcite i.e there is no guarantee of physical predictability from such a problem . for a hyperbolic - type system one
further requires a domain of dependence ( dod ) property to enforce a sensible notion of causality .
if the data is known only on a closed achronal set @xmath59 _ ( that is , a piece of a spacelike hypersurface ) , then the evolution can only be predicted within a region @xmath60 ( the future dod ) , defined as the set of points such that all past - inextendible causal curves through each point [ represented by the @xmath61 s leading to the point @xmath62 in fig .
4a ) ] intersect @xmath59 . in our view causality can effectively be studied only in settings where the dynamical variable is time .
one reason for this is that given information on an arbitrarily - thin ( 3 , 1 ) hypersurface , the ( 4 , 1)-dimensional dod is negligible because of this thinness ( see fig .
other reasons are discussed below .
now we note that the cauchy
kovalevskaya theorem has no control whatsoever over these last two properties for the analytic functions .
there are furthermore two reasons why the analytic functions are undesirable .
first , the analytic functions fail to be appropriate for any causal study since they are rigid ( a change in a small region of an analytic data set induces an unwanted change of the entire data set ) . without them
one absolutely can not have anything like the extremally general ( and in particular signature - insensitive ) cauchy kovalevskaya theorem .
second , realistic and ( or ) interesting situations seldom involve analytic or smooth functions .
for example astrophysical bodies have boundaries , thin matter sheets are discontinuous and the approach to spacetime singularities can be of low differentiability .
bruhat showed that the gr cp is well - posed in 1952 ( @xcite ) and 1956 ( @xcite ) , by making use of how the ( @xmath10 , 0 ; 1 ) efe system ( [ gauss][evk ] ) when cast in harmonic coordinates is of the correct quasilinear hyperbolic form @xmath63 ( where @xmath64 is a lorentzian metric ; both this and the function @xmath65 are smooth ) to enable use of leray s theorem @xcite , which guarantees local existence and uniqueness , and furthermore continuous dependence on the initial data and the dod property .
the above proofs of the four well - posedness criteria for the gr cp are now usually done using _
sobolev spaces @xcite .
these are appropriate because of the hyperbolic character of the einstein evolution equations . at first
they lend themselves to less involved proofs than bruhat s , although if one seeks yet stronger results the functional analysis becomes extremally unpleasant . to recollect simply why sobolev spaces are appropriate @xcite
, it suffices to consider a klein gordon scalar in minkowski spacetime .
given data on a bounded region @xmath14 of a spacelike hypersurface @xmath66 , one can draw the future dod @xmath67 [ fig .
4a ) ] which is the region affected solely by this data due to the finite propagation speed of light .
one can then consider the values of @xmath68 and its first derivatives on @xmath69 .
then by gauss s ( divergence ) theorem and energy - momentum conservation , _ s_0t_act^at^c + _
bt_acl^at^c = _ s_tt_act^at^c , and the second term @xmath70 provided that the matter obeys the dominant energy condition ( dec ) -t^a_cu^c u^a and that @xmath71 is timelike .
then the definition of the energy - momentum tensor gives _
s_t(()^2 + ( ) ^2 + m^2 ^ 2 ) _
s_0(()^2 + ( ) ^2 + m^2 ^ 2 ) [ sums ] because each integrand is the sum of squares ( which are necessarily positive ) , this means that control over the data on @xmath14 gives control of the solution on @xmath72 .
the idea of a sobolev norm is a generalization of these last two ` energy ' integrals @xcite . for later use
, the sobolev class @xmath73 has a bounded norm of this type including up to @xmath74th order derivatives . _
whereas the gr cp comes with important well - posedness theorems @xcite , the ( @xmath1 , 1 ; 1 ) problem is instead a complicated instance of a _ sideways cauchy problem , a category of problem for which the possibility of such well - posedness theorems remains notably undeveloped .
there is quite simply no established way to proceed .
there is no known sideways ( @xmath9 ) analogue of leray s theorem .
there is no known notion of what function spaces might be appropriate for sideways cauchy problems .
we can explain however why the above introduction of sobolev spaces does not carry over to sideways cauchy problems .
first there is no sideways notion of dod to make the construction .
second even if we assume the higher - dimensional dec holds , it would not give an inequality because the perpendicular vector @xmath75 in now spacelike [ fig .
third , we obtain a difference of squares rather than the sums in ( [ sums ] ) , so the equivalent of the energy method s use of sobolev norms is simply of no use to control the evolution " given the data . _
several methods have been proposed for the data construction step , including the elimination and conformal methods .
the former methods are intuitive in that the prescribed quantities are all physical , but have a number of undesirable mathematical features .
in contrast , lichnerowicz s conformal method @xcite is counterintuitive , since some of the prescribed quantities are unphysical , but exploits the mathematical properties of the constraint system in order to decouple it .
bruhat s argument @xcite was in favour of the latter , however , some of her criticisms need to be elaborated upon to cover magaard s elimination method @xcite .
sec 3.1 covers this question and provides further criticisms in the @xmath9 case . in sec 3.2 ,
we discuss the desirable features of york s development of the conformal method @xcite , paying careful attention to the signature - dependent ones .
the thin sandwich and hamiltonian methods are briefly discussed in sec 3.3 .
an example of this method of data construction is employed in the last part of the cm result ( `` magaard s method '' @xcite ) .
we discuss and compare this method with the standard method used for the gr ivp .
magaard s method ( 1963 ) treats the lower - dimensional metric @xmath76 as a known .
here we use a more transparent notation than magaard s , in which the gauss constraint ( [ gauss ] ) takes the form g^abcdk_abk_cd + r + 2= 0 where @xmath77 is the ( undensitized ) inverse of dewitt s supermetric @xcite .
this is then split w.r.t some coordinate @xmath78 .
then @xmath79 can be isolated as the solution of the linear equation 2k_11g^11uwk_uw + g^1u1wk_1uk_1w + g^uwxyk_uwk_xy = - ( r + 2 ) ( where @xmath80 , @xmath74 , @xmath81 , @xmath82 @xmath83 ) , provided that it is possible to divide by @xmath84 .
thus @xmath79 is eliminated from the codazzi constraint ( [ cod ] ) , which is then treated as a p.d.e system for unknowns @xmath85 , some @xmath86 component denoted @xmath87}. on the face of it , this system satisfies the criteria for the ck theorem if one treats @xmath88 \{all the components of @xmath89 bar @xmath79 , @xmath90 and @xmath87 } as known functions on @xmath78 and provided that the p.d.es coefficients and the data are analytic .
so a unique solution exists . the ( generalized ) cm result @xcite groups this and the local existence of a unique evolution " to form the statement that a ( @xmath1 , @xmath2 ) spacetime with prescribed analytic metric @xmath76 may be embedded with an extra space or time dimension for any analytic functional form of the energy - momentum tensor .
this statement suggests an incredibly rich collection of embeddings exist .
however , we must point out that this is itself one of many difficulties associated with the cm result and its applicability , that we describe below .
bruhat had already considered the above method in 1956 @xcite in the case of the gr ivp and pointed out a limitation on its validity .
we find that her argument can be made for any signature or dimension : @xmath91 must be a function of at least one unknown , @xmath87 , so we have no control over whether it is zero .
therefore we can not guarantee the validity of the elimination procedure for @xmath79 from the gauss constraint .
although magaard @xcite finds a route round this problem , we find that this leads to two prices to pay later on .
he starts with prescribed ` data for the data ' i.e values of @xmath92 on some @xmath93-dimensional set @xmath94 .
the ` data for the data ' is validly picked so that @xmath95 on @xmath96 , whereupon by continuity there is a thin region @xmath97 within which it is guaranteed that @xmath95 .
thus magaard s procedure produces strips of data to be used in the embedding step .
the first price to pay is that in general we can not expect to be able to patch such strips together to make extended patches of data . for ,
since the strip construction ends where @xmath91 picks up a zero for some @xmath98 , while restarting the procedure with @xmath98 in place of @xmath99 is valid , the two data strips thus produced will have a discontinuity across @xmath98 .
so what one produces is a collection of strips belonging to different possible global data sets . the evolution of each of these strips would produce pieces of different higher - dimensional manifolds .
so one is _ not in fact saying that an empty ( @xmath10 + 1)-dimensional manifold surrounds any @xmath10-dimensional manifold .
rather one is saying that any @xmath10-dimensional manifold can be cut up in an infinite number of ways ( choices of the @xmath78 coordinate ) into many pieces , each of which can be separately bent in an infinite number of ways ( corresponding to the freedom in choosing the components of @xmath17 in @xmath100 on each set of ` data for the data ' ) , and for each of these bent regions we can locally find a surrounding ( @xmath10 + 1)-dimensional manifold for every possible analytic function form of @xmath57 ( corresponding to the generalizability of the cm result ) .
this apparent excess richness raises the question of physical significance of such a construction . _
magaard s method does not state enough assumptions to make it rigorous .
first , how far does ` data for the data ' extend along @xmath99 ? clearly the topology of the @xmath10-dimensional manifold is an important input , for if it is not compact without boundary , there is a missing boundary or asymptotics prescription required .
also the topology of the @xmath99 set itself has not been brought into consideration ( for example continuity is not guaranteed were this to contain loops ) .
secondly , we specifically consider the ( @xmath10 , 0 ; 1 ) and ( @xmath1 , 1 ; 1 ) problems as separate cases , since we find that there are implicit ways in which signature plays a crucial role even for the cm result . for ( @xmath10 , 0 ; 1 ) one might worry that the data construction for a strip is flawed because the data problem in question is elliptic and
hence naively requires a global treatment .
however , we are able to dispel this worry once we treat york s method below . for ( @xmath1 , 1 ; 1 ) , the use of the strip @xmath97 as evolutionary " data is generally invalidated by the information leak construction in fig . 5 , unless one has had the luck to construct a full global data set .
this would , however , require the building of the data encountering 1 ) no zeros of @xmath91 ( which is thus the second price to pay ) , 2 ) no asymptotic problems .
thus in general there is a severe problem with building ( 4 , 1 ) spacetimes from ( 3 , 1 ) ones using the cm method .
thirdly , magaard s method lacks any @xmath10-dimensional general covariance since it involves the choice of a coordinate @xmath78 and the elimination of a @xmath101-component of a tensor .
the arguments of lichnerowicz @xcite and bruhat @xcite led to the gr ivp taking a very different route from the above sort of brute - force elimination methods .
the conformal method of lichnerowicz and york is instead preferred .
this leads us to ask to what extent conformal methods can be adapted both to heuristic and to general constructions for e.g ( @xmath1 , 1 ; @xmath4 ) data .
therefore our treatment below is as far as possible for the general ( @xmath1 , @xmath2 ; @xmath4 ) case . in the conformal method , one chooses to treat @xmath102 as a known metric which is moreover not the physical metric but rather only conformally - related to it by _
ij = ^h_ij .
we work in terms of @xmath103 and @xmath104 , and permit @xmath105 , @xmath106 and @xmath34 to conformally transform according to ^ij = ^- 2k^ij , ^i = ^ , = ^ , whilst crucially demanding the constant mean curvature ( cmc ) condition k = holds and is conformally - invariant .
one then demands that the ( raised ) codazzi constraint ( [ acod ] ) is to be conformally - invariant . since _
= ^- 2 one requires that - = - 2= .
[ twiddle ] furthermore one demands that the conformally - transformed gauss equation r - ( n - 1 ) + ( n - 1 ) ( 1 - ) + ( m^2- - ^2^ + 2^+ ) = 0 [ prolich ] ( where @xmath107 and @xmath108 is some york extrinsic dynamical variable " proportional to @xmath104 ) is simplified by being made to contain no @xmath109 term , so that @xmath110 , @xmath111 and @xmath112 [ by making use of ( [ twiddle ] ) ] .
now , regardless of ( @xmath1 , @xmath2 ; @xmath4 ) , provided that the ( @xmath10 + 1)-dimensional dec is to be preserved by the conformal transformation , @xmath113 must conformally - transform like @xmath114 , implying @xmath115 . then ( [ prolich ] ) becomes the ( @xmath1 , @xmath2 ; @xmath4 ) version of the lichnerowicz equation d^2= - ( -r - m^-4 + ^2^ - 2 ^ -2 ) [ ndlich ] for the conformal factor @xmath116 .
we observe that a number of the very attractive features of the york method are ( @xmath1 , @xmath2 ; @xmath4)-independent . first , it decouples the solution of the codazzi constraint from the solution of the gauss constraint .
the former proceeds by a traceless - transverse ( tt)traceless - longitudinal ( tl ) splitting @xcite k^ij = k^ij + k^ij , d_ik^ij 0 , k^ij = 2(d^(iw^j ) - h^ijd_cw^c ) ( for some vector potential @xmath117 ) , which along with the trace - tracefree split is conformally - invariant and thus unaffected by the solution of the latter , which has become the p.d.e ( [ ndlich ] ) for the conformal factor .
the simplest case is @xmath118 , @xmath119 which at most requires solution of a first - order linear equation .
second , the choice in using the scale - scalefree decomposition of the metric and the trace - tracefree and tt tl decompositions of @xmath89 are all decompositions into irreducibles and thus @xmath10-dimensionally generally - covariant choices , a decided advantage over the magaard method .
third , one can attempt to preserve the cmc condition away from the ` initial ' hypersurface @xmath11 , which can be done provided that the equation d^2 + ( k^2 - r - ) = can be solved for the lapse @xmath120 .
this choice of lapse is the cmc slicing gauge .
this deliberately defocusses geodesics , thus by definition preventing the breakdown of the coordinate system due to caustic formation @xcite , and so enhancing the practical longevity of the evolution . however , york s method is not absolutely general , for some spacetimes have no cmc slice to identify with @xmath11 in the first place , and in others the cmc slicing can not be maintained to cover the whole spacetime .
furthermore , these results depend on the asymptotics assumed . however , in the usual ( 3 , 0 ; 1 ) case , this method ( and its variants ) is widely accepted as a practical method by the numerical relativity community , for example in the study of colliding compact astrophysical objects @xcite .
of importance for the viability of this method in this application , recent evidence suggests that problems involving gravity wave emission can be treated thus @xcite .
however , some of the other attractive features used to gain control in the york method are only known to occur for the @xmath7 case and have mainly been studied in depth for the compact without boundary and asymptotically - flat cases of ( 3 , 0 ) data construction .
first , the study of the usual ( 3 , 0 ; 1 ) lichnerowicz equation has been based on its ellipticity ( and elliptic methods are absolutely not generalizable to the ` hyperbolic ' equations ) .
the ( 3 , 0 ; 1 ) lichnerowicz equation is well - studied , including with most fundamental matter fields @xcite , and for sobolev spaces matching those then used in the gr cp @xcite .
second , the study of the cmc slicing equation has been based on its ellipticity , so the theory of existence of cmc slicings @xcite is likely to be signature - dependent .
third , the method depends on the asymptotics used and has only been studied in the ( 3 , 0 ) compact without boundary @xcite and asymptotically flat @xcite cases .
fourth , for @xmath7 there is the useful property that certain local data patches can be proved to suffice for the treatment of an astrophysical problem , by building on the notion of dod .
as mentioned , this sort of technique is also applicable to protect pieces of local data obtained by magaard s method on a ( @xmath1 , 0 ) hypersurface .
investigation of whether the @xmath9 ` wave lichnerowicz equation ' has good existence and uniqueness properties could be interesting .
the natural setting for this is as a cauchy problem ( for even the 2-dimensional wave equation is ill - posed as a dirichlet problem ) . assuming that there exists a cauchy surface in the ( 3 , 1 ) spacetime sense , one can attempt forward and backward evolution to produce a global data set ( assuming also that the decoupled procedure for finding @xmath121 also yields a global solution ) .
one must remember however that the next stage is still to be a sideways cauchy problem . whilst this suggested procedure for the data could conceivably produce global data sets in some subcases and thus avoid the information leak problem , the other difficulties we described in sec 2.4 remain . therefore in our view
the ` wave lichnerowicz equation ' is unlikely to be suitable as a general method .
we favour instead the ( 4 , 0 ; 1 ) approach in sec 6 . that said , the ` wave lichnerowicz equation ' may still serve as the basis of a useful heuristic method .
we consider this in sec 4.7 .
we finally discuss the thin sandwich method and why we do not provide an ( @xmath1 , @xmath2 ; @xmath4 ) hamiltonian formulation . in gr
the lapse may be _ algebraically eliminated from the action by use of its own variational equation to form a reduced baierlein sharp
wheeler action @xcite .
suppose that ( @xmath102 , @xmath122 ) is prescribed on a hypersurface along with @xmath34 and @xmath123 .
the thin sandwich conjecture @xcite is that in reasonably general circumstances there exists a unique solution to solving the variational equation for the shift ( this equation replaces the codazzi constraint ) , as a differential equation for the shift itself .
once this is done , the lapse may be reconstructed from the equation used to eliminate it , and the extrinsic curvature then follows from its definition .
the lapse may then be reconstructed from the equation used to eliminate it , and the extrinsic curvature then follows from its definition ( [ ecd ] ) . _ we note that this procedure is ( @xmath1 , @xmath2 ; @xmath4)-independent .
thus we pose the conjecture for general ( @xmath1 , @xmath2 ; @xmath4 ) .
this is also considered for thin matter sheets in sec 4.7 .
to date we are only aware of ( 3 , 0 ; 1 ) papers @xcite treating this conjecture , which contain some favourable and unfavourable results obtained mostly by use of elliptic methods . in fact , there are two variants of the conjecture : i ) the thick sandwich conjecture in which lower - dimensional metrics are prescribed on two nearby hypersurfaces .
ii ) the thin sandwich conjecture proper , as stated above . a ( 3 , 1 ; 1 ) hamiltonian formulation is not provided because although it is insensitive to @xmath4 @xcite , it is sensitive to @xmath2 through the intimate involvement of the space of geometries . whilst the usual hamiltonian formulation is based on the space of riemannian geometries ( superspace ) @xcite ,
the space of semi - riemannian geometries is reported to be not even hausdorff @xcite .
first , we present a simple instance common in the higher - dimensional literature to which sec 2.1 - 3 is applicable . the _ warpfactor split @xcite g_cd = ( _ ll ^2(x^ , z ) & 0 + 0 & w(x^ , z)f_(x^ ) ) , w(x^ , 0 ) = 1 [ wan ] is a simple subcase of the @xmath23-dynamics scheme , in which the metric is allowed to @xmath23-evolve only in its scale , away from the @xmath124 hypersurface where it is taken to be known .
then ( [ ecd ] ) leads to @xmath125 which gives the equation = - [ wftr ] for the warpfactor .
for example , using the ansatz @xmath126 in ( [ wftr ] ) gives an exponential randall sundrum type warpfactor .
whereas the split ( [ wan ] ) does not cover very many cases , our scheme exhibits generalizations for it : to permit the whole metric to evolve and to recognize the gauge freedom in @xmath127 , which should ideally be used to separate coordinate effects from true physics in the spirit of @xcite .
full , overtly @xmath23-dynamical schemes are used in particular examples for domain walls @xcite , braneworld black holes ( such as for the pancake or cigar bulk horizon shape problem @xcite ) and for braneworld stars @xcite .
whereas in the randall sundrum model the higher and lower - dimensional cosmological constants balance out leaving vacuum ( minkowski ) on the brane , more generally a brane would consist of a thin sheet of matter a junction .
we study such [ shiromizu maeda sasaki ( sms)-type ] braneworlds below , starting first however with a careful recollection of where the underlying israel junction conditions come from .
assume that we have a ( @xmath0 1)-dimensional thin matter sheet in a @xmath0-dimensional bulk .
in all cases considered , the bulk s extra dimension is spatial ( @xmath5 ) .
our discussion follows @xcite most closely whilst keeping the unraised index positions of israel s original work @xcite . whereas the requirement of well - defined geometry dictates that the metric is continuous across the thin matter sheet yielding the junction condition ( j.c ) ^+_- f_^+ - f_^- = 0 , [ jcf ]
discontinuities in certain derivatives of the metric are permissible .
consider then the 3 projections of the einstein tensor @xmath128 .
we use the @xmath5 cases of the codazzi and gauss constraints ( [ cod ] ) and ( [ gauss ] ) for @xmath129 and @xmath130 respectively . for @xmath131 ,
the following construction is used .
one begins by writing down the contracted gauss equation ( [ contg ] ) and subtracts off @xmath132 times the doubly - contracted gauss equation ([gpp ] ) : c r _ + - f _ c - + + c r _
= + r_f _ c r _
+ -f _ c - + + c k k _ + k _
^k _ + f_. [ bfbwefes ] the following steps are then applied .
step 1 : the ricci equation ( [ thirdproj ] ) is used to remove all the @xmath133 .
step 2 : the contracted ricci equation ( [ tpcont ] ) is used to remove all the @xmath55 .
thus ( c r _ + - f _ ) c - + + c + f _ = ( c r _ + -f _ ) c - + + c k k _ + k _
^k _ + f_. [ step0 ] this is then rearranged to form the `` gr cp '' geometrical identity _
= g _ - k k _ + 2 k _ ^k _ + f _ + .
[ grcpid ] @xmath134 _ -^+ = 0 , [ g_]_-^+ = 0 , [ jc0 ] _ -^+ = [ k _ - f_k]^+_- .
[ primjn ] the derivation of this last equation makes use of normal coordinates ( in which case the hypersurface derivative @xmath135 becomes the normal derivative @xmath136 ) and the rearrangement to the ` israel ' geometrical identity _
= g _ + k k _ + 2 k _ ^k _ + f _ + ( k _ - f_k ) [ israel ] via the definition of extrinsic curvature ( [ ecd ] ) to form the complete normal derivative @xmath137 .
step 3 : one then further assumes that the ( 4 , 1)-dimensional efe s @xmath138 hold . if one then additionally @xmath139 0 = y _ , 0 = y _ , [ jc0 m ] _ -^+
= ( y _ - f _ ) [ prez2 ] ( performing a trace - reversal to obtain the last equation ) . we next recall the method sms use to obtain their befe s @xcite .
they begin by forming the ( 3 , 1)-dimensional einstein tensor @xmath140 just like we obtain ( [ bfbwefes ] ) above .
sms then apply three steps to this equation .
step s1 : using the definition of the weyl tensor , @xmath141 is replaced by the electric part of the weyl tensor , @xmath142 and extra terms built from the projections of @xmath143 .
step s2 ( = step 3 of the above subsection ) : the ( 4 , 1)-dimensional efe s are then assumed , which permits one to exchange all remaining projections of @xmath143 for ( 4 , 1)-dimensional energy - momentum terms .
only when this is carried out does ( [ bfbwefes ] ) become a system of field equations rather than of geometrical identities .
we refer to the field equations at this stage as `` timelike hypersurface efe s '' ( thefe s ) , as opposed to the braneworld efe s which arise at the next stage .
step s3 : a special subcase of thefe s are braneworld efe s ( befe s ) , which are obtained in normal coordinates by choosing the ( thin ) braneworld energy - momentum tensor ansatz _
ab = y_ab(z ) - g_ab , _ ab ( t_ab - f_ab ) , t_abz^a = 0 , [ bwem ] where @xmath144 is the energy - momentum of the matter confined to the brane .
this is a specialization due to the specific presence of ( 3 , 1 ) and ( 4 , 1 ) cosmological constants @xmath33 and @xmath145 , and by @xmath145 being the only bulk contribution .
a more precise formulation of this , and generalizations , are the subject of sec 4.5 .
one then adopts the j.cs ( [ jcf ] ) , ( [ jc0 ] ) , and ( [ prez2 ] ) with the additional supposition of @xmath146 symmetry : - k _ _
^+ = - k_^- _ = - ( y _ - f _ ) = - ( t _ - f _ ) , [ 41jc ] where the 5-dimensional gravitational constant @xmath147 has been made explicit .
then sms s befe s read g _
= l^ _ + q^ _ - e _ , where @xmath148 and @xmath149 are the terms quadratic in , and linear together with zeroth order in @xmath150 respectively , given by q^ _ = _ 5 ^ 4 [ befe ] l_^ = -(+ ^2)f _ + t_. as opposed to the ( 3 , 1)-dimensional efe s , sms s befe s are not closed since they contain the unspecified electric part of the weyl tensor @xmath151 .
although it also contains 15 equations , the sms befe gauss codazzi system is not equivalent to the ( 4 , 1)-dimensional efe s : indeed sms write down further third - order equations for the `` evolution '' away from the timelike brane of @xmath151 , by use of the @xmath23-derivative of the contracted gauss equation ( [ contg ] ) , bianchi identities and the ricci equation ( [ thirdproj ] ) .
this then involves the magnetic part of the weyl tensor @xmath152 , the evolution " of which follows from further bianchi identities .
this full brane - bulk sms system is then closed .
step 4 : in practice , however , instead of the difficult treatment of this third - order system , other practitioners have often worked on the sms befe s alone .
this involves either the ad hoc prescription of the functional form of @xmath151 ( sometimes taken to be zero ) .
in fact it is often first decomposed according to a standard procedure @xcite .
because the original functional form is unknown , the functional forms of each of the parts defined by the decomposition is also unknown .
some of these parts are set equal to zero whereas other parts are taken to have other functional forms ( in particular a radiation fluid term ) .
these terms are then argued to be small in the circumstances arising in the inflationary @xcite and perturbative @xcite treatments .
however other interpretations may be possible .
having dealt with @xmath151 in one of the above ways , the form ( 53 ) of @xmath153 is then often taken to be uniquely defined and the starting - point of many works on brane cosmology @xcite .
however , sms s procedure is far from unique .
it turns out that there are many reformulations of the befe arising from geometrical identities .
each has a distinct split of the non - einsteinian befe terms into ` bulk ' and ` brane ' terms .
whereas all these formulations are clearly equivalent , their use helps clarify how to interpret sms s braneworld .
were one to truncate the ` bulk ' terms in each case ( in direct analogy with the usual practice of throwing away the weyl term ) , then the befe s obtained in each case would generally be inequivalent . the weyl term in sms s befe s has been the subject of much mystery .
how should it be interpreted ?
is it right to throw it away and if so under which circumstances ? we emphasize that our stance is broader than merely about what functional form is allocated to the weyl term ( e.g whether it is zero everywhere ) .
it is about how formulations can be chosen in which the befe s are not explicitly formulated in terms of a weyl term .
we first remove some misconceptions as to how sms s procedure leads to a befe containing a weyl term .
does it have anything to do with the modelling of braneworld scenarios ?
no , for the weyl term is already in the sms thefe before the braneworld energy - momentum ansatz is invoked .
furthermore , all the procedures used in sms s method are independent of signature and dimension .
thus this issue of a weyl term must have already arisen long ago in the study of the gr cp .
so why is there no manifest weyl term piece in the gr cp formulation of the efe s ?
the answer is simple .
indeed , we have already seen the answer in sec 4.1 since it is fundamentally tied to how the crucial junction condition ( [ prez2 ] ) is obtained in the first place!. in the `` gr cp '' and israel procedures , _ one uses the ricci equation ( [ thirdproj ] ) to remove the @xmath154 term . if there is no early use of the ricci equation , one is left with @xmath151 in the thefe s , which requires later use of the ricci equation to `` evolve '' it .
_ so there is a choice as to whether one formulates the befe s with or without an explicit weyl term . in the usual treatment of the split of the efe s (
sec 2.2 ) , one does not use an explicit weyl term . in the ( 4 , 1)-d case ,
this gives a well - understood system of 15 p.d.es in the variables @xmath155 .
the option of using an explicit weyl term gives a considerably larger , more complicated system of p.d.es with variables @xmath155 , @xmath151 and @xmath156 .
in fact a similar scheme known as the _ threading formulation @xcite is sometimes used in the usual ( 3 , 0 ; 1 ) application . the idea behind
this formulation is to treat as primary the geodesic congruences perpendicular to the foliation rather than the foliation itself .
one then uses only that information on the hypersurfaces that arrives along the impinging geodesic congruence , on the grounds that this is the physically - significant information .
thus it is a ` deliberately incomplete system ' from the foliation perspective . whereas this and the sms formulation are similar in their use of weyl variables
, the sms system does not appear to be a threading formulation .
also , in any case the idea of having a deliberately partial system in the threading formulation is clearly tied to signature - specific physical reasons which do not carry over to the signature relevant to sms s equations .
we also note that some other third - order reformulations of the @xmath157 split of the efe s are sometimes used to seek to cast the efe s into hyperbolic forms that manifestly have theorems associated with them @xcite .
whereas this is precisely the sort of result that is spoiled by considering instead a sideways split , it serves to illustrate that what at first seems a ` mere reformulation ' of a set of equations can in fact be used to prove highly nontrivial theorems .
so , whereas similar complicated formulations have been used elsewhere in the gr literature , sms s unstated motivation to have a complicated formulation does not coincide with the motivation elsewhere in the literature .
below we bring attention to many reformulations of sms s system , so we ask : what is the motivation for the original sms formulation ?
should the use of some simpler second - order formulation be preferred ? is sms s formulation or any other third- or second - order formulation singled out by good behaviour , either in general or for some particular application ? _ also , from first principles the sms procedure to obtain their befe is quite complicated . for
, since they use the j.c obtained by the israel procedure , their procedure actually entails beginning with the whole israel procedure ( steps 1 to 3 of sec 4.1 ) , and then choosing to reintroduce @xmath158 and @xmath55 by reverse application of steps 1 and 2 .
this is followed by the weyl rearrangement ( step s1 ) , the use of the efe s ( step s2 ) and the substitution of the j.c into the extrinsic curvature terms in the braneworld ansatz ( step s3 ) . however , despite being complicated , all is well with sms s scheme since any befe s obtained by other such combinations of careful procedures will always be equivalent because the different steps are related by geometrical identities .
step 4 however is not an instance of being careful as it is a truncation .
our first point is that whereas in sms s formulation the non - einsteinian terms in the befe might be regarded as a bulk - like @xmath151 and a term quadratic in the brane energy - momentum , in other formulations the content of these two terms can be mixed up . in general , befe s contain a group of non - einsteinian ` bulk ' terms we denote by @xmath159 ( which include both weyl terms and normal derivatives of objects such as the extrinsic curvature ) , and a group of non - einsteinian ` brane ' terms that depend on the brane energy - momentum .
thus any temptation to discard the weyl term in the sms formulation ( on the grounds that it involves the unknown bulk over which one has no control ) should be seen in the light that if one considered instead a reformulation , then there would be a similar temptation to discard the corresponding ` bulk ' term , which would generally lead to something _ other than the weyl term being discarded .
thus for each formulation , the corresponding truncation of the ` bulk term ' would result in inequivalent residual ` braneworld physics ' .
this is because there are geometrical identities that relate ` bulk ' and ` brane ' terms , so that the splits mentioned above are highly non - unique and thus not true splits at all .
we take this as a clear indication that any such truncations should be avoided in general . instead
, the full system must be studied . _
our second point is that each possible bulk spacetime may contain some hypersurface on which a given @xmath159 vanishes . then if one identifies this hypersurface with the position of the brane , one has a solution of the full brane - bulk system and not a truncation .
for example , in any conformally - flat bulk , by definition @xmath160 and therefore @xmath161 on all hypersurfaces . thus any of these could be identified as a brane to form a genuine ( rather than truncated ) @xmath162 braneworld . from this
, we can see that _ the sms formulation is particularly well adapted for the study of conformally - flat bulks such as pure ads .
this motivates sms s formulation as regards this common application .
however , also consider repeating the above procedure with some @xmath163 .
this would correspond to a genuine ( rather than truncated ) braneworld model with distinct braneworld physics from that given by sms s particular quadratic term .
note that given a model with some @xmath164 , the befe formulation for which @xmath159 is the bulk term is particularly well adapted for the study of that model .
thus different formulations may facilitate the study of braneworlds with different braneworld physics . in the context of conformally - flat spacetimes
, it is probably true that the @xmath162 braneworlds outnumber the braneworlds for which any other ( or even all other ) @xmath164 since these other conditions are not automatically satisfied on all embedded hypersurfaces .
rather , each of these other conditions constitutes a difficult geometrical problem , somewhat reminiscent of the question of which spacetimes contain a maximal ( k = 0 ) or cmc slice @xcite
. however , generic spacetimes are not conformally - flat . for a generic spacetime , we see no difference between the status of the condition ` @xmath162 on some hypersurface ' and the condition ` any other particular @xmath164 on some hypersurface ' . because braneworlds constructed in each of these cases have a different residual quadratic term and thus a propensity to have distinct braneworld physics , and because we do not know how frequently each of these cases occur , we question whether anything inferred from conformally - flat models with the sms quadratic term need be typical of the full sms brane - bulk system .
confirmation of this would require study of the full range of difficult geometrical problems @xmath164 , and the construction of concrete examples of non - sms quadratic term braneworld models together with the assessment of whether their braneworld physics is conceptually and observationally acceptable . _ for the moment we study what is the available range of reformulations and thus of @xmath159 . to convince the reader that such reformulations exist , we provide a first example before listing all the steps which are available for reformulating the befe s .
assume we do not perform all the steps implicit within sms s work but rather just the israel steps to obtain the j.c and then use it in the field equation ( [ israel ] ) that gave rise to it ( as done in @xcite ) , or ( as done below ) use it in the `` gr cp '' field equation following from ( [ grcpid ] ) . in other words , why not apply the braneworld ansatz to e.g the israel or `` gr cp '' formulations rather than to the sms formulation ?
in the `` gr cp '' case we then obtain g _
= l _ + q _ + b _ [ cpbefe ] q _ = - [ qcp ] l _ = ( t - 2 ^ 2 ) f _ + _ 5 ^ 2[t _ - ( + ) f _ ] [ lcp ] b _ = f _ - . [ bcp ]
this example serves to illustrate that choosing to use a different formulation can cause the ` brane ' quadratic term @xmath148 to be different .
also note that this formulation makes no explicit use of the weyl term .
thus this befe , along with the gauss and codazzi constraints , forms a small second - order system , in contrast with the much larger third - order sms system .
now we further study the list of steps @xcite which may be applied in the construction of befe s .
steps s1 and 3 together mean that the weyl ` bulk ' term @xmath151 is equivalent to the riemann ` bulk ' term together with matter terms .
this swap by itself involves no terms which are quadratic in the extrinsic curvature .
step 1 says that the riemann ` bulk ' term is equivalent to the hypersurface derivative of the extrinsic curvature together with a @xmath165 term .
steps s1 and 3 together say that the hypersurface derivative of the trace of the extrinsic curvature is equivalent to a matter term together with a @xmath166 term .
furthermore , one can use both steps 2 - 3 and step 1 , on arbitrary proportions ( parametrized by @xmath108 and @xmath167 ) of @xmath168 and of @xmath55 : ll g _ &
= g _ - ( 1 + ) r_+ ( 1 - ) r_f _ + + & + k k _ + ( - 1 ) k_^ k _ - f _ + ( - ) k f _ [ 2param ] this introduces freedom in the coefficients of the @xmath165 and @xmath166 contributions to the quadratic term @xmath169 in the thefe s .
we next find further freedom in @xmath170 by choice of the objects to be regarded as primary .
we are free to choose a ` bulk ' term described by hypersurface derivatives @xmath171 ( which are partial derivatives @xmath136 in normal coordinates ) of objects related to the extrinsic curvature @xmath172 by use of the metric tensor ( including its inverse and determinant @xmath173 ) .
the underlying reason for doing this is that it is just as natural to treat such an object , rather than the extrinsic curvature itself , as primary ( see below for examples ) . upon careful consideration ,
there are three separate ways such objects can be related to the extrinsic curvature : raising indices , removing a portion of the trace by defining @xmath174 , and densitizing by defining @xmath175 .
the hypersurface derivatives of these objects are related to those of the extrinsic curvature by _
k _ = _ ( f_k^ _ ) = f _ _ k^ _ - 2_k^ _ , [ i ] _ k _ = _ k^ _ + ( _ kf _ - 2 _ ) , [ 64 ] _ k _ = ( f)^-_1_(f^_1k _ ) + 2_1kk_. [ 65 ] further useful equations arise from the traces of these : _ k = f^ _ k _ + 2 , [ 63 ] _
k = [ f^ _ _ _ - 2(k^2 - k ) ] , [ new ] _ k = ( f)^-_2_(f^_2k ) + 2_2k^2 , [ 66 ] where the @xmath176 in ( [ new ] ) and ( [ new ] ) has been obtained via ( [ 66 ] ) .
the following examples of @xmath177 illustrate that the use of such objects is entirely natural : @xmath178 is the @xmath121 commonly used in the ivp literature , and @xmath179 appears in the guise of forming the complete normal derivative in the israel procedure .
also , the `` gravitational momenta '' are @xmath180 .
the above thorough consideration of possible ` bulk ' terms permits all four thefe terms homogeneously quadratic in the extrinsic curvature to be changed independently
. one may think that we have a redundancy in providing 8 ways to change only 4 coefficients .
however , one can afford then to lose some of the freedoms by making extra demands , of which we now provide four examples of relevance to this paper .
first , one could further demand that there is no weyl term in the thefe s ( as discussed in sec .
second , unequal densitization of @xmath172 and @xmath181 ( @xmath182 ) corresponds to interpreting the fundamental variable to be some densitization of the metric rather than the metric itself .
whereas this is again a common practice ( for example the scale - free metric of the ivp literature is @xmath183 in dimension @xmath10 ) , the use of such an object as fundamental variable does appear to complicate the isolation of the ( 3 , 1 ) einstein tensor truly corresponding to this fundamental variable .
thus this option is not pursued in this paper .
third , one may start by declaring that one is to use particular well - known primary objects ( such as the `` gravitational momenta '' ) and still desire to be left with much freedom of formulation .
fourth , one could declare that one is to use the raised objects given by ( [ 63 ] ) , in which case the further ability to change coefficients by use of ( [ 64 ] ) is lost , since moving a kronecker delta rather than a metric through the derivative clearly generates no terms quadratic in the extrinsic curvature . as a consequence of the above freedoms , there are many formulations in which all four coefficients vanish , and hence @xmath184 . from this
it follows that @xmath148 is zero [ and it is easy to show that all instances of @xmath185 follow from @xmath184 ] .
thus it suffices to seek for cases of thefe s with @xmath184 to obtain all cases of befe formulations that have no quadratic term .
we now motivate these formulations and then choose to exhibit three that comply with some of the extra demands in the previous paragraph .
the diversity of ` brane'-`bulk ' splits ensures that truncations such as step 4 produce all possible combinations of quadratic terms as residual ` braneworld physics ' .
alternatively , one may suspect that there might be solutions to the full brane - bulk system that just happen to have a particular @xmath164 on some hypersurface which is then identified as a brane .
we speculate that each of these situations will often lead to different answers to questions of physical interest . whereas most friedmann lematre robertson walker ( flrw ) perfect fluid models with equation of state @xmath187 arising thus will be similar , differences will be more salient in models with more complicated equations of state , in perturbations about flrw ( as started in @xcite in the sms - adapted case ) and in anisotropic models ( as started by @xcite in the sms - adapted case ) .
these in turn constitute natural frameworks to seriously justify the late - time emergence of flrw behaviour and the likelihood of inflation @xcite as well as the study of singularities @xcite on the brane .
we emphasize that , for a satisfactory study of whether any particular full brane - bulk ( as opposed to truncated ) case leads to any differences from the hitherto - studied @xmath162 case , one would require a full brane - bulk solution explicitly constructed to satisfy some @xmath164 on some hypersurface within a particular given bulk spacetime .
since we currently have no such example , our arguments currently only support the far simpler idea that truncation should be avoided .
we illustrate that in different formulations , the truncation of the corresponding ` bulk ' terms can lead to big differences in the residual ` braneworld physics ' , without any of the above lengthy calculations .
we do this by formulating the ` bulk ' part so that there is no corresponding @xmath188 term at all .
thus these truncations give the ` @xmath34 ' of standard flrw cosmology rather than the ` @xmath34 and @xmath113 ' of brane cosmology @xcite . as a result whether we have a ` @xmath34 and @xmath113 ' brane cosmology depends on the choice of formulation .
so we argue that since the sms procedure followed by truncating the weyl term is a hitherto unaccounted - for choice out of many possible procedures , then adopting the particular homogeneous quadratic term of sms ( often taken as the starting - point of brane cosmology ) appears to be unjustified .
rather , we conclude that no particular truncation should be privileged as the act of truncation imprints undesirable arbitrariness into the study of the truncated system . whereas the ( 3 , 1)-dimensional trace of @xmath151 happens to be zero ( including use of antisymmetry ) and
thus might phenomenologically look like pure radiation fluid to observers on the brane , other bulk characterizations would typically not look like a pure radiative fluid .
this may open up phenomenological possibilities . also , before further study of sms s
full ( untruncated , third - order ) system is undertaken , some of the reformulations along the lines suggested in sec 4.4 might turn out to be more tractable .
in particular those reformulations which fully eliminate @xmath168 by the early use of the ricci equation are already closed as second - order systems .
these include the israel formulation in @xcite , the `` gr cp '' formulation ( [ cpbefe][bcp ] ) , and our second and third examples below , which contain neither a weyl term nor a quadratic term . for our first example , we take as the primary object the antidensitized extrinsic curvature @xmath189 so that the ` bulk ' term is ( partly ) a combination of this object s normal derivatives . the corresponding befe s are : g _ = l _ + b _ l _ = + ( 5 t _ - t)f _ , b _ = -2e _ + ( - f^ f _ ) ( where we have chosen to remove all projections of @xmath143 by the efe s ) . to derive this ,
take ( [ 2param ] ) in normal coordinates .
choose to convert all of the @xmath190 into @xmath191 by ( [ 63 ] ) : ll g _ & = g _ - ( 1 + ) r _ + ( 1 - ) r_f _ + ( + f _ ) + & + k k _ + ( - 1 ) k_^ k _ - f _ + ( + ) k f _ now choosing the primary object to be some densitized @xmath192 by ( [ 65 ] ) we have ll g _ = & g _ - ( 1 + ) r _ + ( 1 - ) r _ f _ + ( + f^ ) + & + ( 1 + 2)kk _ +
( - 1 ) k_^ k _
+ ( 2- ) k^2f _ + ( + ) k f _ ,
so clearly @xmath193 , @xmath194 and the antidensitization choice of weighting @xmath195 ensure that @xmath184 the following examples arose from asking if it is possible to find examples in which neither @xmath148 nor @xmath151 feature .
we found the following befe s : g _ = l _ + b _ l _
= t _ + ( t_- ) f _ , b _
= ( f _ - f _ ) by considering as our primary object the densitized extrinsic curvature with one index raised , @xmath196 .
we obtained these befe s by arguing as follows . in order for the befe s to contain no weyl term ,
@xmath167 is fixed to be @xmath197 .
then the only control over @xmath165 is from raising by ( [ i ] ) .
it is easy to show that this raising must be applied to the whole @xmath198 in order for the coefficient of @xmath165 to be zero .
then using @xmath199 does not change any terms quadratic in the extrinsic curvature .
also , use of distinct densities for @xmath181 and @xmath200 does not appear to make sense since both quantities are related to @xmath172 by a single use of the inverse metric .
although all these restrictions make the outcome unlikely , the use of ( [ 65 ] ) and ( [ 66 ] ) alone suffices to obtain the above example : ll g _ & = g _ + ( 1 - ) r_f _ + f^- ( - + f _ ) + & + ( 1 -2 ) k k _ + ( 2- ) k^2f _ + ( - ) k f _ which has no quadratic terms if @xmath201 and @xmath202 ( ` densitization ' weight ) .
another possibility is to replace @xmath203 by @xmath204 .
although this does not immediately do anything about the quadratic terms , if we also convert a portion @xmath205 of @xmath206 into @xmath207 we obtain ll g _ & = g _ + ( 1 - ) r_f _ + f^- ( - + ( - ) f _ ) + & + ( 1 -2 ) k k _ + ^2f _ + f _ , which requires @xmath208 , whereupon the two remaining equations become identical : @xmath209 , which clearly has many solutions .
a particularly neat one is to take @xmath210 so that the primary objects are ` gravitational momenta ' and @xmath211 so that only two normal derivative terms appear in the ` bulk ' term .
then @xmath212 so the befe s read g _
= l _ + b _ , l _ = t _ + f _ , b _ = ( f _ - f^ f _ ) . of course , it would make sense to particularly investigate the difficult geometrical problem ` @xmath164 on some embedded hypersurface ' for such @xmath164 corresponding to no quadratic terms , since by the same arguments as above , such a model would be sure to give braneworld physics distinguishable from that hitherto studied .
first , so far we have talked in terms of the @xmath213 split of the matter contribution to relate our work as clearly as possible to its predecessors in the literature . however , from the outset @xcite it was pointed out that this split is not unique . on these grounds
we would prefer to work with the unique trace - tracefree split in which all the @xmath214 contributes to the trace part .
the ( 4 , 0 ) version of this split is used in sec 6 .
second , given a fixed type of bulk energy - momentum such as the @xmath56 of nkk or the @xmath215 , then establishing an embedding requires the existence of a suitable compensatory characterization of the bulk geometry .
the gr line of thought would be to only take results within such schemes seriously if they are robust to the addition of bulk matter fields .
of course privileged choices of bulk could arise from further theoretical input .
we argue below that the theoretical arguments behind some privileged choices in the literature for the bulk energy - momentum are not convincing enough to anchor strongly credible physical predictions .
one would rather require rigorous and general theoretical input following directly from some fundamental theory such as string theory .
the vacuum choice of bulk of the induced matter nkk approach is always possible given the premises of the cm result .
then @xmath216 is claimed to be a complete geometrization of matter @xcite .
however , the cm result holds equally well for any other analytical functional form of the energy - momentum .
furthermore , this approach considers only 1-component ( ` induced ' ) matter ; counting degrees of freedom shows that it can not be extended to many important cases of fundamental matter . whereas allowing for more extra dimensions could improve similar situations @xcite , unification requires geometrization of the fundamental matter laws themselves , whilst this ` induced matter ' approach only geometrizes solutions of the efe s coupled to matter of unspecified field dependence .
the @xmath215 choice of bulk ( an example of which is pure ads ) is clearly also always possible given the premises of a particular case of the generalized cm result .
however , the motivation we wish to discuss is the argument for pure ads bulks from string theory .
this is not generally justifiable since firstly , bulk gravitons are permitted so the bulk geometry would generically contain gravity waves .
secondly , bulk scalars ought to be permitted since they occur along with the graviton in the closed string spectrum @xcite .
the content of the closed string spectrum thus places interesting restrictions on bulk matter rather than completely abolishing it . from the perspective of 5d gr , evidence for the stability of vacuum or ads bulks ( and of any resulting physical predictions ) to the introduction of suitable bulk fields would constitute important necessary support for such models and their predictions . finally note that use of arbitrary smooth bulk @xmath57 does not affect the form of the junction conditions since only the thin matter sheet contribution to @xmath57 enters these .
the conclusion of our second point is that there is no good reason not to explore at least certain kinds of bulk matter in order to have a more general feel for how these thin matter sheet models behave @xcite .
first we explain how our criticism of ( 3 , 1 ; 1 ) methods in general in sec 2.4 are applicable to the case with thin matter sheets .
as regards the evolution " w.r.t @xmath23 , the issue of causality holds regardless of whether thin matter sheets are present , and the issue of well - posedness not being known will become particularly relevant in the study of sufficiently general situations in which rough function spaces would become necessary to describe the evolution of thin matter sheets ( see sec 6.1 ) .
the campbell magaard scheme is of limited use in models with thin matter sheets not only because analytic functions are undesirable ( and definitely inapplicable to sufficiently general situations ) but also because the junction condition imposes restrictions on @xmath172 which prevents these being subdivided into the knowns and unknowns of magaard s method .
these restrictions remain even if one considers suitably well - behaved thick matter sheet models . in the next paragraph
we outline how much we expect can be achieved with the ( 3 , 1 ) version of the york method applied on the thin matter sheet , but recall that even if this does provide data sets , one is next confronted with the difficulties of the ( 3 , 1 ; 1 ) evolution " scheme .
the main message is that the choice of methods which properly respect the difference between space and time is absolutely crucial .
thus , although at the simplest level ( 3 , 1 ; 1 ) methods which build higher - dimensional bulks about the privileged ( 3 , 1 ) worlds may look tempting , general attempts at proceeding thus are hampered by the required mathematical tools simply not being available and by these attempts not being in accord with the usual notion of causality . to date ( 4 , 1 )
worlds built from ( 3 , 1 ) ones have relied on very simple specialized anstze , such as \a ) @xmath23-symmetric surfaces @xmath217 with known metric @xmath155 , whereupon the vacuum codazzi equation is automatically satisfied and then @xmath218 is required from the gauss constraint .
\b ) @xmath126 with known @xmath155 , for example to obtain the randall sundrum bulk @xcite or slightly more general solutions @xcite .
now the maximal subcase @xmath219 of the cmc condition is a generalization of a ) , whereas the full cmc condition itself is a generalization of b ) .
moreover , now the metric is to be treated as only known up to scale . thus one would generally only know
the full metric of each model s ( 3 , 1 ) world once the ` wave lichnerowicz equation ' for the embedding of this world into the ( 4 , 1 ) world is solved .
so one loses the hold from the outset on whether each model will turn out to contain an interesting ( 3 , 1 ) world .
nevertheless , some of the ( 3 , 1 ) worlds will turn out to be of interest .
furthermore , one should question the sensibleness of any ideas involving the prescription of full ( 3 , 1 ) metrics if the most general technique available fails to respect such a prescription .
this point is more significant for ( 3 , 1 ) data than for ( 3 , 0 ) data because conformally - related metrics have different non - null geodesics . for ( 3 , 0 ) data no physical significance
is attached to spatial geodesics , but for ( 3 , 1 ) there are timelike geodesics which are physically interpreted as paths of free motion of massive particles .
so a @xmath220 spacetime which is conformally related to @xmath221 is different physically ( for example one could violate energy conditions the other one does not violate ) .
one can get out of this problem by either attaching no physical significance to one s inspired guesses for @xmath221 or by hoping for unobservably tiny deviations between the geodesic curves of the two geometries .
in the specific case of thin matter sheets , by the j.cs , a ) implies that @xmath222 , whilst b ) implies that @xmath223 is a hypersurface constant .
the maximal condition implies that @xmath224 whilst the cmc one implies that @xmath223 is a hypersurface constant .
so whereas the maximal and cmc ans@xmath225tze are more general than a ) and b ) respectively , they are nevertheless restricted in this braneworld application .
notice , however , that a number of interesting cases are included : vacuum , radiative matter and electromagnetic matter are all among the @xmath224 spacetimes .
it is important to note that unlike the usual gr application , the choice of a hypersurface to be a brane is not a choice of slicing because localized energy
momentum is to be pinned on it .
almost all reslicings would fail to isolate this energy - momentum on a single slice .
we know of no good reason why the brane should be cmc nor what value the cmc should take on it
. however , at least this is a well - defined notion and it is substantially simpler to solve for than in general because of the decoupling of the gauss and codazzi constraints . as regards possible use of either of the two forms of the thin sandwich conjecture to treat branes , we first distinguish between thin sandwiches between 2 nearby branes and thin sandwiches which have a brane on one side and an undistinguished hypersurface on the other
. one should be aware that non - intersection requirements may be different in these two cases , and also different from that of the original tsc setting of 2 unprivileged spacelike hypersurfaces .
second , one would have to take @xmath226 as unknown until it can be deduced from the @xmath172 evaluated from the ts procedure .
finally we caution that tsc schemes need not always exist .
they require the `` lapse '' to be algebraically - eliminable , which for example is not the case for the analogous @xmath22 of the kk split .
we have argued in secs 2 - 3 in favour of ( 4 , 0 ; 1 ) schemes over ( 3 , 1 ; 1 ) schemes .
however , it is often suggested that ( 3 , 1 ; 1 ) schemes may be useful since in certain particular examples ( 3 , 1 ) spacetime singularities are ` removed ' by embedding into nonsingular ( 4 , 1 ) spacetimes @xcite .
whereas embeddings are undoubtedly useful tools in the study of ( 3 , 1 ) spacetimes ( for the purpose of algebraic classification @xcite ) , we ask what is the status of their use to remove cosmological singularities .
is it mathematically rigorous , generally applicable and physically meaningful ? by ` singular ' we mean geodesically - incomplete @xcite . in the study of singularities , timelike and
null geodesics ( t.g s and n.gs ) are the only objects of fundamental importance since they are the curves privileged by the free travel of massive and massless particles respectively .
the other objects associated with the study of singularities arise as a matter of convenience rather than being fundamental .
such objects include the expansion and shear of geodesic congruences , and curvature scalars .
because these objects are closely - linked to material in this paper , we focus first on them ( sec 5.1 ) , explaining how they arise in the study of singularities and then demonstrating how the embedding and embedded versions of these objects can be very different ( sec 5.2 ) .
we then discuss the more important issue of geodesic incompleteness in the context of embeddings ( sec 5.3 ) .
an example is used to illustrate some of these points and others ( sec 5.4 ) .
a lack of rigour in singularity - removal claims is uncovered in sec 5.5 .
we finally treat the case with thin matter sheets in sec 5.6 . for ( @xmath227 , 1)-dimensional cosmology , smooth congruences of past - directed normal t.gs with normalized tangents denoted by @xmath228 are considered . the decomposition _
ab d_au_b = h_ab + _ ab + _ ab [ bsplit ] provides the _ expansion @xmath229 and _ shear @xmath230 .
the twist @xmath231}$ ] is zero for the normal congruences considered here , in which case @xmath232 is an extrinsic curvature @xmath233 .
the corresponding normal raychaudhuri equation [ c.f ( [ normray ] ) ] can now be considered .
although for @xmath234 this would usually mean that a caustic develops , under certain global conditions a contradiction about the existence of conjugate points arises .
singularity theorems are thus obtained ; in a cosmological context the simplest of these is ( hawking , theorem 1 of @xcite ) . _ _ for globally - hyperbolic ( 3 , 1 ) gr spacetimes obeying the sec and such that @xmath235 on some smooth ( spacelike ) cauchy surface @xmath236 , then no past - directed timelike curve from @xmath236 can have greater length than @xmath237 . by the definition of cauchy surface ,
all past - directed t.gs are among these curves and are thus incomplete , so the spacetime is singular .
such theorems generalize to ( @xmath227 , 1 ) spacetimes ( provided that these obey analogous energy conditions ) in the obvious way since raychaudhuri s equation clearly behaves in the same way for all @xmath227 and the global part of the arguments uses topological space methods that do not care about dimension . it must be noted that the singularity theorems are about existence whilst saying nothing about the nature of the singularity .
ellis and schmidt began to classify singularities according to their properties @xcite , a difficult study which may never be completed @xcite .
below we consider only the most elementary type of genuine singularities : _ curvature singularities , for which at least one spacetime curvature scalar such as @xmath238 or @xmath239 blows up .
_ here we explain how knowledge of gauss hypersurface geometry renders it unsurprising that there are instances of singular spacetimes being ` embeddable ' in nonsingular ones .
this is because the behaviour of higher and lower dimensional objects used in the study of singularities clearly need not be related .
as a first example , consider expansion . in the normal case
, @xmath240 and @xmath241 from which follows = - ^_. so a blowup in @xmath242 need not imply a blowup in @xmath243 . thus the ( 4 , 1 ) spacetime perspective on focusing of geodesics
can be completely different from the perspective on some ( 3 , 1 ) hypersurface .
figure 6 shows how one s ( 3 , 1 ) notion of expansion generally corresponds to expansion and shear from the perspective of an embedding ( 4 , 1 ) spacetime .
in particular , what look like caustics or singularities in the ( 3 , 1 ) spacetime could well correspond to no such feature in some surrounding ( 4 , 1 ) spacetime . as a second example , consider curvature scalars , in particular the ricci scalar .
preliminarily , gauss outstanding theorem is a clear indication that extrinsic curvature can compensate for differences between higher and lower - dimensional intrinsic curvature .
our simple idea is to consider the implications for rigorous embedding mathematics of the generalization of this to the case where lower - dimensional curvature scalars become infinite . from the form ( [ agauss ] ) for the ( 3 , 1 ; 1 ) gauss constraint , 2= 2r _ - r = ^2 - k^ ^ - r [ singrem ] clearly - at least for some ( 3 , 1 ) worlds which have ricci - scalar curvature singularities @xmath244 - there will be surrounding ( 4 , 1 ) worlds in which @xmath238 ( and the 5-d @xmath34 ) remain finite . for it could be that @xmath245 suffices to compensate for @xmath244 .
thus it could be shear , expansion or both that dominate the compensation for @xmath244 .
if this involves @xmath246 , it means a ( 4 , 0 ) caustic forms in the ( 4 , 1 ) spacetime corresponding to the ( 3 , 1 ) spacetime s singularity . yet
this is not the only nonsingular possibility ; if @xmath247 it may be the shear that compensates , including cases in which the blowup is pure shear .
finally , clearly the higher - dimensional singularity theorems hold so even if one were to succeed in excising singularities from a lower - dimensional model , one would typically expect singularities to occur elsewhere in the resultant higher - dimensional models .
the nature of these higher - dimensional singularities may not be the same as that of the excised lower - dimensional ones .
so , interestingly , by allowing extra dimensions , one would be even less certain of the character of singularities .
the fundamental importance of geodesics is problematic for embeddings since in general the ( 3 , 1)-dimensional geodesics are not included among those of an embedding ( 4 , 1)-dimensional spacetime .
there is then the dilemma of which of these sets of curves should be the physically privileged ones and thus be the set of curves whose extendibility is in question .
if one wishes to postulate ( 4 , 1)-dimensional gr , in addition to the ( 4 , 1)-dimensional efe s one must require that the matter follows the ( 4 , 1)-dimensional geodesics .
then the issue of extendibility of the original ( 3 , 1 ) geodesics becomes irrelevant since they are unprivileged curves in the ( 4 , 1 ) spacetime .
but this is at the price of altering what the effective ( 3 , 1)-dimensional physics is : often it will be of affine - metric type and thus in principle distinguishable from ordinary ( 3 , 1 ) physics .
one can not then necessarily claim that the recovered ( 3 , 1 ) physics is as close to the usual ( 3 , 1 ) physics as is widely claimed in the braneworld literature if these do not include the ( 3 , 1 ) geodesics , then the ( 3 , 1 ) question of what happens to these becomes physically irrelevant .
one would not then be extending one s ( 3 , 1 ) geodesics but replacing them with distinct ( 4 , 1)-dimensional geodesics , which might be viewed as unsatisfactory .
there is the further complication in braneworld models that one might wish for ( 3 , 1 ) and ( 4 , 1 ) geodesics to both be relevant , as the paths of brane - bound and brane - unbound matter or gravity waves .
it is then not clear what is meant by a singularity
exactly which curves are supposed to be incomplete ?
we end with two loose possibilities about incompleteness which might sometimes be related : the ( 3 , 1 ) singularities might be interpretable in terms of a ( 3 , 1 ) hypersurface that becomes tangent to the characteristics of the bulk equations , and incompleteness might correspond to t.gs and n.gs being forced to exit the ( 3 , 1 ) hypersurface .
the second possibility is a variant on the theme of ( 4 , 1 ) and ( 3 , 1 ) geodesics whereby the ( 3 , 1 ) geodesics are pieces of ( 4 , 1 ) geodesics which are extendible only by replacing a piece of the original foliation with a new one that extends into what was originally regarded as the extra dimension .
the following example illustrates many of the above points and leads us to further comments .
consider a ( 4 , 1 ) spacetime with metric ( see e.g @xcite ) g_ab = ( g_tt , e _ , g_zz ) = ( -z^2 , t^ z^ _ , ) , q > 1 .
[ funnymink ] this is simple to treat because : \1 ) if we foliate it with constant @xmath23 hypersurfaces , a portion of each @xmath248 const hypersurface has induced on it a flat flrw cosmology . in particular
, the coordinates have been chosen so that the @xmath249 hypersurface is the flrw cosmology with scale factor @xmath250 .
we restrict attention to @xmath251 due to the dec . for @xmath252
there is a ( 3 , 1 ) ricci scalar curvature singularity ( @xmath253 is the radiation universe , whence @xmath254 ) .
\2 ) the ( 4 , 1 ) spacetime is in fact minkowski .
first , foliating the flat flrw cosmology with constant @xmath255 surfaces , @xmath256 as @xmath257 , so there is ( 3 , 0 ; 1 ) focusing as the big bang is approached .
furthermore , _ only focusing occurs : @xmath258 .
next foliate the ( 4 , 1 ) spacetime with constant @xmath255 surfaces .
the ( 4 , 1 ) spacetime contains many flrw worlds on constant-@xmath23 surfaces .
build a congruence by collecting the flrw geodesics on each @xmath248 const slice .
then the ( 4 , 0 ; 1 ) expansion @xmath259 also blows up as @xmath260 , but there is also a blowup of the corresponding ( 4 , 0 ; 1 ) shear : ^_ab ( -e _ , 3g_zz ) , so we have case 2 of fig 7 . also , whilst still both @xmath261 and @xmath262 in this case , we know that the former viewed from within the @xmath249 ( 3 , 1 ) hypersurface corresponds to a genuine ( 3 , 1 ) singularity , whilst the latter in ( 4 , 1 ) would be a mere caustic .
_ second , consider the @xmath249 slice of the ( 4 , 1 ) spacetime . here
the ( 3 , 1 ; 1 ) expansion and shear are = , ^ _
= -(3 , t^ _ ) .
thus for the physical range , both blow up as @xmath257 .
so the spacetime includes a @xmath124 , @xmath263 point , at which there is a ( 4 , 0 ) caustic .
these blow - ups combine in @xmath264 to cancel @xmath265 for all values of @xmath227 ; for @xmath253 the shear and expansion contributions exactly cancel each other .
the unphysical @xmath266 case is an example of a pure shear blowup . although the congruence two paragraphs up is built to naturally include the geodesics of all the included flrw universes , these turn out not to be the ( 4 , 1 ) geodesics ( nor pieces of them ) .
rather , by considering the coordinate transformation between ( [ funnymink ] ) and standard ( 4 , 1)-dimensional minkowsi coordinates , it is easy to show that the minkowski geodesics pierce the @xmath248 constant surfaces that are the flrw universes ( more significantly in the early universe than today ) . as @xmath257 the @xmath267 constant surfaces approach the null cone
so the example is an illustration of the foliating ( 3 , 1 ) hypersurfaces becoming tangent to the characteristics at the point of interest .
also the foliation breaks down as @xmath257 because the family of hypersurfaces of constant @xmath23 intersect at @xmath263 .
all these points are illustrated in figure 6 b ) .
finally , the example can not be taken to be typical by the ( 4 , 1 ) version of genericity .
indeed , we know that not many ( 3 , 1 ) spacetimes can be even locally embedded into ( 4 , 1 ) minkowski @xcite .
the value of particular examples is limited to exhibiting _ possible behaviours .
one would require this to be upgraded to the study of large classes of examples to assess _ probable behaviours .
indeed , one would expect higher - dimensional cosmological models to exhibit a range of behaviours distinct from that of the big bang ( including nonsingularness ) .
this follows from knowledge of particular examples ( from the algebraic study ) of inhomogeneous cosmologies in standard gr @xcite .
we especially note that if highly - symmetric cosmologies are haphazardly embedded , one would often expect the resulting higher - dimensional models to be less symmetric i.e more anisotropic and inhomogeneous than the original lower - dimensional models [ as naively suggested by the appearance of ( 4 , 0 ; 1 ) shear from models containing no ( 3 , 0 ; 1 ) shear ] .
so the resulting higher - dimensional models are expected occasionally to exhibit more unusual behaviour than the lower - dimensional models from which they arose , and this includes the possibility of nonsingular cosmologies .
indeed , we would expect that an increase of dimension increases the variety of possible behaviours in inhomogeneous cosmology .
but the bottom line in the study of 4-d and 5-d singularities is the same : one is ultimately interested in generic behaviour and this requires more sophisticated means of study than building individual particular examples .
_ _ there are arguments against the use of embedding results toward making general statements about singularity removal .
our secs 2.4 and 3 provide arguments against the use of the cm result ( some of which also hold against the use of further embedding theorems ) .
in particular , given a spacetime , there are so many possible embeddings . were nature higher - dimensional , why should it choose a particular nonsingular embedding out of an infinity of ( nonsingular or singular ) embeddings that mathematically exist ?
the more extra dimensions are present , the more severe this nonuniqueness is .
this makes singularity removal by such embeddings physically - questionable .
the cm scheme , and other embedding schemes such as the 10-d minkowski embedding theorem @xcite ) , are _ local .
this localness includes the notion that the theorem is only applicable to a small region of the original manifold .
we question whether this need include the approach to a singularity , since these are edges of the lower - dimensional spacetime and thus have no neighbourhood .
_ singularities are global features of spacetime .
however even some of the global embedding theorems ( which require very many extra dimensions ) are stated for the analytic functions @xcite .
but low differentiability may be typical in the approach to singularities @xcite .
it becomes much harder to make any general statements once one accepts that spacetime is not analytic ! _ _
one may view at least some ( 3 , 1 ) singularities as projective effects due to taking a badly - behaved foliation ( for example a foliation that becomes tangent to the characteristics ) , manifested by the formation of caustics or pure shear blow - ups . the question is why there is confusion in reversing this projection procedure to embed ( 3 , 1 ) singularities into nonsingular ( 4 , 1 ) spacetimes .
if an embedding of the ( @xmath1 , @xmath2 ; @xmath4 ) type such as those considered in this paper is to be used , it is mandatory that the region of hypersurface to be embedded be entirely of one signature @xmath2 ( the ck theorem demands a _ nowhere - characteristic lower - dimensional manifold ) .
also , from the lower - dimensional perspective , the singularity can not be included in the set of points on which data is prescribed because it is an edge of spacetime , not a point of spacetime .
were one to try and include it by providing data ` right up ' to that edge , one would expect that the data would become badly - behaved e.g some components of @xmath172 could be forced to be infinite . under the various circumstances above , embedding theorems become inapplicable to reconstruct the higher - dimensional spacetime .
_ finally , we have two comments about the application to thin matter sheets .
first , there is less scope for singularity removal along the lines of ( [ singrem ] ) . for ,
as it is the energy - momentum on the brane , @xmath226 is presumably finite for a nonsingular ( 4 , 1 ) world so @xmath172 is finite and so can not cause blowups in @xmath268 .
however , blowups in this quantity can still occur if the inverse metric is badly - behaved ( corresponding to @xmath173 = 0 ) .
note however that the application of the ck theorem to gr requires this not to be the case everywhere within the region of applicability .
second , one would of course require the investigation of more elaborate ( e.g ` warped ' ) embeddings than the example discussed above in order to investigate whether some embeddings lead to 5-d geodesics that exit the brane .
if the ( 3 , 1)-dimensional geodesics on the brane are not included among those of the ( 4 , 1 ) dimensional bulk , it is not clear at all what is meant by ` singular ' since one then one would have to simultaneously consider some notion of extendibility two congruences of privileged curves of different dimensionality . under the decomposition ( [ adms ] )
, the 5-d geodesic equation @xmath269 becomes ^ + ^_^^ + 2k^ _ = 0 , [ geo1 ] ^ + k_^^ = 0 .
[ geo2 ] if one upholds the use of normal coordinates .
in order for ( [ geo1 ] ) to reduce in all cases to the ( 3 , 1 ) geodesic equation on the brane , one requires everywhere on the brane @xmath270 or @xmath271 . by ( [ geo2 ] ) , maintenance of @xmath270 along all geodesics
is impossible unless @xmath272 .
thus in all cases @xmath272 is required on the brane . but
this means that @xmath273 or equivalently @xmath274 i.e the absence of a brane .
thus in sms - type braneworlds the ( 3 , 1 ) geodesics are not generally among the ( 4 , 1 ) geodesics .
so if one wishes to adhere to the postulate that bulk - traversing fields follow ( 4 , 1 ) geodesics and fields confined to the brane follow ( 3 , 1 ) geodesics , one should accept that the restriction of the former to the brane generally does not coincide with the latter .
thus there may be observable consequences in the ( 3 , 1 ) physics : freely - falling bodies composed of confined matter would follow distinct paths from freely - falling bodies composed of matter capable of traversing the bulk but which happens to be moving along within the brane , which represents a violation of the equivalence principle , and the causal structures within the brane due to confined photons and due to gravitons which happen to be moving along within the brane would generally be distinct . also the above new conceptual difficulty about the definition of singularities
is thus indeed relevant to brane cosmology .
we are interested in the study of ( @xmath10 + 1)-dimensional gr models with thin matter sheets using only well studied , well behaved mathematical techniques which also make good physical sense and are applicable to full efe systems . in our view a good way of achieving this is via the ( @xmath10 , 0 ) ivp , followed where possible by the heavily - protected ( @xmath10 , 0 ; 1 ) cp .
we provide a hierarchy of modelling assumptions together with associated sources of difficulty , and argue that evolution is more problematic than data construction .
we concentrate on the latter , making use of all the attractive features of the york method mentioned in sec 3 , albeit this elliptic mathematics is in a new setting ( see sec .
6.2 - 3 ) .
the differences come about due to the following difficulties of relevance to the hierarchy of modelling assumptions .
firstly , the roughness of function spaces required in the rigorous treatment of certain thin matter sheet models . secondly , the requirement of novel asymptotics and boundary conditions ( b.cs ) . note that the increase in dimensionality itself does not affect much the data construction nor the study of well - posedness and stability of the gr cp , although there is some dimensional - dependence when the rougher function spaces come into use , due to the following two contrasting results .
the first result is that in all dimensions , the presence of thin matter sheets means that the metric is rougher than @xmath275 ( the functions with continuous second derivatives ) because of the jump in the riemann tensor at the sheet .
this corresponds to the metric being too rough to belong to the sobolev class @xmath276 .
the second result is that the definition of a product for sobolev spaces for the efe s in harmonic coordinates according to hughes , kato and marsden ( hkm ) @xcite requires the sobolev class of the metric to be no rougher than @xmath277 [ i.e @xmath278 for standard gr and @xmath279 for ( 4 , 1)-dimensional gr ] .
thus hkm s mathematics , the strongest used for both the gr ivp and cp in the review @xcite , is not generally powerful enough to deal with thin matter sheets .
more powerful mathematics is thus desirable , and more may well appear over the next few years ( see the program starting with @xcite ) .
realizations of part of the hierarchy of models below would be an interesting application for this sort of mathematics , which further motivates the study of increasingly rough sobolev spaces , since the higher the dimension the greater the improvement to the established hkm mathematics is required for a full study of thin matter sheets .
we must also mention that in spaces rougher than @xmath276 there is no guarantee of geodesic uniqueness @xcite , so complete conventional physical sensibleness is generally lost no matter what rougher sobolev class results are proved .
this shortcoming is clearly dimension - independent ( from the dimension - independence of the form of the geodesic equation ) .
at least one need not go below @xmath280 whereupon geodesics need not even exist ! the asymptotics problem @xcite follows from supposing that ` ads ' bulks are desirable for string - inspired scenarios .
the study of ` ads ' bulks by relativists would entail the study of asymptotically ads bulks , to permit a more general study of disturbances close to the thin matter sheet .
the following comments are in order here .
first , in contrast to ds , ads is infinite and hence possesses asymptotics .
second , whereas we will for the moment seek to avoid the function space difficulty , the asymptotics problem will remain in the ` tractable scenarios ' . in the usual gr ,
the use of ads asymptotics for the application of the york method to small - scale astrophysics was neglected because of the negligible effect of a cosmological constant on such small scales and because of dod arguments .
the braneworld application is however substantially different , so one might in this case have to study the york method with ` ads asymptotics ' .
this could affect the tractability of the lichnerowicz equation , and also whether the existence of the crucial maximal or cmc slices is as commonplace for these new spacetimes as it was for the ( 3 , 1 ) gr ones .
furthermore , since one is interested in infinite planar branes , the asymptotics is _ directional , ` perpendicular ' to the brane .
however , in general there is no such notion of perpendicularity in gr . whereas one can locally define geodesics and draw hypersurfaces perpendicular to them , this procedure in general eventually breaks down , for example due to ( spatial ) caustic formation . _
we start our hierarchy by envisaging the most general situation possible for thin matter sheets within the framework of gr .
it is entirely legitimate to construct a ( 4 , 0 ) initial data snapshot with whatever shape of thin matter sheet , but it is not legitimate to assume that any features of this are maintained over time unless it can be shown that the full evolution equations robustly maintain these features .
strictly , this is an _ initial - boundary problem , such as occurs for water waves or for the surface of a star .
this is a very hard and quite new problem in gr @xcite .
thus the evolution step is particularly hard both in its full generality and in the justifiability of simplifications such as a fixed boundary . _
consider a thin matter sheet in an asymmetric ( not @xmath146 symmetric ) bulk .
one would ideally want to follow objects that crash into and possibly disrupt the thin matter sheet .
one can then imagine asymmetric crashes which might punch though the thin matter sheet .
the sheet could thicken or disperse with time .
it could emit a significant amount of gravitational radiation , which could moreover include gravitational shock waves that spread out the @xmath280 character that initially pertained only to the thin matter sheet and not to the smooth surrounding bulk .
that is , the @xmath280 character of a thin matter sheet data set @xmath281 could typically infect the whole of its causal future @xmath282 ( fig .
7 ) . in simple words
, why should the junction remain unscathed ?
why should the bulk in the immediate vicinity of the junction at later times be simple , smooth or known ?
addressing these questions goes beyond the reach of present - day techniques .
rather we note that some usually - tacit assumptions are used to definitely avoid these difficulties , and that we do not know whether any of these are justifiable in the physics governed by the gr evolution equations .
one can presuppose a thin matter sheet exists at all times to prevent it being created or destroyed .
one can allow only bulks with regular thin matter sheet boundaries to preclude shock waves .
one could also preclude asymmetric crashes by presupposing @xmath146 symmetry .
even if discontinuous emissions are precluded , one can imagine smooth emissions and absorptions ( symmetric or asymmetric ) whereby the bulk interacts with the thin matter sheet .
this would entail material leaking off or onto each side of the thin matter sheet .
this can be precluded by the presupposition that the thin matter sheet s energy - momentum resides at all times on the sheet , which is encapsulated by the well - known ` equation of motion ' of the thin matter sheet @xcite .
this is a significant restriction on the dynamics of thin matter sheets in gr , often carried out in the name of tractability but which may not be realistic . in the randall sundrum scenario ,
the brane can be placed anywhere in the bulk without affecting it .
we strongly suspect that this is not a typical feature in gr - based braneworlds , nor is it desirable since it dangerously marginalizes the ontological status of the bulk .
we take this feature to be too specialized to be included in the developments below .
we close this section by briefly discussing modelling assumptions which avoid some of the above difficulties .
we then implement these in two particular classes of tractable problems .
the desirability of each modelling assumption strategy and of other modelling assumptions from a string - theoretic perspective is discussed in sec 7 .
\i ) the choice of @xmath283 symmetry may well look and indeed be arbitrary from a gr point of view and indeed from a string - theoretic point of view ( see sec 7 ) .
but without such a choice one simply can not establish any specific embeddings if there are thin matter sheets ( in the gr study of stars one requires the absence of surface layers to perform matchings ) .
thus one studies the restricted case with @xmath283 symmetry , in which one integrates only up to the b.c provided by the j.c .
in the ivp part one can assume whatever configuration of thin matter sheets .
thus this step is more justifiable than the subsequent evolution , which may require some of the above _ ad hoc assumptions as to the existence and good behaviour of the thin matter sheet at all future times .
so for the moment we are merely after the construction of ( 4 , 0 ) initial data up to the junction . by themselves
these data are useful in addressing issues such as the shape of the extensions of black hole horizons into the bulk , but the staticity and stability of these configurations remain unaddressed in the as - yet intractable evolutionary step of the problem @xcite . _ \ii ) one could choose instead to work with thick matter sheets i.e ones that are finitely rather than infinitely thin @xcite .
then the above troubles with the function spaces need not occur , since the walls would then not be rough in the above sense , although they could still have the tractable level of roughness that is able to accommodate astrophysical objects just as in ordinary gr .
it is then easier to envisage the study of the evolutionary step of the thick matter sheet problem quite high up our hierarchy , since this more closely resembles the stellar surface problem .
\iii ) one could investigate closed shells rather than open sheets in order not to require directional asymptotics .
we cast the initial value constraints as - ( ) = 0 , = - c(r - m^c + ^2^b - 2^a ) , m = ^ ^ [ nonlinlich ] d_a^ab = j^b b = , c = , a = \ { ll & j^b = 0 + & j^b 0 . + .
we first consider the @xmath284 case
. then by definition @xmath285 suffices in order to solve the codazzi constraint .
furthermore , in that case a particular scaling of @xmath286 is not required to keep the codazzi constraint conformally - invariant , so preserving the dec does not enforce a particular scaling of @xmath34 .
so we are left to solve just the lichnerowicz equation with the scaling of @xmath34 to be determined by our convenience .
we begin by considering the linearized lichnerowicz equation .
set @xmath287 , @xmath288 in ( [ nonlinlich ] ) and equate the @xmath289 terms to obtain [ - ( _ 0 ) ] _ 1 = ( _ 0 ) , ( _ 0 ) = -c[r -(1 + c)m^c_0 + ^2(1 + b)^b_0 - 2(1 + a)_0^a ] , ( _ 0 ) = 2c_1 ^ 1 + a_0 .
[ linlich ] if asymptotically @xmath290 and @xmath291 one obtains @xmath292\phi_1 = \mbox{inh}(\phi_0^{\mbox{\scriptsize a\normalsize}})$ ] from which the asymptotic behaviour of @xmath116 for the full equation may be obtained .
( [ linlich ] ) is a linear elliptical equation _ 1 = ( _ 0 ) [ linell ] ( where @xmath293 ) .
many b.v.ps for such equations are well - studied @xcite . to give an idea of the techniques available , consider the dirichlet problem : ( [ linell ] ) in @xmath236 together with b.c @xmath294 on @xmath295 . on sufficiently - benign function spaces ,
the well - known uniqueness proof based on green s identity holds , provided that lin @xmath70 everywhere .
this uniqueness can also be proved from the _ maximum principle , the use of which applies to more general function spaces .
one then proves existence from uniqueness @xcite , by setting up a sequence of problems interpolating between the flat - space laplace equation and ( [ linell ] ) .
one can then interpolate between quasilinear elliptic equations such as the full lichnerowicz equation and their linearizations .
then one has topological theorems about the fixed points of suitable maps , such as leray
schauder degree theory @xcite , to prove existence of the solutions to the quasilinear elliptic equation s dirichlet problem .
similar techniques are available for further b.v.ps @xcite . _
thus the chances of having a well behaved b.v.p for the lichnerowicz equation are much enhanced if the corresponding lin @xmath70 . to this
end , notice that @xmath296 and @xmath297 are always positive whereas @xmath265 is of variable sign .
contrary to what york assumes in the ordinary gr context , we must permit @xmath298 since our application requires a negative bulk cosmological constant , which further complicates the analysis of which cases are guaranteed to be well - behaved .
there are three further simplifications one can try : \i ) conformal flatness , which replaces @xmath299 by the flat - space @xmath300 and wipes out the arbitrary - sign @xmath265 term .
\ii ) maximality : @xmath301 removes the highest - order term of nonlin(@xmath116 ) .
\iii ) setting @xmath302 .
if this is combined with ii ) , one has time symmetry @xmath303 , which is regarded as overly simple in the standard applications of numerical relativity as it corresponds to the absence of gravitational momentum on the slice @xmath236 .
one can sometimes also use a trick : to scale @xmath34 like @xmath296 , @xmath265 or @xmath297 and then fix it so as to cancel with this term . because we are interested in @xmath298 , only the first two of these possibilities are available . in the @xmath304 case ,
the cancellation trick above is not available if one wishes the procedure to respect the dec .
simplification i ) is convenient in this case .
whereas it is important to develop this case , it lies beyond the scope of this paper .
we work with the unphysical line element @xmath305 for some known trial function @xmath306 , so as to model a @xmath307-symmetric object such as a black hole or a star on the brane .
for the @xmath308 case , the ivp essentially reduces to the solution of the lichnerowicz equation for some @xmath309 .
we begin with the @xmath310 case of ( the spatial projection of ) ( [ bwem ] ) .
we first find the b.cs for the lichnerowicz equation , then provide algorithms and then comment on the underlying analysis .
note that the b.cs are to be imposed on the physical metric and then written in terms of the unphysical quantities to have b.cs in terms of the objects we start off with and the unknown conformal factor @xmath116 .
we begin with the inner conditions and then discuss asymptotic outer conditions . to establish this sort of inner radial b.c , it is commonplace to use an isometry @xcite .
for example , one could use @xmath311 @xmath312 @xmath313 @xmath312 @xmath314 leading component - by - component to the homogeneous neumann b.c ( the so - called reflection b.c ) .
|_r = 0 = 0 , [ irbc ] and the restriction @xmath315 on the valid form of the known function .
one might worry in the case of black holes that @xmath316 is singular .
for ordinary gr , one excises the singularity by use of an inversion - in - the - sphere isometry about some throat at radius @xmath317 of the black hole within the apparent horizon ( see fig 8a ) .
this gives an inner robin condition @xcite .|_r = a = 0 .
for braneworld black holes one does not know how far the black hole extends into the bulk .
the main point of the study is to find this out ( pancakes versus cigars ) .
our idea is then to guess a profile @xmath318 along which to excise .
about each point on @xmath318 an isometry in the corresponding 2-sphere may be applied ( fig 8b ) , leading to an inner robin condition .|_r
= f(z ) = 0 .
the profile could be chosen so that it matches up smoothly with the @xmath316 and @xmath249 boundaries .
once the problem is solved , one can find out whether this profile was a good choice or not ( see fig 8c ) . by picking a 1-parameter family of @xmath318 curves
, one could then iterate until a satisfactory profile is found . for the inner-@xmath23 ( @xmath249 ) b.c
, we impose the j.cs ( [ jcf ] ) and ( [ prez2 ] ) , which now read _ ^+ = _ ^- , _ ^+ - _ ^- = -_5 ^ 2 ( _ - _ ) [ scndjn ] ( n.b these are imposed on the physical quantities ) . imposing @xmath146 symmetry , the second of these becomes . | _ z = 1 = 0 [ nonlinbc ] by use of the definition of the physical @xmath319 in normal coordinates and then factorizing out the @xmath320 .
this @xmath146 isometry b.c is more complicated than usual because of the presence of the thin matter sheet . that the @xmath146 symmetry is termed a reflection
appears to generate a certain amount of confusion .
this is because there is also the reflected wave notion of reflection , taken to give a neumann b.c ( such as in the @xmath316 condition above ) . in the case of shiromizu and shibata ( ss ) @xcite
( see also @xcite ) , who furthermore treat our known @xmath74 as their unknown , and our unknown @xmath116 as the perfect ads @xmath321 , it has the nice feature that the other 2 terms in ( [ nonlinbc ] ) cancel leaving a neumann b.c .
|_z=1 = 0 , [ funnyref ] which might look like a reflection in both of the above senses .
however , this is only possible for @xmath322 which may be somewhat restrictive .
note also that ( [ funnyref ] ) is an unusual notion of reflection in this context , since it is of the conformally - untransformed ( 3 , 0)-metric , whereas the natural notion of reflection isometry would be for the conformally - transformed ( 4 , 0)-metric .
thus we conclude that the presence of the notion of reflection ( [ funnyref ] ) in ss s work is happenstance : it is separate from the @xmath146 notion of reflection encapsulated in ( [ scndjn ] ) and no analogue of it need exist in more general situations .
our b.c . ( [ nonlinbc ] ) contains a reflection part @xmath323 ; the other part being a ( nonlinear ) absorption .
this has the following implications .
first , the presence of absorption terms is interesting since pure reflection is the boundary condition of a perfect insulator in the potential theory of heat .
this suggests that pure reflection b.cs such as for pure ads have built - in non - interaction of the bulk with its bounding brane , whereas our b.c may lead to ( evolution models ) with brane - bulk interactions .
whereas neither the planar symmetry in ( 4 , 1 ) spacetime of the underlying thin matter sheet nor the @xmath307 symmetry within the ( 3 , 1 ) thin matter sheet of a simple astrophysical object generate gravity waves , we expect that spherical bumps on approximately planar branes in ( 4 , 1 ) spacetimes to admit gravity waves , potentially giving rise to instabilities in the approximately ads bulk braneworlds in which the branes contain astrophysical objects
. our b.c could permit the modelling of such interactions .
second the nonlinear absorption ( much as arises in the theory of heat with temperature - dependent conductivity ) is likely to complicate both analytic and numerical treatments of our b.v.p .
the first point above is one reason to favour our approach over that of ss ( subject to overcoming the complications due to the second point ) .
we now provide a further reason why the approach we present should be favoured on the long run . whereas one could try to generalize ss s work by keeping the notion that @xmath74 be unknown and @xmath116 known but not @xmath324 , whereupon our b.c ( [ nonlinbc ] ) is interpreted as an inhomogeneous neumann b.c in @xmath74 ,
the main trouble with ss s work is that there is no good reason for their method to be directly generalizable away from its restrictive assumption @xmath303 . whereas treating @xmath74 not @xmath116
as the unknown may give nicer b.cs , it is not tied to a method known to be amenable to less trivial solutions of the codazzi equation than @xmath325 .
our proposed method seeks a workable extension to the significantly more general case @xmath326 , and speculatively to completely general @xmath327 .
another nice feature in ss s work is that the matter is scaled so that the linearized p.d.e obtained is a poisson equation .
this p.d.e is simply invertible to obtain detailed asymptotics @xcite . for our scheme ,
the linearized equation ( [ linlich ] ) is of helmholtz - type rather than poisson , complicating such a procedure .
we merely demand instead that @xmath328 for @xmath23 large wherever possible so that our models are ` directionally asymptotically ads ' . as for large @xmath1 , at least on the brane , one can impose asymptotic flatness as @xmath329 .
it is less clear what one should prescribe in this respect off the brane .
a more detailed study of the asymptotics for our b.v.ps should be required as part of the actual construction of examples of data sets .
our proposed b.v.p is thus the mixed , nonlinear problem \ {
l = ( ) , + + ( - ) = 0 z = 1 r f(1 ) , z , + . .
= 0 r = 0 z f^-1(0 ) , 1 r + + = 0 r = f(z ) . .
one can use simplifications i ) and/or ii ) on this and still have a more general case than ss .
simplification i ) might be plausible because 5-d ads clearly admits conformally - flat spatial sections [ @xmath330 ) ] or maybe not , since it is regarded with suspicion in the ( 3 , 1 ) gr 2-body problem @xcite .
the trick at the end of sec 6.2 can not be used to obtain particularly simple examples due to the following argument . unless one puts @xmath331 ( constant ) and
has this maintained by non - scaling ( i.e @xmath332 ) , it is overwhelmingly probable that the emergent @xmath333 is not constant .
thus our freedom in @xmath317 is used up to ensure that @xmath333 is constant .
our algorithm for solving this problem is as follows : \i ) prescribe the following unphysical quantities : the metric @xmath76 and matter density @xmath34 .
\ii ) pick any suitable @xmath334 to solve the codazzi constraint .
it follows that @xmath296 is known .
\iii ) thus we can attempt to solve our b.v.p for the lichnerowicz equation to obtain @xmath116 .
\iv ) then we can compute the physical bulk metric @xmath335 and induced brane metric @xmath336 of our snapshot .
this assumes that we have correctly guessed the profile .
one would now check whether this is the case by solving for the apparent horizon .
if this is unsatisfactory ( fig 8c ) then one would repeat the algorithm with adjusted profile . for nonvacuum branes ( @xmath337 ) , a similar argument enforcing @xmath332 holds .
the j.c may now be split into a trace b.c part , . | _ z = 1 = 0 and a restriction on the tracefree part of the matter on the brane , which is y_^ = 0 [ 95 ] for the metric ansatz and coordinate choice used .
our algorithm now becomes \i ) one now requires the prescription of @xmath338 everywhere and of @xmath339 .
\ii ) pick any suitable @xmath334 and hence @xmath296 .
iii)attempt to solve our b.v.p for the lichnerowicz equation .
\iv ) then we can compute @xmath335 and @xmath340 . in this particular case , the simultaneous imposition of normal coordinates and an isotropic line element forces the restriction ( [ 95 ] ) on the braneworld matter .
this is because one is applying too many coordinate restrictions ; in this light the habitual practice of working in braneworlds using normal coordinates may be seen as a poor choice of gauge .
if the above coordinate choices are not simultaneously made , the algorithm would contain further nontrivial equations [ in place of ( [ 95 ] ) ] to solve before the braneworld matter content can be deduced .
we assume that this nonvacuum application is for a star or clump of dust on the brane in which case there is no need to excise a corner with a profile as was done above to deal with the black hole singularity .
notice the lack of control of the physical metrics characteristic of theoretical numerical relativity .
our non - scaling of the matter at least gives visible control over the matter .
the sensibleness of doing this is in fact tied to the negativity of the bulk @xmath34 : for @xmath341 and @xmath342 , lin receives a negative contribution with its tendency to encourage ill - posedness , whereas for predominantly @xmath343 , the danger @xcite lies in @xmath344 .
a simpler example of b.v.p set up along the lines we suggest is that of nakamura , nakao and mishima @xcite .
their metric is cylindrical not spherical and they have a slightly simpler version of ( [ nonlinbc ] ) .
their case is still time - symmetric .
the linearized equation is then precisely helmholtz , facilitating its inversion and the consequent more detailed knowledge of the asymptotics .
we also propose the 2-brane version of the above , in which the @xmath345 condition is replaced by another ` parallel ' brane boundary at @xmath346 .
this problem is close to that considered by piran and sorkin @xcite .
in contrast however , their study involves a periodically - identified fifth dimension to which there does correspond an isometry - based reflection neumann condition @xmath347 .
we can provide a local ( in function space ) uniqueness proof for our proposed type of b.v.p , at the same level as that for standard gr black hole data in @xcite . in the conformally - flat case ,
suppose that @xmath348 for @xmath80 small .
then the homogeneous linearized b.v.p in @xmath80 is applicable .
we then have _ ( |u|^2 + u^2)d^4x = _ ( |u|^2 + uu)d^4x = _ uds
= l _ z = z uds + _ r = r uds + - _ z = 1 + uds - _ r = uds by green s theorem . for @xmath349
at least as fast as @xmath324 , the first surface integral tends to zero . for @xmath349
at least as fast as @xmath350 the second surface integral tends to zero .
the fourth integral is zero by the neumann reflection b.c . upon applying the linearization of the nonlinear b.c ( [ nonlinbc ] ) , .
z=1 = 0 , [ linrob ] the third surface integral is non - positive provided that the restriction @xmath351 holds .
thus since lin @xmath70 u must be zero .
for the non - conformally flat case , this argument requires @xmath352 .
the argument is not seriously changed in the black hole case when a corner is excised by a profile with a robin condition on it . stronger existence and uniqueness proofs for our proposed problems
are complicated by four factors : unboundedness , ( albeit of the most benign kind ) , the boundary having corners , mixed b.cs ( i.e a piecewise prescription on the boundary ) and a portion of these being nonlinear .
the first is however of the simplest kind and the second and third are commonplace .
the last is less usual but perhaps not so bad either because by @xmath353 the problem can be written in divergence form , for which b.v.ps with nonlinear b.cs are treated in ch .
10.2 of @xcite .
the trouble is that the treatment there , unlike here , is neither for mixed b.cs nor for an unbounded region with corners .
thus , the strongest point we can make at present is that there is good hope of obtaining existence and uniqueness results even for quite rough function spaces by more - or - less conventional , entirely rigorous mathematics for our proposed problem .
this should be contrasted with the hopeless state of affairs with sideways prescriptions !
once one begins to get good numerical results , it becomes worthwhile to explicitly prove the well - posedness of the method used , in order to support those results and the ongoing production of more such results . in this particular case
, these results would serve to support and understand the crucial numerical step iii ) in the above algorithms . to permit the even greater generality required to have momentum flows ( @xmath304 ) , by our route one
is forced to forfeit the dec control over the matter .
this is because at least the @xmath145 part of @xmath34 can not scale in accord with the scaling of @xmath354 .
we leave this further development , which additionally requires posing and solving the b.v.p following from the codazzi equation , for a future study .
many of the difficulties with evolution discussed above stem from the thinness of the matter sheets causing function - space - related problems .
but ordinary physics should not be sensitive to which function space is used .
one would hope that models with thick matter sheets would be more amenable to study , be good approximations to the thin matter sheets and in any case could be closer to reality than thin matter sheets .
bonjour , charmousis and gregory have used the ( 2 , 1 ; 1 ) version of the general split ( [ gauss ] ) , ( [ cod ] ) , ( [ evk ] ) with scalar matter and consider both thick walls and the thin - wall limit perturbatively in the presence of gravity .
they further specialize to the case of a spherical domain wall , which they show collapses
. it would be worthwhile if this sort of example , which combines the modelling assumptions ii ) and iii ) , is investigated using the ( @xmath10 , 0 ; 1 ) split subject to flat and to ads asymptotics . for the particular example mentioned above
, the matter profile is sigmoidal .
however , to approximate a thin brane , one would want instead a hump profile , as indicated in fig 9a .
one could still use an inner b.c at @xmath249 , where now @xmath355 so that the linear robin b.c . | _ z = 1 = 0 [ thickrob ] holds .
one could in fact consider 2 problems ( see fig 9b ) : the flat sheet with directional asymptotics and the spherical sheet with a suitable inversion - in - the - sphere in place of the above inner b.c .
both of these problems avoid the main obstacles to existence and uniqueness proofs by possessing linear b.cs .
the second problem is also invertible to an inner problem which is on a bounded region with no corners , for which results in @xcite apply .
these problems are for ` plain ' thick branes as opposed to branes containing spherically - symmetric objects .
the first problem ought to more easily admit this extension .
one might wish to study ` plain ' thin branes too , e.g in the context of branes in relative motion .
finally , suppose instead one attempted to use magaard s method to construct ( 4 , 0 ) thin - matter - sheet data .
then one benefits from @xmath7 and the identification of @xmath356 with the @xmath249 brane prescribes the topology , removes some of the sources of nonuniqueness and the brane s energy - momentum endows physical significance upon the ( 3 , 0 ) covariance of the elimination procedure . in the thin matter sheet case , one is blocked by the nonanalyticity of @xmath34 , but at least some thick matter sheet models could be built in this way .
we first address * q2 by discussing how the work in this paper may be generalized and which of these generalizations may be regarded as stringy .
we end with the conclusions of this paper as regards * q1 . *
* we did not consider the possibility of a higher codimension @xcite which could substantially alter the nature of the embeddings used . for models with 2 or more times ,
it is simply not possible to stick entirely to our suggestion to build on established mathematical physics .
finally , our study could be complicated by having more than the einstein tensor in the bulk .
for example , one can have a nontrivial lovelock tensor term @xcite in dimension 5 , which arises from the variation of a gauss bonnet term in the action , or one could have genuine higher - derivative terms @xcite . whilst our incipient results such as the correspondence between ( @xmath10 , 0 ; 1 ) and ( @xmath10 - 1 , 1 ; 1 ) schemes at the simplest level or our befe ambiguity will have counterparts in these theories , these theories equations are much larger than the efe s , and their cp / ivp is much less well - studied than that of gr .
higher derivatives or non - standard bulk matter might lead to the violation of the energy conditions used in this paper .
we next discuss which features of gr - based braneworld models are desirable for string theory .
first for some caveats .
we introduced the sms braneworld as a gr generalization of the randall sundrum scenario , moreover one that was as yet not sufficiently general for the purposes of gr in which the nature and existence of thin matter sheets at all times might not be a representative presupposition .
it could be that it is other aspects of the randall sundrum scenario that are of interest in string theory since it also has particle physics aspects to it and because it is a toy model of hoava
witten theory .
one may prefer to consider only such stringy scenarios rather than randall sundrum or sms .
also a great source of tension between string theorists and relativists is that the latter believe gr is suggestive of the importance of background
independent theories , whilst there is as yet no background independent formulation of string theory available .
such ideas as conserved charges , flat sheets and simple bulks arise from assuming minkowski or ads backgrounds .
since gr is background - independent , one is interested there in generic solutions to the full field equations .
this makes the notion of asymptotics necessary and subtle , and not necessarily compatible with flat sheets nor conserved charges . from a general perspective
it is doubtful whether highly - specialized solutions such as minkowski or ads are likely to reflect the gravitational physics of the universe , unless it can be shown that large classes of generic solutions behave likewise or are attracted toward such solutions .
string theory might favour many extra dimensions , although it is not clear how many of these should be large and how many compactified . among the corrections to the einstein tensor mentioned above , string theorists might favour the presence of the gauss bonnet correction , since this is the only first - order correction for a heterotic string @xcite ; an infinite series of correction terms is predicted .
string theorists might favour very flat walls on the grounds that these could be stabilized by being the lowest - energy carriers of bps charge @xcite .
these arguments would rule against the use of highly symmetric bulk manifolds / orbifolds , although string theorists might try to invoke them as fixed backgrounds so as to perturb about them . there may be different reasons for stability in gr and string theory .
for example , in contrast to string - theoretic arguments favouring reflection - symmetric orbifolds , a gr domain wall could be stable precisely because the ( 3 , 1 ) bulk is different on each side , as the wall separates domains in which symmetry is differently broken . on the whole
, the string theorist would agree that bulks could be complicated by the presence of bulk matter from the closed string multiplet , as mentioned in sec 4.5 .
at least in some models , gauge fields would be expected to occur only as fields confined to branes since they belong to the open string multiplet and some branes are where open strings end @xcite . as strictly , the branes are of planckian thickness , so it is not a disaster that thick branes are favourable toward rigorous mathematics for gr - based scenarios ( this is recycling the argument in footnote 11 ) . of course , one could worry that they are thin enough for quantum gravitational effects to be important .
we now address * q1 . we feel that it is secure to say that gr - based braneworld models only make sense if realistic examples of them are stable to the dynamics which governs gr .
we have investigated two schemes within which this study might be furthered : the ( 3 , 1 ; 1 ) and ( 4 , 0 ; 1 ) schemes . * as regards ( 3 , 1 ; 1 ) schemes , before pressing on with something like the sms braneworld by perturbation theory and dynamical systems analysis , we think that it is important to check that the sms braneworld s interpretation is properly understood .
we have demonstrated that there exist many reformulations of sms s befe s , thus making it clear that it is a choice as to whether the befe s contain sms s weyl term ( requiring closure as some complicated third - order system such as in sms s paper ) , and that it is a separate choice as to whether any terms homogeneously quadratic in the thin matter sheet energy - momentum tensor feature in the befe s .
we have used this to illustrate the danger in truncating the befe s by setting bulk terms to be zero or to take a particular form . the full ( 4 , 1)-dimensional picture of the brane is required . building this picture by sideways evolution appears unlikely to be well - behaved mathematically or to make sense from a physical ( causal ) point of view .
we hope that this paper will either stimulate new research where these concepts are carefully thought about or force an admission that it is preferable to use a ( 4 , 0 ; 1 ) scheme .
we have argued that the removal of singularities by the ( 3 , 1 ; 1 ) scheme is unlikely to be generally possible in a meaningful way . whichever of the two schemes one works in , one is well - advised to heed the old , careful literature about solving the gauss codazzi system rather than starting afresh .
we have carefully detailed which parts of this traditionally ( @xmath10 , 0 ) scheme easily pass over to the ( @xmath1 , 1 ) case .
as regards the ( 4 , 0 ; 1 ) scheme that we have argued is favourable , we have argued that its ivp step is cleaner than its cp step and by itself serves to answer a number of questions .
we have provided a survey of obstacles and ideas to extend this ivp with a thin matter sheet away from the relatively trivial time - symmetric cases hitherto considered .
we also suggest that thick branes are more easily treated rigorously and that this choice should not affect the underlying physics if it is respectably robust .
we feel that the next stage should be to solve for such data numerically . as this field grows
, the underlying analytic mathematics will deserve detailed attention .
we note that the scope of the ( 4 , 0 ; 1 ) scheme would go as far as the study of colliding black holes , which are expected to be an important source of gravity waves .
one requires the kind of methods that we promote in this paper for gr - based braneworlds for the full study of such high - curvature processes .
whilst this study is a tall order even for the usual gr @xcite , these calculations may be required if braneworld effects can be large enough to substantially alter the extreme physics of colliding black holes . finally , we summarize the aspects of * q2 that should be asked as the partial answers to * q1 begin to be built up . does the introduction of stringy features change the outcome to any of issues ? and
so , are gr - based braneworlds adequate or typical frameworks for string theory ? what scenarios should one explore in order to furnish string theory with reasonably unique predictions ?
+ * acknowledgements + ea is supported by pparc .
we acknowledge useful discussions on a wide range of topics with julian barbour , henk van elst , james lidsey , roy maartens , malcolm maccallum , marc mars , nikolaos mavromatos , niall murchadha and carlos romero . * * * k. akama , lect . notes phys . * 176 267 ( 1982 ) , hep - th/0001113 ; v.a . rubakov and m.e .
shaposhtikov , phys .
b * 159 22 ( 1985 ) ; n. arkani - hamed , s. dimopoulos and g. dvali , phys . lett .
b * 429 263 ( 1998 ) , hep - ph/9803315 ; m. gogberashvili , europhys .
* 49 396 ( 2000 ) , hep - ph/9812365 . * * * * j. khoury , b.a .
ovrut , n. seiberg , p.j .
steinhardt , n. turok , phys . rev . * d65 086007 ( 2002 ) , hep - th/0108187 ; j. khoury , b.a .
ovrut , p.j .
steinhardt , n. turok , phys . rev . * d66 046005 ( 2002 ) , hep - th/0109050 . * * r. arnowitt , s. deser and c.w .
misner , in _ gravitation : an introduction to current research ed l. witten ( wiley , new york 1962 ) ; p.a.m .
dirac , _ lectures on quantum mechanics ( yeshiva university , new york 1964 ) . _ _ k. nakao , k. nakamura and t. mishima , preprint gr - qc/0112067 ; k. nakamura , k. nakao and t. mishima , progress of theoretical physics supplement * 148 _ brane world : new perspective in cosmology , gr - qc/0302058 .
_ * j.a wheeler , in _ groups , relativity and topology , ed .
dewitt and c.m .
dewitt ( gordon and breach , new york 1963 ) ; j.a .
wheeler , in _ battelle rencontres ed . c.m
dewitt and j.a .
wheeler ( benjamin , new york 1968 ) . _ _ the result is in m.d .
stern s princeton a.b .
senior thesis , may 1967 ( unpublished ) ; the result is reported in the second reference in @xcite and in a.e .
fischer , in _ relativity , ed .
m. carmeli , s.i .
fickler and l. witten ( plenum press , new york london 1970 ) .
_ e.j .
copeland , a.r .
liddle and j.e .
lidsey , phys .
* d64 023509 ( 2001 ) , astro - ph/0006421 ; v. sahni , m. sami and t. souradeep , phys . rev .
* d65 023518 ( 2002 ) , gr - qc/0105121 ; s. kanno and j. soda , hep - th/0303203 . * * a. campos and c.f .
sopuerta , phys . rev . * d63 104012 ( 2001 ) , hep - th/0101060 ; r. maartens , v. sahni and t.d .
saini , phys . rev . *
d63 063509 ( 2001 ) , gr - qc/0011105 ; r.j .
van den hoogen and j. iba@xmath357ez , preprint gr - qc/0212095 ; r.j .
van den hoogen , a.a .
coley and y. he , preprint gr - qc/0212094 . *
* the lovelock tensor is discussed in d. lovelock , j. math .
* 12 498 ( 1971 ) ; for the ( 3 , 1 ; 1 ) application to braneworlds , see for example c. charmousis , j - f .
dufaux , class .
quantum grav .
* 19 4671 ( 2002 ) , hep - th/0202107 .
for an incipient ( n , 0 ; 1 ) study of the system , see c. teitelboim and j. zanelli , class .
quantum grav .
* 4 l125 ( 1987 ) . * * * \a ) for unspecified or general signature , we use the following notation .
let @xmath358 be a ( @xmath10 + 1)-dimensional spacetime .
let @xmath359 be an embedded spacelike hypersurface at parameter value @xmath360 , with unit normal @xmath361 .
the induced metric on @xmath236 is @xmath362 .
consider another nearby spacelike hypersurface @xmath363 at parameter value @xmath364 .
then nearby slices fit together in accordance with the figure , where the lapse is @xmath365 and the shift is @xmath366 . in the particular case of a ( @xmath10 , 0;1 ) split
, we denote @xmath28 by @xmath236 , @xmath361 by @xmath53 and @xmath33 by @xmath255 whilst for a ( @xmath10 - 1 , 1 ; 1 ) split we denote we denote @xmath28 by @xmath367 , @xmath361 by @xmath75 and @xmath33 by @xmath23 .
\b ) when we simultaneously talk of several splits in this paper , we use the notation presented in this figure .
the white arrows represent the extrinsic curvatures associated with each of the embeddings .
a general means of study of hyperbolic p.d.es is to require that the data be prescribed on spacelike surfaces [ ( @xmath10 , 0 ; 1 ) scheme ] a ) , not on timelike surfaces [ ( @xmath1 , 1 , 1 ) scheme ] .
the line @xmath281 in a ) represents the inclusion of a snapshot of a thin matter sheet , whilst the whole of the surface in b ) could be the worldsheet of an arbitrarily thin matter sheet .
* figure 3 ( domain of dependence property ) we should not claim to know too great a portion of the future .
a ) given data for a hyperbolic system on a piece of a spacelike surface , we can predict the future only in the forward wedge ( domain of dependence ) within which all causal curves , i.e allowed paths of information flow , emanate from the data . outside this wedge , for all we know , large disturbances ( represented in the picture by gravity waves ) could influence the future arbitrarily soon .
* \b ) if we claim to know the 4-dimensional dod , we preclude the influence by the extra dimension , thus making it largely redundant .
the 5-dimensional dod of a thin sheet is small because of the thinness of the sheet .
thus gravity waves could be lurking very close in the bulk .
this corresponds to the 4-d branewold efe system not being closed .
\a ) @xmath367 is a ( @xmath1 , 1 ) ( i.e timelike ) hypersurface containing a data strip @xmath150 built by the magaard method from prescribed ` data for the data ' on a timelike @xmath99 .
the shaded region @xmath368 is the region of the ( @xmath369 , 1 ) spacetime supposedly entirely controlled by @xmath150 i.e for which the campbell method can validly provide the evolution .
typically @xmath150 will not cover all of @xmath367 in the @xmath78-direction , so there will be nearby points like @xmath62 outside @xmath150 . consider the future light - cone with apex @xmath62 .
no matter how thin in the @xmath23-direction one considers @xmath368 to be , it is always pierced by causal curves such as @xmath61 on or in the light - cone , by taking @xmath61 to be at a sufficiently slender angle @xmath116 to @xmath367 .
thus information can leak into @xmath368 from elsewhere than @xmath370 , which is a contradiction .
therefore , in parts at least , @xmath368 is arbitrarily thin in the @xmath23-direction .
one can envisage that sometimes there will be enough points like @xmath62 that there is no region @xmath368 at all .
\b ) suppose @xmath99 is spacelike and is a section of the entire hypersurface @xmath367 and @xmath371 is the data strip built from it by the magaard method .
there is now no room for points like @xmath62 !
however , typically @xmath371 will not cover all of @xmath367 in the @xmath78-direction , so there will be nearby points like @xmath227 in the causal past @xmath372 not in @xmath373 . no matter how thin one considers @xmath368 to be , it is pierced by causal curves @xmath374 from @xmath227 _ which do not pass through @xmath373 , by taking these to be at a sufficiently slender angle @xmath375 .
so again , in parts at least , @xmath368 is arbitrarily thin and sometimes there will be enough points like @xmath227 that there is no region @xmath368 at all .
small pieces of timelike hypersurfaces need not hold useful information for hyperbolic - type systems ! _
\b ) diagram of the embedding of flat flrw universes into minkowski spacetime in the standard 5-d minkowski coordinates @xmath376 .
the curved surfaces are the flrw spacetimes .
as one approaches @xmath377 ( corresponding to @xmath263 in flrw coordinates ) , each of these surfaces becomes tangent to the light cone ( characteristics of minkowski ) .
the foliation by these surfaces also becomes bad here because the surfaces intersect .
note also that the flrw geodesics move within each of the surfaces whereas the minkowski geodesics clearly pierce these surfaces .
thus the 4-d and 5-d geodesics in this example are not the same .
the spread of the @xmath280 region of a snapshot @xmath378 containing the @xmath280 junction @xmath281 between the 2 spacelike bulk pieces @xmath379 and @xmath380 upon which the prescribed data is smooth or analytic
. then the evolutions in the domains of dependence @xmath381 and @xmath382 are smooth or analytic , but the causal future of @xmath281 will typically be @xmath280 .
\a ) excision of the inner region in ( 3 , 0 ) black hole data construction by an inversion - in - the - sphere isometry identification between two copies of the black hole .
this is along the lines of the method of images in electrostatics .
\b ) excision of the inner region in ( 4 , 0 ) braneworld black hole data construction would involve making a guess @xmath318 for the excision region and then performing inversion - in - the - sphere isometries pointwise to identify two copies of the black hole .
\c ) one is then to numerically solve for the data .
one can then find the apparent horizon .
if this intersects @xmath318 ( horizon 1 ) then the data may be bad in that the part of the shaded portion of the excised region may be in causal contact with its surroundings .
if the apparent horizon extends too far past the end of the excision region ( horizon 2 ) , one might worry that the singularity also extends past the excised region in which case some of the data prescribed along @xmath316 should not have been prescribed since it lies on the singularity . | we explore some foundational issues regarding the splitting of @xmath0-dimensional einstein field equations ( efe s ) with respect to timelike and spacelike ( @xmath0 - 1)-dimensional hypersurfaces , first without and then with thin matter sheets such as branes .
we begin to implement methodology , that is well - established for the gr cauchy and initial value problems ( cp and ivp ) , in the new field of gr - based braneworlds , identifying and comparing many different choices of procedure .
we abridge fragmentary parts of the literature of embeddings , putting the campbell magaard theorem into context .
we recollect and refine arguments why york and not elimination methods are used for the gr ivp .
we compile a list of numerous mathematical and physical impasses to using timelike splits , whereas spacelike splits are known to be well - behaved .
we however pursue both options to make contact with the current braneworld literature which is almost entirely based on timelike splits . in our study of timelike splits ,
we look at the shiromizu
maeda
sasaki braneworld by means of reformulations which emphasize different aspects from the original formulation .
we show that what remains of the york method in the timelike case generalizes heuristic bulk construction schemes .
we formulate timelike ( brane ) versions of the thin sandwich conjecture .
we discuss whether it is plausible to remove singularities by timelike embeddings .
we point out how the braneworld geodesic postulates lead to futher difficulties with the notion of singularities than in gr where these postulates are simpler .
having argued for the use of the spacelike split , we study how to progress to the construction of more general data sets for spaces partially bounded by branes .
boundary conditions are found and algorithms provided . working with ( finitely ) thick branes would appear to facilitate such a study .
* pde system approach to large extra dimensions + + * electronic address : [email protected] , [email protected] pacs numbers : 04.20ex , 04.50+h |
while the cumulative evidence for neutrino oscillations is very striking , the final proof that the observed anomalies are actually due to neutrino oscillations is still outstanding . in particular ,
the current observations of atmospheric neutrinos @xcite are all consistent with the hypothesis of maximal @xmath2 oscillations , but do not yet exclude some alternative unconventional explanations @xcite .
the main physics goal of the monolith experiment @xcite is to establish the occurrence of neutrino oscillations in atmospheric neutrinos through the explicit observation of the full first oscillation swing in @xmath2 disappearance @xcite , and to investigate and presumably exclude alternative explanations .
this also yields a significantly improved measurement of the oscillation parameters with respect to previous measurements .
the monolith detector will be located at the gran sasso laboratory in italy , and the measurement of the oscillation pattern can be supplemented by measurements in the cern to gran sasso neutrino beam .
a proposal is currently in preparation @xcite .
if approved promptly , a first part of the detector could be operational towards the end of 2004 .
the physics results described in the following sections correspond to an exposure of 4 years with the full detector .
the goals quoted above can be achieved with a high - mass tracking calorimeter with a coarse structure and magnetic field .
a large modular structure has been chosen for the detector ( figure [ fig : module ] ) .
one module consists in a stack of 120 horizontal 8 cm thick iron planes with a surface area of @xmath3 , interleaved with 2 cm planes of sensitive elements .
the height of the detector is thus 12 meters .
thinner plates , 2 and 4 cm thick , were also considered in the past , however the 8 cm plate thickness resulted to be best compromise between physics result and detector costs .
the magnetic field configuration is also shown in figure [ fig : module ] ; iron plates are magnetized at a magnetic induction of @xmath4 t. the detector consists of two modules .
optionally , the downstream module could be complemented by an end cap of vertical planes to improve the performance for non - contained muons from the cngs beam .
the total mass of the detector exceeds 34 kt .
glass spark counters ( resistive plate chambers with glass electrodes ) have been chosen as active detector elements .
they provide two coordinates with a pitch of 3 cm , and a time resolution of 2 ns .
finally , an external veto made of scintillation counters reduces the background from cosmic ray muons .
in the two flavour approximation , the survival probability for neutrino oscillations in vacuum can be expressed by the well known formula @xmath5 where @xmath6 is the distance travelled in km , @xmath7 is the neutrino energy in gev , @xmath8 is the neutrino mixing angle , and @xmath0 is the difference of the mass square eigenvalues expressed in ev@xmath9 . [ cols="^,^ " , ]
provided that the neutrino oscillation hypothesis is confirmed , another goal of the experiment is to further investigate the nature of these oscillations . depending on the oscillation parameters , oscillations into active ( @xmath10 ) or sterile ( @xmath11 ) neutrinos can be distinguished through their different effects on the up / down ratio of neutral current ( nc)-like events , and/or through the presence or absence of matter effects yielding a distortion of the observed oscillation pattern as a function of energy and/or muon charge . even in the absence of sterile neutrinos ,
matter effects are present in the case of a small contribution from @xmath12 oscillations at the `` atmospheric '' @xmath0 .
the corresponding msw resonance might be observable @xcite as a localized @xmath2 rate suppression either in @xmath2 or in @xmath13 .
due to its ability of in situ measurement of the energy of every muon in the multi - tev range , monolith will also be a unique facility for pioneer investigations of cosmic ray muons in the unexplored 100 tev energy region .
the results of these studies will give information which is relevant for the solution of the problem of the knee in the cosmic ray energy spectrum .
other potential physics topics include studies of the primary atmospheric neutrino flux , the search for astrophysical point sources , and a search for a neutrino `` line '' from wimp annihilation in the center of the earth .
neutrino beams from future muon storage rings @xcite ( neutrino factories ) will be essentially pure beams of either @xmath14 or @xmath15 .
the occurence of @xmath16 or @xmath17 oscillations would therefore manifest itself via the appearance of wrong sign muons .
a massive magnetized iron detector like monolith , with good muon charge separation and momentum measurement , could therefore be well suited @xcite for the observation of such oscillations .
as pointed out in @xcite this kind of beam will in particular offer the possibility to measure the @xmath18 mixing angle , currently only constrained by the super - kamiokande and chooz results , and the sign of @xmath0 through matter effects . depending on which of the solar neutrino solutions is correct it might also open the way for the study of cp violation in the neutrino system .
interestingly , the optimization of detectors for the neutrino factory , focusing on wrong sign muon appearance measurements , has yielded a detector @xcite whose basic parameters are very similar to those of monolith .
this is true in particular when the source is far enough away to impinge at a sizeable angle from below ( horizontal geometry of monolith ) .
for instance , a beam from fermilab ( l=7300 km ) would impinge at an angle of 35@xmath19 , and be almost aligned with the gran sasso hall axis , and therefore perpendicular to the magnetic field axis .
the results obtained in the physics studies of ref .
@xcite concerning the measurements of @xmath20 , sign of @xmath0 , and cp violation therefore qualitatively apply to monolith used as a neutrino factory detector .
of course the potential timescale of a neutrino factory is quite different from the one of the current atmospheric neutrino program .
nevertheless , it might be interesting to consider that such a facility might become reality within the lifetime of the monolith project , and that its useful life might be extended accordingly .
monolith is a 34 kt magnetized iron tracking calorimeter proposed for atmospheric neutrino measurements at the gran sasso laboratory in italy .
its main goal is the proof of the neutrino oscillation hypothesis through the explicit observation of a sinusoidal oscillation pattern ( @xmath2 reappearance ) .
other goals include auxiliary measurements in the cern to gran sasso beam , and the investigation of potential @xmath12 and @xmath21 contributions . in the long term
, the detector could also be used in a potential neutrino factory beam .
monolith progress report , lngs - loi 20/99 , cern / spsc 99 - 24 , august 1999 ; + monolith proposal , lngs p26/2000 , cern / spsc 2000 - 031 , august 2000 ; + m. ambrosio et al . , the monolith prototype , proceedings of the bari rpc workshop , october 1999 , ftp://netview.ba.infn.it/rpc/proceedings/gustavino.ps , submitted to nucl . instr . and meth .
super - kamiokande collaboration , y. fukuda et al . , phys .
* 81 * ( 1998 ) 1562 ; + super - kamiokande collaboration , y. fukuda et al .
* b 436 * ( 1998 ) 33 ; + super - kamiokande collaboration , y. fukuda et al .
, phys . lett .
* b 433 * ( 1998 ) 9 .
r. barbieri et al . , p. creminelli , and a. strumia , ifup - th-2000 - 00 , hep - ph/0002199 , february 2000 ; + e. lisi , a. marrone , and d. montanino , hep - ph/0002053 , february 2000 .
v. barger et al .
, phys . lett .
* b 462 * ( 1999 ) 109 . c. albright et al .
, fermilab - fn-692 , may 2000 .
+ v. barger et al .
, hep - ph/0003184 , march 2000 , hep - ph/0004208 , april 2000 .
+ a. cervera et al .
, hep - ph/0002108 , february 2000 .
+ a. bueno , m. campanelli , and a. rubbia , hep - ph/0005007 , may 2000 .
+ m. freund , p. huber , and m. lindner , tum - hep-373/00 , hep - ph/0004085 , april 2000 .
d. cline and d. neuffer , aip conf .
* 68 * ( 1980 ) 846 ; reproduced in aip conf . proc .
* 352 * ( 1996 ) 10 ; + s. geer , phys . rev . * d 57 * ( 1998 ) 6989 ; erratum - ibid .
d 59 ( 1999 ) 039903 .
gomez - cadenas and a. cervera - villanueva , talks given at @xmath22-fact 99 , lyon , 5 - 9 july 1999 .
a. de rujula , m.b .
gavela and p. hernandez , nucl . phys .
* b 547 * ( 1999 ) 21 . v. barger , s. geer and k. whisnant , hep - ph/9906487 . | 8.0 cm 5.0 cm monolith is a proposed massive ( 34 kt ) magnetized tracking calorimeter at the gran sasso laboratory in italy , optimized for the detection of atmospheric muon neutrinos .
the main goal is to establish ( or reject ) the neutrino oscillation hypothesis through an explicit observation of the full first oscillation swing .
the @xmath0 sensitivity range for this measurement comfortably covers the complete super - kamiokande allowed region .
other measurements include studies of matter effects and the nc / cc and @xmath1 ratio , the study of cosmic ray muons in the multi - tev range , and auxiliary measurements from the cern to gran sasso neutrino beam .
depending on approval , data taking with part of the detector could start in 2004 .
the detector and its performance are described , and its potential later use as a neutrino factory detector is addressed . -1.6 cm 16.5 cm -1 cm a. geiser , hamburg university for the monolith collaboration ( * m*assive * o*bservatory for * n*eutrino * o*scillations or * li*mits on * th*eir existence ) bologna , bonn , cnr torino , columbia , lnf frascati , hamburg , humboldt berlin , inr moscow , laquila , lnf frascati , lngs gran sasso , mephi moscow , milano , mnster , napoli , roma , torino , tunis -0.7 in 6.0 in 9.0 in |
granular media are simultaneously one of the most common and complex forms of matter on earth .
this complexity stems , in part , from heterogeneous structure and highly nonlinear particulate interactions @xcite . over the past thirty years
, mechanical wave propagation in ordered granular media has become an active field of research as it provides a setting for the broader understanding of granular media dynamics @xcite .
ordered granular media have also been shown to enable a wide array of novel passive wave tailoring devices that leverage the nonlinear response stemming from the hertzian relationship between elastic particles in contact @xcite , in conjunction with dispersion induced by periodicity @xcite or local resonances @xcite .
experimental configurations used to study mechanical wave propagation in ordered granular media typically involve spherical particles confined by elastic media .
this type of arrangement is particularly common in one- and two - dimensional configurations , and includes macro- to microscale particles .
for example , at the macroscale , elastic rod structures , tracks , and tubes have been used to confine the particles in one - dimensional @xcite and quasi - one - dimensional @xcite configurations , and elastic plates have been used in two dimensions @xcite .
more recently , the dynamics of a two - dimensional monolayer of 1 @xmath0 silica particles adhered to an elastic substrate was studied using a laser ultrasonic technique @xcite .
analytical models used to describe the dynamics of these systems typically only include the interaction between the particles ( often just the normal hertzian contact interaction ) and disregard the effect of the substrate . in reality , even for the simple case of a particle monolayer on a substrate , more complex dynamics involving interactions between the particles and elastic waves in the substrate should be expected .
indeed , a recent experiment @xcite showed that a monolayer of microspheres adhered to a substrate strongly interacts with rayleigh surface waves in the substrate , leading to the hybridization between rayleigh waves and a microsphere contact resonance .
the results of this experiment were analyzed with a simple model involving only vertical ( normal to the substrate surface ) vibrations of isolated particles , following the approach adopted in earlier theoretical works on the interaction of surface oscillators with rayleigh waves @xcite .
however , in reality , the particle motion is not confined to the vertical direction , and the rayleigh wave has a significant horizontal component .
furthermore , the interaction between neighboring particles in contact is expected to significantly influence the dynamics . a notable theoretical work @xcite provided a model for the dynamics of a monolayer adhered to an elastic substrate which accounted for both normal and horizontal motion and interaction between the particles .
however , this study did not take into account particle rotation . a more recent study @xcite demonstrated that the rotational degree of freedom has a profound effect on the dynamics of granular monolayers .
however , the analysis of monolayers on substrates in ref .
@xcite only accounted for normal contact interactions between the particles and the substrate , and the substrate was considered rigid .
the aim of this work is to provide a theoretical model for the contact - based dynamics of a two - dimensional layer of spheres on a substrate , accounting for the elasticity of the substrate , translational and rotational motion of the spheres , and both normal and shear stiffnesses of sphere - to - sphere and sphere - substrate contacts .
we focus on a system with microscale particles that interact with each other and with the substrate via van der waals adhesion forces . rather than postulate the contact stiffness constants ,
we derive them from hertzian contact models .
this imposes certain constraints on the values of the constants : for example , the ratio of the normal and shear contact stiffness between the spheres is a constant only weakly dependent on poisson s ratio .
we consider contact - based modes having frequencies significantly below the intrinsic spheroidal vibrational modes of the spheres , such that they can be described as spring - mass oscillators .
furthermore , we focus on dynamics involving particle and substrate displacements in the sagittal plane , as would be detectable in a laser - based experiment , such as that of ref .
@xcite .
we start with the case of a rigid substrate , where we find three eigenmodes involving vertical , horizontal and rotational motion of the spheres . in the long - wavelength limit
these modes yield three contact resonances , for which simple analytical expressions are obtained .
one of the resonances only involves motion of the spheres normal to the substrate surface , whereas the other two involve mixed horizontal - rotational motion .
we then present our effective medium model , which describes the interaction between the spheres and the substrate .
the results show that the contact resonances interact with rayleigh surface waves , which leads to mode hybridization and avoided crossings .
we discuss the behavior for cases involving both isolated ( non - touching ) and interacting spheres , and demonstrate the important role of rotations in both cases .
we also examine the validity of the effective medium approximation , by comparing the calculations using discrete and effective medium models .
finally , we discuss the implications of our findings for past and future studies on granular monolayer systems .
we consider a monolayer of elastic spheres on a substrate , which can be either close - packed and in contact , or isolated , as shown in fig .
[ model_schem](a ) . in either case , the spheres are assumed to form a square lattice , with the wave propagation direction aligned with the lattice vector , as shown in fig .
[ model_schem](b ) .
we model the layer as an infinite lattice of rigid spheres with diameter @xmath1 and mass @xmath2 , coupled to a semi - infinite , isotropic elastic substrate by normal and shear stiffnesses @xmath3 and @xmath4 , and to nearest - neighbor spheres by stiffnesses @xmath5 and @xmath6 , as schematically shown in fig .
[ model_schem](c ) .
the subscript @xmath7 corresponds to forces acting normal to the surface of the sphere , and @xmath8 to forces acting transverse to the surface of the sphere .
the shear springs generate an associated torque about the sphere center , while the normal springs do not .
the absolute horizontal , vertical , and angular displacements of sphere @xmath9 from its equilibrium state are given by @xmath10 , @xmath11 , and @xmath12 , respectively , and the displacements of the substrate are denoted by @xmath13 , corresponding to displacement in the @xmath14-direction , and @xmath15 , corresponding to displacements in the @xmath16-direction .
we derive the stiffnesses @xmath3 , @xmath4 , @xmath5 , and @xmath6 using derjaguin - muller - toporov ( dmt ) @xcite and mindlin contact models @xcite .
the dmt theory is typically applicable in weakly - adhesive systems with small , stiff particles @xcite , and assumes that the deformation profile is hertzian .
the mindlin model describes the shear stiffness of particles in contact , assuming an applied normal force @xcite . at the microscale ,
adhesive contact models have been explored experimentally in the quasi - static regime using atomic force microscopy and nanoindentation approaches @xcite .
for contact between two spheres ( or a sphere and a halfspace ) having elastic moduli @xmath17 and @xmath18 , and poisson s ratios @xmath19 and @xmath20 , the hertzian restoring elastic force @xmath21 corresponding to displacement @xmath22 of the particle center in the direction normal to the contact surface is given by @xmath23 where @xmath24 is the effective radius of contact ( equal to @xmath25 for sphere - sphere contacts and @xmath26 for sphere - halfspace contacts ) , and @xmath27^{-1}$ ] is the effective modulus . considering the dmt adhesive force @xmath28 acting normal to the contact surface @xcite ( where @xmath29 is the work of adhesion between two surfaces ) , the net normal force is given by @xmath30 to describe the shear contact
, we utilize the mindlin model @xcite , which assumes small relative displacements and no slip at the contact surface . for two elastic bodies with shear moduli @xmath31 and @xmath32 ,
the restoring elastic force @xmath33 to displacement @xmath34 of the particle center in the direction transverse to the contact normal is given by @xcite @xmath35 where @xmath36^{-1}$ ] is the effective shear modulus .
here , the factor of @xmath37 arises from the hertzian relation between the contact radius and @xmath21 . by substituting the relative displacements @xmath38 and @xmath39 into eqs .
[ contactforcennet ] and [ contactforces ] , and linearizing about the equilibrium configuration of @xmath40^{2/3}$ ] and @xmath41 , we derive linearized normal and shear contact stiffnesses @xmath42_{sph}\\
k_s & = 8 \left(\frac{3}{4 } \frac{g^{*^3}}{e^ * } r_c f_{dmt}\right)^{1/3},\label{springconst } \end{split } \end{aligned}\ ] ] with @xmath43 and @xmath44 given by equations of the same form , but with @xmath45 , @xmath46 , @xmath47 , and @xmath48 adjusted for sphere - sphere contacts . in the special case where the spheres and substrate are composed of the same material ,
the relative magnitudes of the stiffness constants are determined exclusively by poisson s ratio @xmath49 of the material , @xmath50 where @xmath51 . + assuming small displacements ( i.e. @xmath52 , @xmath53 , and @xmath54 are much less than @xmath55 )
, the @xmath56 sphere obeys the equations of motion @xmath57\\ i \ddot{\theta}_j = & - k_s r(q_j - u_{0,j } + r \theta_j)\\ & - g_s r [ r(\theta_{j+1 } + 2 \theta_j + \theta_{j-1})\\ & - ( z_{j+1 } - z_{j-1 } ) ] , \label{eom_disc } \end{split } \end{aligned}\ ] ] where @xmath58 and @xmath59 are horizontal and vertical displacements of the substrate surface at the point of contact , respectively .
considering wavelengths much longer than the sphere diameter , we treat the monolayer as an effective continuous medium . by substituting the center difference formulas
@xmath60/(2d ) \approx \partial ( \cdotp ) / \partial x$ ] and @xmath61 / ( d^2 ) \approx \partial^2 ( \cdotp ) / \partial x^2 $ ] into eq .
( [ eom_disc ] ) , we arrive at the equations of motion of the monolayer in effective medium form : @xmath62 the coupling between the monolayer and substrate is described by the following boundary conditions at the surface @xmath63 , which describe the average force acting on the surface due to the motion of the spheres : @xmath64 where @xmath65 and @xmath66 are components of the elastic stress tensor @xcite and @xmath67 = @xmath68 is the area of a primitive unit cell in our square - packed monolayer . the combination of eq .
( [ eom_cont ] ) and the linear elastic wave equations describing waves in the substrate @xcite , coupled by the boundary conditions of eq .
( [ bc ] ) , comprises the complete effective medium model .
we first consider the discrete model , which accurately captures the structural periodicity of the monolayer .
we assume spatially - discrete traveling wave solutions of the form @xmath69 ( with similar terms for the other displacements ) and set the displacements of the substrate surface @xmath58 and @xmath70 to zero ( rigid substrate ) . here
, @xmath71 is the amplitude of a plane wave in the displacement variable @xmath72 , @xmath73 is the angular frequency , and @xmath74 is the wave number . after algebraic manipulation , eq .
( [ eom_disc ] ) is reduced to a homogeneous system of three linear algebraic equations in the amplitudes @xmath71 .
this system possesses non - trivial solutions only for pairs of @xmath75 and @xmath76 that cause the determinant of the system to vanish . enforcing this condition
, we arrive at the following dispersion relation , where the three rows of the determinant correspond to the three equations of eq .
( [ eom_disc ] ) : @xmath77 \end{array}\right| = 0 , \label{drdet_disc}\ ] ] where @xmath78 , @xmath79 , @xmath80 , @xmath81 and @xmath82 are the longitudinal and transverse long - wavelength sound speeds of the discrete monolayer , @xmath83 , @xmath84 , @xmath85 , and @xmath86 . in the effective medium model , which approximates the discrete system at long wavelengths , we substitute spatially - continuous traveling wave solutions of the form @xmath87 ( with similar terms for the other displacements ) into @xmath88 , @xmath89 , and @xmath90 in eq .
( [ eom_cont ] ) with @xmath91 .
following the same procedure as in the discrete model , we arrive at the dispersion relation for the effective medium model : @xmath92 it is particularly instructive to examine the behavior of the effective medium model in the long wavelength limit @xmath93 . in this limit , the displacements vary slowly in space , and the spatial derivative terms of eq .
( [ eom_cont ] ) may be neglected . for the case of a rigid base , eq .
( [ eom_cont ] ) then reduces to the form @xmath94 the equation for @xmath89 decouples from the other two equations and yields a vertical vibrational mode .
the two other equations remain coupled , yielding two modes containing both horizontal and rotational motion . using the moment of inertia of a solid sphere
@xmath95 , we find three resonance frequencies @xmath96^{1/2}\\ \omega_{rh } & = \left[\left(\frac{k_s}{4m}\right ) \left(20 \gamma + 7 + \sqrt{400 \gamma^2 + 120 \gamma + 49}\right)\right]^{1/2}\\ \omega_{hr } & = \left[\left(\frac{k_s}{4m}\right ) \left(20 \gamma + 7 - \sqrt{400 \gamma^2 + 120 \gamma + 49}\right)\right]^{1/2}. \label{res_longwave } \end{split } \end{aligned}\ ] ] where @xmath97 . here , @xmath98 corresponds to a mode with exclusively vertical motion , described by eq .
( [ eom_cont_nosd ] ) .
the other two modes @xmath99 and @xmath100 exhibit both rotational and horizontal ( but not vertical ) motion , with relative amplitudes determined by @xmath101 .
the higher of the two horizontal - rotational modes is predominantly rotational and the lower is predominantly horizontal , hence we have used the notations @xmath102 and @xmath103 , where the first letter in the subscript denotes the dominant motion . if the spheres and substrate are made of the same material , then , by using eq .
( [ springconstreduced ] ) , we can relate the horizontal - rotational frequencies of eq .
( [ res_longwave ] ) to the vertical resonance frequency , with the expressions @xmath104 and @xmath105 . in the limiting case of isolated spheres ( described by @xmath106 ) , @xmath99 and @xmath100 of eq .
( [ res_longwave ] ) become @xmath107 and @xmath108 , respectively . for identical materials , @xmath109 . the dependence of @xmath99 and @xmath100 on @xmath101
is shown in fig .
[ gammaplot ] ( a ) , where it can be seen that @xmath99 originates at @xmath110 for @xmath106 and grows unbounded , while @xmath100 originates at @xmath108 , and approaches @xmath111 asymptotically . in fig .
[ gammaplot ] ( b ) , we plot the horizontal and rotational displacement amplitudes as functions of @xmath101 for these two modes .
different signs of the rotational amplitude indicate that the @xmath102 and @xmath103 modes have different motion patterns . in the former ,
a positive displacement is accompanied by a counterclockwise rotation , while in the latter , it is accompanied by a clockwise rotation .
we note that the zero - frequency mode , @xmath112 , corresponds to the rolling motion of an isolated sphere . with the inclusion of a bending rigidity
, the sphere would not be allowed to freely roll , and instead would undergo rocking motion of a finite frequency . while non - zero bending rigidity is expected to exist in real systems ( for instance
, a microsphere adhered to a substrate does not freely roll ) , the frequency of resulting rocking vibrations is predicted to be orders of magnitude lower than the other contact resonances discussed here @xcite .
bending rigidity would thus act as a small perturbation to the predictions of our model , and we do not include it in our analysis .
to illustrate the importance of particle rotations in the model , we note that in the limiting case of @xmath113 , when there is no rotation , eq .
( [ eom_cont_nosd ] ) yields two resonances : a vertical resonance with frequency @xmath98 , and a horizontal resonance having frequency @xmath114 . for isolated spheres , the effect of rotations can be thought of as a reduction of the effective mass " of the sphere to @xmath115 , which increases the horizontal resonance frequency . for interacting spheres ,
on the other hand , rotations drastically change the dynamics , yielding two horizontal - rotational modes whose frequencies depend on the relative strengths of the sphere - to - sphere and sphere - substrate interactions .
( red line ) and @xmath100 ( black line ) as functions of the stiffness ratio @xmath101 .
( b ) displacement amplitudes of the resonant modes with frequencies @xmath116 ( red lines ) and @xmath100 ( black lines ) , as functions of the stiffness ratio @xmath101 .
solid and dotted lines correspond to @xmath117 and @xmath118 , respectively . for each resonance ,
the amplitudes are normalized such that the sum of squares is unity .
the positive sign of @xmath119 corresponds to counterclockwise rotation . ] as in the case of the effective medium approximation for a rigid substrate , we assume traveling wave solutions of the form @xmath87 ( with similar terms for the other displacements ) into @xmath88 , @xmath89 , and @xmath90 in eq .
( [ eom_cont ] ) .
likewise , we express the variables @xmath120 , @xmath121 , @xmath65 @xmath66 in terms of surface wave solutions for the elastic potentials @xcite @xmath122 and @xmath123 , and then substitute these expressions into eq .
( [ eom_cont ] ) and eq .
( [ bc ] ) . here , @xmath71 is the amplitude of a plane wave in the displacement or potential variable @xmath72 , @xmath124 , @xmath125 , and @xmath126 and @xmath127 are the longitudinal and transverse sound speeds of the substrate , respectively . after algebraic manipulation , eq .
( [ eom_cont ] ) and eq .
( [ bc ] ) are reduced to a homogeneous system of five linear algebraic equations in the five plane wave amplitudes @xmath71 , with coefficients depending on @xmath74 and @xmath73 .
we reach the dispersion relation by seeking nontrivial solutions of this system , which exist only for pairs of @xmath74 and @xmath73 that cause the determinant of the following coefficient matrix to vanish : @xmath128 where @xmath129 is the density of the substrate , and @xmath130 is the area of a primitive unit cell in our square - packed monolayer .
we note that the coupling between the spheres and the substrate is represented by elements @xmath131 , @xmath132 , and @xmath133 of the matrix in eq .
( [ drdet ] ) .
thus , the strength of the coupling can be quantified by the ratio @xmath134 ; if this term is made to vanish ( e.g. by making the mass of each sphere much less than that of the portion of the substrate below it , extending to the depth of material influenced by rayleigh waves ) , then the substrate and monolayer will be effectively decoupled .
we note that if rotations are disregarded ( e.g. by letting @xmath135 ) , eq .
( [ drdet ] ) reduces to the same form as that of the adsorbed monolayer of ref .
@xcite .
it is instructive to consider the long - wave limit when the spatial derivatives in eq .
( [ eom_cont ] ) can be disregarded . in this case , we find the simplified dispersion relation @xmath136 for isolated spheres , there is no approximation in eq .
( [ drdet_reduced ] ) with respect to eq .
( [ drdet ] ) , because in this case the terms generated by the spatial derivatives in eq .
( [ eom_cont ] ) are identically zero . for interacting spheres , the accuracy of dispersion relations calculated with eq .
( [ drdet_reduced ] ) will be assessed below by a comparison with results obtained with eq .
( [ drdet ] ) .
we will see that eq .
( [ drdet_reduced ] ) essentially describes the interaction of contact resonances given by eq .
( [ eom_cont_nosd ] ) with rayleigh surface waves .
in the following calculations , we consider silica spheres of @xmath137 @xmath138 m diameter on a silica substrate , and use the elastic constants ( ref .
@xcite ) @xmath139 73 gpa , @xmath140 0.17 , and work of adhesion ( ref .
@xcite ) @xmath141 0.063 j / m^2^. ( black dotted lines ) , @xmath89 ( red dotted lines ) , and @xmath142 ( blue dotted lines ) , corresponding to the branches of the same numeral for the dispersion of the discrete monolayer adhered to the rigid base shown in ( a ) .
the amplitudes are normalized such that the sum of the squares is unity . ]
we plot numerical solutions of eq .
( [ drdet_disc ] ) , to obtain the dispersion curves for the discrete model of interacting spheres on a rigid base , as shown in fig .
[ dr_layer](a ) . in our description of a rigid substrate
, we assume that no elastic waves propagate in the substrate , but allow local deformation at the points of contact for the purpose of the contact stiffness calculation ; this preserves the same contact stiffnesses as in the elastic substrate analysis .
we note that due to the periodicity of the system , all three branches have zero - group velocities at the edge of the first irreducible brillouin zone @xcite of the monolayer . by substituting the solutions shown in fig .
[ dr_layer](a ) into the coefficient matrix of the corresponding algebraic system , we numerically determine the amplitudes of the sphere displacements , which we plot in fig .
[ dr_layer](b - d ) . by comparing the calculated displacements of with the dispersion curves
, we see that each branch takes on the character of its respective contact resonance in the limit @xmath93 .
one can see that each of the three modes generally involves both vertical and horizontal , as well as rotational motion ( albeit the rotational component of mode ii is quite small ) .
the existence of the three modes with mixed displacements is a consequence of the inclusion of the rotational degree of freedom : without rotations , there would be two modes , one vertical and one horizontal .
we note that , in the special case @xmath143 , the mode originating at @xmath100 becomes purely horizontal and decouples from the other two modes .
the remaining modes ( characterized by vertical translation and rotation ) are generally consistent with the results of ref .
@xcite , for the case of normal contact with a rigid surface and no bending rigidity . since ref .
@xcite considered hexagonal packing , the behavior is analogous at long wavelengths , but diverges at short wavelengths due to discrete lattice effects .
the dotted lines in fig .
[ dr_layer](a ) show dispersion curves calculated with the effective medium model as per eq .
( [ drdet_cont_rb ] ) .
the effective medium approximation yields accurate results at long wavelengths but fails at shorter wavelengths with the unphysical behavior of the first mode , whose frequency goes to zero . at even shorter wavelengths , as shown in fig .
[ dr_em_compare ] , the effective medium dispersion curves of modes ii and iii asymptotically approach straight lines with slopes given by the longitudinal and transverse sound speeds in the monolayer .
this asymptotic behavior has been described by kosevich and syrkin @xcite . however , as can be seen from the dispersion curves generated using the discrete model in fig .
[ dr_em_compare ] , this asymptotic behavior does not occur in our system due to the spatial periodicity of the monolayer . as a result , the inclusion of the first- and second - order spatial derivative terms of eq .
( [ eom_cont ] ) does not deliver much additional understanding of the dynamics of our system .
black dash - dotted lines denote the long - wavelength longitudinal and transverse sounds speeds of the monolayer . ]
we numerically solve eq .
( [ drdet_reduced ] ) for the isolated spheres case using @xmath144 and all other parameters derived in sec .
[ contact ] , and plot the resulting dispersion relation for the effective medium model , as shown in fig .
[ dr_iso ] .
this dispersion relation exhibits classic avoided crossing " behavior @xcite about the resonance frequencies @xmath98 and @xmath145 . in this model , surface acoustic waves ( saws ) in the substrate behave as classical rayleigh waves at frequencies far from the contact resonances , and the dispersion curves follow the line corresponding to the substrate rayleigh wave speed @xmath146 @xcite .
conversely , sphere motions dominate those of the substrate at frequencies close to the contact resonances . for phase velocities greater than @xmath127 , which correspond to the region @xmath147 , the wave numbers that solve eq .
( [ drdet ] ) are complex valued ; these solutions are leaky " modes for which energy leaves the surface of the substrate , and radiates into the bulk .
this isolated spheres case is particularly applicable in systems where adhesion between particles is negligible , e.g. for : macroscale particles without lateral compression where the dominant static compression is due to gravity and is between the particles and substrate ; or for microscale particles , if the separation distance between particles is larger than the range of adhesion forces .
( black solid lines ) , @xmath121 ( red solid lines ) , @xmath88 ( black dotted lines ) , @xmath89 ( red dotted lines ) , and @xmath142 ( blue dotted lines ) , corresponding to the branch denoted by the same numeral in ( a ) .
the amplitudes are normalized such that the sum of the squares is unity . ] in fig .
[ dr_inter](a ) , we plot numerical solutions of eq .
( [ drdet_reduced ] ) for the long wavelength limit of the effective medium model with interacting spheres .
the amplitudes of the sphere and substrate displacements are calculated in the same manner as in fig .
[ dr_layer ] , and are plotted in fig .
[ dr_inter](b - e ) . in fig .
[ dr_inter](a ) , we observe features qualitatively similar to the dispersion relation for isolated spheres in fig .
[ dr_iso ] , with the exception of a third avoided crossing at frequency @xmath103 .
the mode shapes reveal the ways in which each of the branches are influenced by the contact resonances , as well as long and short wavelength asymptotic behavior of our system . in the long wavelength limit
, the substrate motions closely resemble rayleigh saws @xcite , with a mix of vertical and horizontal motions .
since the frequencies of waves in this regime are well below the contact resonances , the effect of the spheres is negligible , and the monolayer moves in phase with the substrate surface . at short wavelengths , it can be clearly seen that the first , second , and third lowest branches exhibit motions dominated by the displacements @xmath88 , @xmath89 , and @xmath90 , respectively ( each corresponding to a resonant mode of the monolayer ) , while the highest branch tends toward the rayleigh saw .
the effects of proximity to the contact resonances are well illustrated , for example , by branch iii of fig .
[ dr_inter](a ) , which exhibits large vertical sphere motions at its starting point near @xmath98 , resembles the rayleigh saw as it approaches and crosses the @xmath146 line , and transitions into large rotational sphere motions after bending around the avoided crossing with @xmath102 . in order to examine the behavior of our system throughout the entire brillouin zone , we superimpose the dispersion curves for the effective medium model of interacting spheres on an elastic base including higher order spatial derivative terms ( the full eq .
( [ drdet ] ) ) with the dispersion curves for the discrete monolayer on a rigid substrate ( eq . ( [ drdet_disc ] ) ) , as shown in fig . [ dr_re_compare ] . at long wavelengths the discreteness of the monolayer is insignificant , and
the dispersion is well described by the effective medium model .
furthermore , we note that at long wavelengths the dispersion curves calculated using the effective medium model including higher order terms shown in fig .
[ dr_re_compare ] , only slightly deviates from the dispersion calculated using the effective medium model with the higher order terms neglected shown in fig .
[ dr_inter](a ) .
the only noticeable effect is a downshift in frequency of the avoided crossing between the rayleigh wave and the @xmath102 resonance ; since the latter intersects at the highest wave vector of the three contact resonances , calculations with eq .
( [ drdet_reduced ] ) in this case are the least accurate . in fig .
[ dr_re_compare ] , at short wavelengths , beyond the avoided crossings with the rayleigh wave branch , the elasticity of the substrate has little effect on the dynamics , and the dispersion can be described using the discrete model for interacting spheres on a rigid substrate .
we suggest that by `` stitching together '' the effective medium model for spheres on an elastic substrate with the discrete model for spheres on a rigid substrate , we can simultaneously capture the interaction of saws with the monolayer at long wavelengths and effects caused by the discreteness of the spheres at short wavelengths .
past the avoided crossings , the two sets of curves in fig .
[ dr_re_compare ] stitch together smoothly , resulting in a full picture of the monolayer dynamics on the elastic substrate . ( a ) . ]
we expect the presented results to be useful for predicting complex dynamic responses and extracting effective contact stiffnesses from measurements of acoustic dispersion in a manner similar to boechler _ et al .
the findings described above invite several questions , including whether our model of a square lattice is applicable to results on hexagonally packed monolayers , and why horizontal - rotational resonances were not observed in the experiment @xcite .
we believe that the assumption of the square lattice is not essential . for isolated spheres , eqs .
( [ eom_cont ] ) and ( [ bc ] ) with @xmath43 and @xmath44 set to zero can be obtained for any arrangement of the spheres , periodic or random , with the only parameter depending on the arrangement being the surface area per sphere a. for interacting spheres , the results generally do depend on the lattice structure and the propagation direction .
however , the contact resonances given by eq .
( [ res_longwave ] ) correspond to the @xmath148 limit and , consequently , do not depend on the propagation direction .
the relative positions of the three contact resonances may be different in the long wavelength limit between a hexagonal and square packed lattice , but their presence should still be expected in both cases .
we suggest that the reasons why horizontal - rotational resonances were not observed in ref . @xcite may be the following .
since the measurements were not sensitive to horizontal motion , the @xmath99 and @xmath100 resonances could only be detected when they hybridized with saws near avoided crossings , and since the avoided crossings with @xmath99 and @xmath100 resonances are more narrow than the one with the @xmath149 resonance , they could have been missed .
furthermore , our model assumes that all spheres are either connected by identical springs or are isolated . if the contact stiffness between spheres were to vary widely ( some neighboring spheres being in contact and others not , for example ) , then distinct resonances may be absent .
in addition , the upper ( @xmath99 ) resonance may have been outside the range of the measurements in ref . @xcite .
further experimental studies of monolayer dynamics in conjunction with exploration of ways to control sphere - to - sphere contacts should help resolve the discrepancy between the theory and experiment .
while the main focus of this work has been on microgranular monolayers , our theory is equally valid for macroscale systems . in this case , the contact springs would be determined by gravity and , possibly , applied lateral static compression @xcite , rather than by adhesion forces .
we note that several past experimental works on macroscale granular systems @xcite have observed systematic upshifts in frequency relative to theoretical predictions , and have suggested uncertainties in material parameters and experimental setups , as well as deviations from hertzian contact behavior as possible causes . as per the results from our model ,
the presence of additional degrees of freedom and interactions between spheres and substrate may also be possible causes . in the absence of the external lateral compression , highly nonlinear `` sonic vacua '' @xcite
should also be expected .
generally , as amplitudes are increased , interesting nonlinear dynamics are expected for both micro- and macroscale monolayers due to nonlinearity of hertzian contacts between the particles @xcite and between the particles and the substrate @xcite .
we have developed a model for wave propagation in granular systems composed of a monolayer of spheres on an elastic substrate .
our model expands on those used in previous works by including the elasticity of the substrate , horizontal and rotational sphere motions , shear coupling between the spheres and substrate , and interactions between adjacent spheres .
we have shown that a monolayer of interacting spheres on a rigid substrate supports three modes involving vertical , horizontal , and rotational motion . in the long - wavelength limit , these modes yield three contact resonances , one purely vertical and two of mixed horizontal - rotational character . on an elastic substrate ,
these resonances hybridize with the rayleigh surface wave yielding three avoided crossings . for isolated spheres
, the frequency of the lower horizontal - rotational resonance , in the absence of bending rigidity , tends to zero and only two contact resonances with two respective avoided crossings remain . by comparing the effective medium ( valid for long wavelengths ) to the discrete formulation of our model
, we have demonstrated that for the presented microsphere monolayer example , the effective medium model can be used to describe the interaction of the contact resonances with the rayleigh waves in the substrate , but loses accuracy at shorter wavelengths . in that case , the substrate can be considered rigid , and the discrete model is more appropriate .
this model is scalable in that it can be adapted for use with both macro- and microscale systems , and provides a means to experimentally extract contact stiffnesses from dynamic measurements .
opportunities for future studies include exploration of analogous models for granular monolayers in the nonlinear regime , as well as analysis of the transverse modes of a monolayer of spheres on an elastic substrate ( the latter involves transverse horizontal displacement and rotations of the spheres , as well as shear horizontal acoustic waves in the substrate ) .
further experiments with macro- and microscale granular monolayers will help guide the modeling effort .
the authors greatly appreciate discussions with vitaly gusev . s.w . and n.b . gratefully acknowledge support from the us national science foundation ( grant no .
cmmi-1333858 ) and the us army research office ( grant no . w911nf-15 - 1 - 0030 ) .
the contribution by a.a.m . was supported by the national science foundation grant no .
che-1111557 . a. n. lazaridi and v. f. nesterenko , _ prikl
* 26 * , 115 ( 1985 ) ; c. coste , e. falcon , and s. fauve , _ phys .
e _ * 56 * , 6104 ( 1997 ) ; s. job , f. melo , a. sokolow , and s. sen , _ phys .
lett . _ * 94 * , 178002 ( 2005 ) ; c. daraio , v. f. nesterenko , e. b. herbold , s. jin , _ phys . rev .
e _ * 73 * , 026610 ( 2006 ) ; n. boechler and c. daraio , in _ proceedings of the asme 2009 design engineering technical conferences , san diego , 2009 _ , pp .
271 - 276 ; n. boechler , g. theocharis , s. job , p.g .
kevrekidis , m.a .
porter , and c. daraio , _ phys .
lett . _ * 104 * , 244302 ( 2010 ) ; n. boechler , j. yang , g. theocharis , p. g. kevrekidis , and c. daraio , _ j. appl
. phys . _ * 109 * , 074906 ( 2011 ) .
h. p. rossmanith and a. shukla , _ acta mech . _
* 42 * , 221 ( 1982 ) ; c. daraio , d. ngo , v. f. nesterenko , and f. fraternali , _ phys . rev .
* 82 * , 036603 ( 2010 ) ; j. yang , s. dunatunga , c. daraio , _ acta mech . _
* 223 * , 549 ( 2012 ) ; b. gilles and c. coste , _ phys .
* 90 * , 174302 ( 2003 ) ; a. leonard , f. fraternali , and c. daraio , _ exp .
* 53 * , 327 ( 2011 ) ; n. boechler j. eliason , a. kumar , a. a. maznev , k. a. nelson , and n. fang , _ phys .
lett . _ * 111 * , 036103 ( 2013 ) . | we present a model for wave propagation in a monolayer of spheres on an elastic substrate .
the model , which considers sagittally polarized waves , includes : horizontal , vertical , and rotational degrees of freedom ; normal and shear coupling between the spheres and substrate , as well as between adjacent spheres ; and the effects of wave propagation in the elastic substrate . for a monolayer of interacting spheres ,
we find three contact resonances , whose frequencies are given by simple closed - form expressions .
for a monolayer of isolated spheres , only two resonances are present .
the contact resonances couple to surface acoustic waves in the substrate , leading to mode hybridization and `` avoided crossing '' phenomena .
we present dispersion curves for a monolayer of silica microspheres on a silica substrate , assuming adhesive , hertzian interactions , and compare calculations using an effective medium approximation to a discrete model of a monolayer on a rigid substrate . while the effective medium model does not account for discrete lattice effects at short wavelengths
, we find that it is well suited for describing the interaction between the monolayer and substrate in the long wavelength limit .
we suggest that a complete picture of the dynamics of a discrete monolayer adhered to an elastic substrate can be found using a combination of the results presented for the discrete and effective medium descriptions .
this model is potentially scalable for use with both micro- and macroscale systems , and offers the prospect of experimentally extracting contact stiffnesses from measurements of acoustic dispersion . |
stars form due to the local collapse of material in molecular clouds .
the conditions prior to the collapse result from a complex interplay between self - gravity , turbulence , magnetic fields and thermodynamics . understanding the effects of each of these influences leads to a better knowledge of how stars form . in this paper
we are interested in the role played by interstellar turbulence and it s effect on the structure of giant molecular clouds ( gmcs ) .
the subject of astrophysical turbulence is complex and not fully understood , though there are many useful reviews on the topic ( e.g. elmegreen & scalo , scalo & elmegreen , brandenburg & nordlund @xcite ) .
here we are interested in comparing the structure of nearby gmcs to investigate whether it has an influence on the observed star formation properties . ultimately this might also be used to determine the nature of the turbulent field , i.e. driven by a compressive forcing or solenoidal driving ( e.g. federrath et al . ) .
we will probe the column density structure in nearby gmcs by means of extinction maps derived from near - infrared observations .
this is the least biased way to estimate column density ( goodman et al .
@xcite ) . in the first paper in this series ( rowles & froebrich @xcite , hereafter paperi )
, we presented new all - sky extinction maps derived using data from the 2micron all - sky survey ( 2mass , skrutskie et al .
we used the _
median _ near - infrared colour excess technique ( nice ) to calculate the extinction ( see lada et al .
the nearest 25 , 49 and 100 stars to the centre of each pixel were used , hence the noise can be considered constant throughout the map .
these extinction maps are therefore referred to as _ con - noise _ maps . in our second paper ( froebrich & rowles @xcite , hereafter paperii ) , we analysed the column density and mass distributions of a selection of 16 nearby gmcs . to facilitate this we determined new extinction maps using only the stars within each pixel ( i.e. no oversampling ) , which therefore have a constant spatial resolution and
are referred to as _ con - res _ maps . as a result of the analysis we found a universal star formation threshold of about [email protected] @xmath3 .
this threshold separates two different regions in the clouds . below the threshold , at low column densities , turbulence dominates the structure , while at higher column densities gravity is the dominant force . the low @xmath3 part of the clouds could be fitted by a log - normal distribution .
there were significant differences in the slopes of the column density and mass distributions when considering only the low @xmath3 regions .
this shows that the properties of the turbulence differ depending on the environment of the cloud .
regarding the high @xmath3 regions , we found no such differences , implying that gravity solely dominates these parts . using our extinction maps we can derive column density structure functions similar to velocity structure functions ( e.g. padoan et al .
@xcite ) for each molecular cloud . this will allow us to perform a comparison with models of interstellar turbulence .
predictions of structure functions resulting from a turbulent medium have been presented e.g. by schmidt et al .
@xcite , kolmogorov @xcite , she & leveque @xcite and boldyrev @xcite ( hereafter s08 , k41 , sl94 and b02 , respectively ) . in this paper
we test for correlations between structure function parameters and the properties of the clouds .
we also examine the cloud structures using the @xmath1-variance technique ( see stutzki et al . and ossenkopf et al . ) and in particular the mass spectral index scaling coefficient . in sec.[sfmeth ] we describe the methods used for our analysis . in sec.[results ] we give the results for the clouds selected .
we discuss these results and give conclusions in secs.[discuss ] and [ conclusions ] , respectively .
the general definition of the structure function is given in eq.[struct1 ] , ( e.g. lombardi et al . ;
padoan et al . @xcite and padoan et al .
@xcite ) . here
the equation is expressed in terms of our observable the column density or optical extinction @xmath3 .
@xmath4 @xmath5 is the distance between points , @xmath6 represents a position in the map , @xmath3 is the optical extinction at @xmath6 ( or @xmath7 ) and @xmath8 is the order of the structure function . for @xmath9
the equation is the two - point correlation function of the extinction map .
the brackets @xmath10 denote that the average over all pixel positions @xmath6 and all possible directions for the separation @xmath5 of points is applied . for each order
@xmath8 we find the scaling exponent @xmath11 by fitting a power - law to values of @xmath12 against @xmath5 .
this assumes that the scaling exponents are related to the structure functions by eq.[struct2 ] ( e.g. padoan et al .
@xcite ) : @xmath13 equation[struct2 ] allows us to determine the scaling exponents @xmath11 which are then normalised to the third order @xmath14 , as a universal behaviour should be exhibited at low reynolds numbers ( determined by benzi et al .
@xcite ) . for each gmc investigated we used the range of spatial scales @xmath5 from 0.1pc to 1.0pc to fit the power law exponent , in order to be able to compare the results for all clouds .
we then follow padoan et al .
@xcite and assume the column density scaling exponents @xmath15 are equivalent to the velocity scaling exponents @xmath16 ( dubrulle @xcite ) .
they can then be expressed by a relation of the form as shown in eq.[dubeq ] .
@xmath17 here @xmath18 is the intermittency and @xmath1 is related to the co - dimension @xmath19 and intermittency by : @xmath20 .
the fractal dimension @xmath21 of the cloud is related to the co - dimension by @xmath22 . using this equation ,
the k41 relation can be expressed by setting @xmath23=0 .
as mentioned previously we use the projected column density scaling parameter @xmath24 instead of the velocity scaling parameter @xmath25 to keep the same description of the structure function as in padoan et al .
@xcite , though it is not known _ a priori _ whether the parameters correspond to each other .
however , padoan et al .
@xcite found that the scaling exponents derived from integrated intensity images follow the velocity scaling using the b02 relation .
we may then derive the parameters @xmath1 , @xmath19 , @xmath18 and @xmath21 for each cloud and compare with the velocity field structure functions obtained from the aforementioned theoretical works by k41 , sl94 , b02 and s08 .
when fitting our data to eq.[dubeq ] , we varied @xmath1 in the range 0.02 to 1.20 with increments of 0.01 , and @xmath19 ranged from 0.02 to 3.00 with increments of 0.01 to find the best match . only fits with an @xmath26 better than 0.1 ( in units of @xmath15 ) were considered a good fit .
this is justified given that @xmath26 is not an absolute measure of the goodness of fit .
regarding the previously published models , the simplest to consider is that of k41 .
the energy over a wide range of lengths ( known as the ` inertial range ' ) in turbulent flows is redistributed from larger scales into ever smaller scales until the effects of viscosity become important . in k41 , kolmogorov considered the structure velocity functions s@xmath27 and s@xmath28 , showing that both are power laws of the form of eq.[struct2 ] , with the exponents being 2/3 and 1 , respectively . due to the self - similarity of turbulence at different scales ,
this law can be extended to all powers of @xmath8 such that @xmath29 .
however , experimental measurements for the scaling exponents show a deviation from the k41 relation for turbulence , for higher orders of @xmath8 - known as _ intermittency_. this is exhibited in the non - gaussian tails when plotting the probability density functions ( pdfs ) of e.g. the column density , and anomalous scaling of higher order structure functions ( e.g. anselmet et al .
therefore , we need a relation for the scaling that considers the effects of intermittency .
sl94 derived a scaling relation that accounts for intermittency .
their relation shows good agreement when compared to simulations of structures that are not influenced by magnetic fields . for incompressible turbulence
sl94 derived that @xmath30 and @xmath31 .
therefore , the sl94 model assumes that @xmath32 , i.e. that the most intermittent structures are filaments .
b02 extended the sl94 relation to model highly supersonic turbulence and also take account of magnetic fields .
they found that @xmath33 with @xmath31 as with the sl94 model . under the b02 model
it follows that @xmath34 , i.e. that the structures are sheet - like in form . in s08 , the scaling exponents were found to be well described by log - poisson models . in these models
@xmath35 , rather than @xmath36 .
the modelling applied two types of forcing to drive the turbulence , solenoidal ( divergence - free ) and compressive ( rotation - free ) .
values for the parameters @xmath1 and @xmath19 differ depending on the type of forcing applied .
perhaps the nature of forcing in real clouds can be found by comparing the observational data with the s08 model . to ascertain the effect of noise in the data when calculating structure functions from an image we undertook a number of tests . in particular
we desired to know whether ( increased ) random noise leads to a significant and/or systematic deviation of @xmath15 compared to a test image without noise , or a low noise @xmath3 map .
in the first instance we made an artificial @xmath3 map of 400@xmath37400 pixels containing a circle of radius 124 pixels , where the pixel values in the circle are varied linearly from 0.81mag ( edge ) to 10.0mag ( centre ) . this may be considered analogous to a perfectly spherical cloud with a smooth density variation from the core to the edge ( admittedly , an unlikely scenario in practice ) .
all other pixels in the map were set to zero .
the structure function of this map happened to correspond to kolmogorov type turbulence .
we then added 0.28mag gaussian noise ( corresponding to the 1@xmath38 noise in our 49@xmath39 nearest neighbour map see paperi ) to this map and fit the parameters @xmath1 , @xmath19 , @xmath18 and @xmath21 .
this process was continued by adding further amounts of noise to the basic map with 0.28mag ( 1@xmath38 ) noise .
the quantities added were a further 0.0625@xmath38 , 0.125@xmath38 , 0.1875@xmath38 and 0.25@xmath38 , generating five images in total .
the results of this exercise are discussed in sec.[noires ] .
in addition to investigating the effect of noise on artificial clouds , we added random noise to a selection of real clouds .
the clouds chosen were chameleon , circinus , corona australis , oriona and orionb .
we used the nearest 49@xmath39 nearest neighbour @xmath3 maps and added 0.0625@xmath38 , 0.125@xmath38 , 0.1875@xmath38 and 0.25@xmath38 noise .
we performed the structure function analysis as with the artificial clouds .
these results are also discussed in sec.[noires ] . the @xmath1-variance method for analysing molecular cloud structures in astronomical images ( i.e. in two dimensions )
was introduced by stutzki et al . and improved by ossenkopf et al . .
the method works by measuring the quantity of structure on a particular length scale ( e.g. @xmath5 ) and filtering the data @xmath40 with a spherically symmetric ` down - up - down ' type function .
the function ( denoted @xmath41 ) , which treats different regimes of @xmath5 separately , is given by eq.[frehat1 ] below .
@xmath42 the rhs of eq.[frehat1 ] is defined using eqs.[mexhat1 ] and [ mexhat2 ] .
@xmath43 @xmath44\ ] ] the @xmath1-variance , @xmath45 is then defined in eq.[delva1 ] .
@xmath46 the filter function @xmath41 denotes the fourier transform of the filter function with size @xmath47 and diameter ratio @xmath48 .
two particular filter types were found to best probe the structure in molecular clouds , a ` mexican hat ' filter and a ` french hat ' filter ( ossenkopf et al . ) .
both filters treat the core and the annulus separately ( eq.[frehat1 ] ) .
the optimal filter function for molecular clouds was found to be either a mexican hat filter with a diameter ratio of 1.5 , ( or a french hat filter with a diameter ratio of 2.3 - ossenkopf et al . ) .
we used a mexican hat filter with a diameter ratio of 1.5 for our analysis .
we applied this method to our _ con - noise maps_. to further investigate the dimensions of structures in the clouds ( i.e. filaments etc . )
we also applied this method to the corresponding star density maps .
the structure shown in the star density maps is enhanced since the extinction in the @xmath3 maps is roughly proportional to the @xmath49 of the star density .
therefore , both the @xmath3 and star density maps were used for this analysis .
the results are shown in sec.[delvrs ] .
using the @xmath1-variance technique allows us to also calculate the mass spectral index scaling exponent ( denoted @xmath50 ) of each cloud .
this is found by fitting a power - law to @xmath51 against scale .
the power law is fitted over the range of size of the cloud , i.e. between the smallest and largest scales of the cloud .
the smallest scale is determined by eliminating the pixels affected by oversampling . for the largest scale we chose the point where the increase in @xmath51 begins to tail off before the peak .
these results are also shown in sec.[delvrs ] . as with the structure functions it is also necessary to consider the effects of noise on the @xmath1-variance results .
this was investigated in a similar way as for the structure functions ( see sec.[noistr ] ) .
the same set of test images of 400@xmath37400 pixels containing a circle , as well as real clouds were used .
the @xmath1-variance of these image was calculated .
increments of gaussian noise were added to each of the images in exactly the same way as for the structure function analysis .
the results are presented in sect.[noiresdvar ] .
we estimate the average physical density of the cloud material in order to relate it to the cloud properties obtained from our analysis of the column density and the star formation properties . the average density of material can be estimated from the mass of the cloud @xmath52 above a certain @xmath3 threshold ,
the number of pixels @xmath53 the cloud is covering in our @xmath3 map with values above this threshold , and the distance @xmath54 to the cloud in the following way : @xmath55 = \frac{8.78 \cdot 10^{15 } \cdot m [ m_{\odot } ] } { n^{\frac{3}{2 } } \cdot ( x [ '' ] d [ pc ] ) ^3},\ ] ] here the average density is expressed in solar masses per cubic parsec , the mass in solar masses , the pixel size @xmath56 in arcsec and the distance @xmath54 in parsec .
the power of @xmath57 at the number of pixels of the cloud assumes that the two visible dimensions in the plane of the sky are a good representation of the dimension along the line of sight .
note that if one wants to express the density in units of particles per cubic centimeter , then the constant 8.78@xmath5810@xmath59 needs to be multiplied by a factor of 40 .
we determine two average densities for the purpose of this paper .
i ) the average density of the entire cloud .
this includes all material which is above the self shielding column density threshold of 1mag @xmath3 .
ii ) the average density of the material above the cloud s star formation threshold .
this corresponds to the density of the material which is potentially involved in star formation .
the results are shown in sect.[denres ] .
for the analysis of the structure functions we used the three available _ con - noise _ maps ( see paperi ) for the following reasons : \i ) the noise has a less predictable effect on the results when using the _ con - res _ maps .
this is caused by the fact that each pixel has a different number of stars contributing to the @xmath3 value , and hence variable noise . in some cases
only very few stars are used .
one drawback of using the _ con - noise _ maps for this analysis is the variable spatial resolution between pixels .
this is most notable in regions of high extinction .
however , we only include pixel separations above 0.1pc ( see above ) , which are generally above the spatial resolution .
furthermore , we determine the structure functions only for the parts of the cloud where turbulence dominates , i.e. where @xmath3 is greater than 1mag and less than the cloud s individual star formation threshold ( generally below 8mag , see paperii ) .
\ii ) the obtained values for @xmath60 and @xmath14 are very low when using the _ con - res _ maps .
this could be a result of the generally higher and , more importantly , variable noise in these maps . with the _ con - noise _ maps we found much more reasonable values for @xmath60 and @xmath14 .
we calculate a mean value for @xmath60 of [email protected] and @xmath14 of [email protected] . for individual clouds , using taurus as an example , we find @xmath60=0.98 and @xmath14=1.41 , compared to @xmath60=0.77 and @xmath14=1.10 ( padoan et al .
@xcite ) . for perseus
we find @xmath60=1.24 and @xmath14=1.75 , compared to @xmath60=0.83 and @xmath14=1.18 .
thus , we determine the structure function of every gmc for each of the three available @xmath3 maps , utilising only extinction values above the self - shielding limit ( 1mag ) and below the star formation threshold of the particular cloud .
note that every individual extinction value used for these calculations has a signal - to - noise ratio of at least three ( paperi ) .
hence , the obtained structure functions are highly reliable , as indicated by the small uncertainties for @xmath15 ( e.g. fig.[sfeg1 ] ) . for each of the three structure functions we determine the best fitting set of parameters @xmath1 , @xmath19 , @xmath18 and @xmath21 , and a scatter for each parameter by considering only the fits with an @xmath26 of less than 0.1 ( in units of @xmath15 ) .
the final structure function parameters for each cloud are then determined as weighted averages of the three obtained values . an example plot showing @xmath15 against @xmath8 for the auriga1 cloud together with the best fit
is shown in fig.[sfeg1 ] , where we used the 49@xmath39 nearest neighbour map .
similar plots for all clouds are shown in the appendix[indplt ] . a plot showing the variation of @xmath15 against @xmath8 for the auriga1 cloud and the 49@xmath39 nearest neighbour map . our data ( + signs )
is compared against the models of k41 ( solid line ) , sl94 ( dotted line ) and b02 ( dashed line ) .
the best fit values of @xmath1 , @xmath19 , @xmath18 , @xmath21 and the @xmath26 value are shown in the upper left corner .
a plot of the difference of the fit from the data against @xmath8 is shown as a red solid line in the lower panel.,width=302 ] the determined parameters of the structure function ( @xmath1 , @xmath19 , @xmath18 and @xmath21 ) for all the clouds and their uncertainties are listed in table[strrestab ] .
the @xmath26 of the fit is also shown .
the values for @xmath1 lie in the range 0.45 to 1.08 with a mean of [email protected] .
therefore , even after allowing for uncertainties , the values are more or less in the middle of those quoted in the literature ( @xmath1=2/3 in e.g. padoan et al .
@xcite , sl94 , b02 ; @xmath1=1 in e.g. schmidt et al .
@xcite ) . the co - dimension values @xmath19 lie between 1.18 and 2.51 with a mean of [email protected] . in turn
the fractal dimension of our sample of clouds has a value of [email protected] , rendering them filament like rather than sheet like .
the values of @xmath18 ( the degree of non - intermittency ) in our sample range from 0.38 to 0.65 with a mean of [email protected] .
the literature usually quotes a values for @xmath18 of either 2/3 ( sl94 ) or 1/3 ( b02 ) .
therefore , the value for our gmc sample lies about halfway between these values . [
cols="<,^,^,^,^,^,^,^,^,^,^,^,^,>,>,>,>,^,^,^ " , ] we added noise to a test image and to a selection of the extinction maps of real clouds to investigate the effect on the structure functions and their fitted parameters ( @xmath1 , @xmath19 , @xmath18 , @xmath21 ; see sec.[noistr ] ) .
the results of this exercise are shown in table[tesnoires ] . in the top part of the table we list the values of the structure function parameters obtained for our test image and the test image plus noise .
as one can see , the artificial cloud corresponds to a structure in agreement with kolmogorov type turbulence ( @xmath61 ) .
this also implies that the actual values of the other parameters are meaningless ( see eq.[dubeq ] ) .
once our typical noise of 0.28mag @xmath62 has been added the parameters change significantly .
the values for @xmath1 are 0.5 and the structure corresponds to an almost two - dimensional object ( @xmath21=1.9 ) . when the noise is further increased no systematic and/or significant changes
are observed .
this has two implications : i ) our typical noise can change the structure function parameters significantly compared to a column density map which is free of noise .
ii ) once the image has some observational noise , a small increase in the noise will not change the values of the determined structure function parameters significantly or systematically .
all changes are well below the uncertainties of the parameters .
this implies that it does not matter which of our @xmath3 maps we use ( and indeed justifies that we average the results from all three maps ) , as the variation of a factor of two in the number of stars leads only to an increase / decrease of the noise in the maps by 33% ( caused in part by the covariance , see paperi ) .
the bottom part of table[tesnoires ] illustrates the latter part of the above discussion for a real cloud ( corona australis ) . with the standard noise in the 49@xmath39 nearest neighbour map of 0.28mag @xmath3
we obtain the listed values for the structure function parameters .
when increasing the noise by another small amount ( at maximum 25% ) , again we do not find any significant and/or systematic changes in the determined structure function parameters .
c|cccccccccc noise & @xmath1 & @xmath63 & @xmath19 & @xmath64 & @xmath18 & @xmath65 & @xmath21 & @xmath66 & @xmath26 & @xmath67 + @xmath68 $ ] & & & & & & & & & & + + 0.0000 & 0.02 & 0.01 & 2.03 & 0.58 & 0.99 & 0.01 & 0.97 & 0.58 & 0.01 & 0.01 + 1.0000 & 0.50 & 0.06 & 1.18 & 0.41 & 0.55 & 0.11 & 1.82 & 0.41 & 0.04 & 0.01 + 1.0625 & 0.50 & 0.06 & 1.21 & 0.42 & 0.55 & 0.10 & 1.79 & 0.42 & 0.04 & 0.01 + 1.1250 & 0.48 & 0.06 & 1.09 & 0.40 & 0.53 & 0.12 & 1.91 & 0.40 & 0.05 & 0.01 + 1.1875 & 0.48 & 0.06 & 1.17 & 0.43 & 0.56 & 0.11 & 1.83 & 0.43 & 0.04 & 0.01 + 1.2500 & 0.44 & 0.06 & 0.96 & 0.38 & 0.49 & 0.14 & 2.04 & 0.38 & 0.05 & 0.01 + + 1.0000 & 0.85 & 0.03 & 2.64 & 0.25 & 0.68 & 0.02 & 0.36 & 0.25 & 0.09 & 0.01 + 1.0625 & 0.83 & 0.06 & 2.35 & 0.41 & 0.64 & 0.04 & 0.65 & 0.41 & 0.03 & 0.01 + 1.1250 & 0.83 & 0.06 & 2.39 & 0.41 & 0.65 & 0.04 & 0.61 & 0.41 & 0.03 & 0.01 + 1.1875 & 0.86 & 0.06 & 2.42 & 0.40 & 0.64 & 0.04 & 0.58 & 0.40 & 0.03 & 0.01 + 1.2500 & 0.92 & 0.05 & 2.57 & 0.33 & 0.64 & 0.03 & 0.43 & 0.33 & 0.04 & 0.01 + example of the delta variance @xmath51 for oriona .
we show the results for all three available @xmath3 maps ( _ dash - dot _ 25@xmath39 nearest neighbours , _ dotted _ 49@xmath39 nearest neighbour , _ dash - dot - dot _ 100@xmath39 nearest neighbours ) .
the corresponding vertical lines indicate the spatial resolution limit for each map , below which the values are meaningless .
the typical scale of the structure in this cloud can be identified by the peak at about 11pc . for clarity
, error bars are only overplotted on the 49@xmath39 nearest neighbour data .
the weighted correlation coefficient for the determined slope is 0.976.,width=302 ] we applied the @xmath1-variance technique ( as described in sec.[delmet ] ) to all gmcs of our sample . as an example
we show the results of the @xmath51 plot for oriona in fig.[dvex1 ] .
similar plots for all the individual clouds can be found in the appendix[indplt ] .
these plots also contain a cut - out of the 49@xmath39 nearest neighbour @xmath3 map , as well as the structure function of the cloud , determined from this map .
the statistical error bars for the @xmath1-variance in some clouds appear rather large .
this has also been recognised by ossenkopf et al . .
however , they also note that ( quote ) : `` in spite of the large error bars , the general scaling behaviour can be accurately traced . '' since we will measure the mass spectral index scaling exponents @xmath50 in these diagrams we determined the weighted correlation coefficients @xmath69 ( listed in the last column of table[strrestab ] ) .
they show , except for two clouds ( @xmath70-ori , lupus1 , 2 ) that there is a very good correlation , with @xmath69 generally significantly larger than 0.9 .
using our _ con - noise _ maps we investigated the positions of the peaks in the @xmath1-variance plots .
these peaks reveal the scale at which a change in the structure occurs , e.g. the size of the cloud , the length of filaments etc .. these results are shown in table[strrestab ] .
we find peaks in some of the clouds ( cepheus , chamaeleon , @xmath70-ori , oriona , lupus , monoceros , taurus ) while for the remaining clouds no clear - cut peak can be identified .
the identified peaks appear over a range of scales from 1.5pc ( lupus1 and 2 and ophiuchus ) to 16pc ( cepheus ) . using the star density instead of the column density maps
, we find peaks in the @xmath1-variance for the clouds auriga1 , circinus , corona australis , @xmath70-ori , ophiuchus , oriona , perseus , serpens and taurus .
in general , the peaks occur at smaller scales ( from 0.5pc in taurus to 10pc in auriga1 ) .
however , only for four clouds ( @xmath70-ori , ophiuchus , oriona , taurus ) do we detect a clear peak in both maps . in each case
we find that the peak determined when using the column density map is larger , indicating that the star density maps trace smaller scales when the @xmath1-variance technique is applied to them . by means of the _ con - noise _ maps we also measure the mass spectral index scaling exponents @xmath50 for each cloud .
these values and uncertainties are listed in the last two columns of table[strrestab ] .
the listed values are the mean of the results for the three available column density maps .
the values for @xmath50 range from 0.17 ( @xmath70-ori ) to 1.08 ( corona australis ) with a mean for all clouds of [email protected] .
similar to the investigations of the structure function , we determined the influence of noise on our analysis of the cloud structure with the @xmath1-variance method .
the same noise increments as for the structure function analysis were added to the extinction maps and the change to the peak position of the @xmath1-variance and the mass spectral index scaling exponent were determined .
neither the peak position nor the mass spectral index seem to be significantly and/or systematically influenced , if an additional noise of the level discussed in sect.[noistr ] is added to the extinction maps .
this applies to both , the use of our artificially generated image as well as the real cloud .
hence the obtained values can be considered robust .
the determined average densities of the entire gmcs , as well as the densities of the material above the star formation threshold can be found in table[strrestab ] .
the average densities for the entire cloud range from 1.3@xmath71 ( for monoceros ) to 100@xmath71 ( for serpens ) .
the low value for monoceros might be due to small scale structures not being detected because of the distance to the cloud . for the serpens cloud the apparently high density is due to a general offset in the extinction values ( either foreground or background to the cloud ) . when not considering the very low and very high values , we find a mean value for our sample of about 15@xmath71 , which corresponds to 600@xmath72 .
the densities of material above the star formation threshold are naturally higher .
we find a range from 23@xmath71 ( monoceros ) to 6500@xmath71 ( taurus ) .
the median value is about 750@xmath71 or 3@xmath5810@xmath73@xmath72 . here
the values for the individual clouds should be considered with great care , as we find larger values for clouds with smaller distances , indicating the non - detection of higher column density material in more distant clouds ( due to limited spatial resolution ) and thus an underestimate of the density .
to investigate possible influences of cloud structure on the star formation properties we evaluate the current and potential future amount of star formation of our sample .
\i ) in paperii ( e.g. bottom left panel of fig.4 ) we found that our sample of clouds can be divided into two groups when considering their column density and mass distribution in the low extinction regions ( a@xmath0 above 1mag and below the star formation threshold ) . we indicate which cloud belongs to which group in table[strrestab ] .
clouds in group1 have a shallow column density and mass distribution while clouds in group2 have a steeper distribution .
hence , group1 members have a larger amount of material at high column densities , and thus more mass potentially available for star formation .
one finds that the average fraction of mass potentially involved in star formation is about two to three times higher for clouds in group1 ( see table4 in paperii ) .
\ii ) we use literature data to estimate the current number of young stars and mode of star formation for each cloud .
in particular the work by kainulainen et al . and reipurth @xcite have been used .
it turns out that clouds in group1 contain typically more than 100 young stellar objects and the majority of these clouds form at least one cluster of stars .
clouds in group2 on the other hand , typically only form a few tens of stars and no clusters . thus
, group1 contains clouds which intensely form stars , preferably in a clustered mode , while clouds in group2 show weaker star formation activity mostly in a distributed mode .
taurus , without a known cluster of young stars , seems to be an exception as it belongs to group1 . for our structure function analysis we selected only pixel values between the self shielding column density threshold of 1mag @xmath3 and the individual star formation threshold for each cloud ( see paperii ) , as this range represents the part of the clouds where turbulence is expected to dominate the structure .
we calculated the structure functions for each cloud using the three _ con - noise _ maps and a spatial scale range from 0.1pc to 1.0pc ( traceable for all clouds ) . using the data shown in table[strrestab ] we investigated possible correlations between the parameters @xmath1 , @xmath19 against the molecular cloud densities ( @xmath74 and @xmath75 ) as well as a range of other parameters from table4 of paperii .
no correlations could be found .
the analysis was also performed by considering two separate groups of clouds .
there are no significant differences for any of the structure function parameters between the two groups .
hence , our determined structure functions are not able to distinguish between clouds that form stars in clusters or in a distributed mode
. one might expect that the values of the co - dimension @xmath19 increase with distance ( i.e. the fractal dimension @xmath21 is decreasing ) , due to the change in appearance of the cloud .
more distant clouds will appear to have a more simplistic structure due to details not being resolved .
however , in our analysis no such trend could be found .
this can be attributed to the fact that we determine the structure functions in the scale range 0.1pc to 1.0pc , which is resolved for all investigated clouds .
our mean value for all clouds of @[email protected] is much higher than the values of @xmath1=0 expected by k41 .
furthermore , it is inbetween the predictions of @xmath1=2/3 ( sl94 , b02 ) and @xmath1=1.0 ( s08 ) .
however , the scatter in our results means the value could be as low as @xmath1=0.69 or as high as @xmath1=1.01 .
for the parameter @xmath19 our average of [email protected] is close to the sl94 value of @xmath19=2.0 , while b02 theorised a value of @xmath19=1.0 . in s08 @xmath19
is found to be @xmath761.1 for compressive forcing and @xmath761.5 for solenoidal forcing , both below our average value for the co - dimension .
our average is closer to the predicted value for solenoidal forcing ( in particular considering the uncertainties ) .
hence , we conclude that the average cloud of our sample has a structure function hinting to solenoidal forcing of the turbulent field , rather than compressive driving . using the @xmath1-variance technique , we identified peaks in plots of @xmath51 against @xmath5 to quantify the scale of structure in our investigated gmcs .
we were only able to find peaks in @xmath51 for about half the clouds . in the remainder
no peaks could be identified .
this could be caused by the fact that there are no dominant scales in those clouds , or that the size of the map around each cloud was too small ( due to neighbouring clouds ) .
note that the @xmath1-variance will only detect dominant scales if they are significantly smaller than the map size .
scales identified average at about 5pc .
these generally correspond to the width of a filament or sheets in the cloud rather than the length or size of the entire cloud . in the region lupus3 , 4 , 5
, 6 the peaks correspond to the length of the different filaments .
when we perform the @xmath1-variance analysis with the star density maps , dominant scales are found as well for about half the clouds .
these scales are smaller and average at about 2pc .
this indicates that the star density maps trace more compact , higher extinction regions , generally the width of dense filaments . for the mass spectral index scaling exponent @xmath50 , federrath et al .
found values of @xmath50=0.55 for solenoidal forcing of turbulence and @xmath50=1.34 for compressive driving . we find a considerable variation in the measured values for @xmath50 amongst our clouds .
all values are below the predicted amount for pure compressive driving . however , the mean for all clouds is @[email protected] , hence very close to the predictions for pure solenoidal driving of turbulence .
when considering the average values of @xmath50 for our two groups of clouds , significant differences can be found . while the cluster forming clouds have an average of @[email protected] , the clouds with a more distributed star formation mode have a mean of @[email protected] ( or , when excluding the clouds with low correlation coefficient , @[email protected] ) .
based on the predictions from federrath et al . , the larger value of @xmath50 for the cluster forming clouds indicates an enhanced component of compressive driving compared to the clouds with a more distributed mode of star formation .
we refrain from trying to estimate a more quantitative statement about the amount of contribution from compressive driving in the cluster forming clouds , since the uncertainties and scatter in our data are clearly large .
we present the homogeneous investigation of a selection of 16 nearby giant molecular clouds .
utilising near infrared extinction maps made from 2mass data with the near infrared colour excess technique , we determine structure functions and perform a @xmath1-variance analysis .
the same clouds as in froebrich & rowles @xcite were investigated .
they are all reasonably nearby , away from the galactic plane , and to the best of our knowledge there is only one cloud along each line of sight . using our extinction maps
we identify two groups of clouds based on the slope of their column density and mass distributions of the turbulence dominated ( low a@xmath0 ) part of the cloud .
we find that clouds with shallower mass distributions form stars preferably in a clustered mode , while clouds with steep mass distributions form fewer stars and show a more distributed mode of star formation . structure functions determined for all clouds homogeneously within a range from 0.1pc to 1.0pc can not be used to distinguish clouds with different star formation modes .
comparing the structure function parameters to model calculations suggests that the turbulence in the investigated clouds is governed preferably by solenoidal forcing .
our results of the @xmath1-variance analysis also indicate that the average cloud in our sample is governed by solenoidal forcing .
however , clouds which form a large number of stars in clusters have an enhanced component of compressive driving of the turbulent field , in comparison to clouds with isolated star formation .
hence , compressive driving seems to lead to a more clustered mode of star formation . to quantify these qualitative findings
, i.e. to determine the fraction of solenoidal and compressive driving in each cloud , a more detailed comparison of the numerical simulations with the observational data needs to be performed .
in particular viewing angle and resolution / distance effects in the model data need to be investigated .
jr acknowledges a university of kent scholarship . 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 .
anselmet , f. , gagne , y. , hopfinger , e. j. , antonia , r. a 1984 , j.fluid mech , 140 , 63 benzi , r. , ciliberto , s. , tripiccione , r. , baudet , c. 1993 , phrve , 48 , 29 boldyrev , s. 2002 , , 569 , 841 brandenburg , a. and nordlund , a. 2009 in reports on progress in physics , 74 , 1 dubrulle , b. 1994 , , 73 , 959 elmegreen , b. g. & scalo , j. 2004 , , 42 , 211 federrath , c. , roman - duval , j. , klessen , r. s. 2010 , , 512a , 81f froebrich , d. and rowles , j. 2010 , , 406 , 1350f goodman , a. a. , pineda , j. e. , schnee , s. l. 2009 , , 692 , 91 kainulainen , j. , beuther , h. , henning , t. , plume , r. 2009 , , 508 , 35 kolmogorov , a. 1941 , dokl.akad.nauk sssr , 30 , 301 lada , c. j. , lada , e. a. , clemens , d. p. , bally , j. 1994 , 429 , 694l lombardi , m. , lada , c. j. , alves , j. 2008 , , 489 , 143 ossenkopf , v. , krips , m. , stutzki , j. 2008a , , 485 , 719 ossenkopf , v. , krips , m. , stutzki , j. 2008b , , 485 , 917 padoan , p. , cambrsy , l. , langer , w. 2002 , , 580 , 57 padoan , p. , boldyrev , s. , langer , w. , nordlund , , 2003 , , 583 , 308 reipurth , b. , 2008a , handbook of star forming regions , volume i : the northern sky asp monograph publications , vol .
4 . edited by bo reipurth reipurth , b. , 2008b , handbook of star forming regions , volume ii : the southern sky asp monograph publications , vol . 5 . edited by bo reipurth rowles , j. & froebrich , d. 2009 , , 395 , 1640 scalo , j. & elmegreen , b. g. 2004 , , 42 , 275 schmidt , w. , federrath , c. , klessen , r. 2008 , , 101 , 4505 she , z .- s . &
leveque , e. 1994 , , 72 , 336 skrutskie , m. f. , cutri , r. m. , et al . , 2006 , , 131 , 1163s stutzki , j. , bensch , f. , heithausen , a. , ossenkopf , v. 1998 , , 336 , 697
ticks in long numbered from -12 to -3 by 5 short unlabeled from -12 to -3 by 1 / right label ticks in long unlabeled from -12 to -3 by 5 short unlabeled from -12 to -3 by 1 / bottom label l [ ] ticks in long numbered from 156 to 173 by 5 short unlabeled from 156 to 173 by 1 / top label ticks in long unlabeled from 156 to 173 by 5 short unlabeled from 156 to 173 by 1 / at 171 -10 at 171 -11 at 140 . -8 . | we analyse extinction maps of nearby giant molecular clouds to forge a link between driving processes of turbulence and modes of star formation .
our investigation focuses on cloud structure in the column density range above the self shielding threshold of 1mag a@xmath0 and below the star formation threshold the regime in which turbulence is expected to dominate .
we identify clouds with shallow mass distributions as cluster forming .
clouds that form stars in a less clustered or isolated mode show a steeper mass distribution .
structure functions prove inadequate to distinguish between clouds of different star formation mode .
they may , however , suggest that the turbulence in the average cloud is governed by solenoidal forcing .
the same is found using the @xmath1-variance analysis which also indicates that clouds with a clustered mode of star formation show an enhanced component of compressive driving in the turbulent field .
thus , while star formation occurs in each cloud , independent of the turbulent driving mechanism , compressive forcing appears to be associated with the formation of stellar clusters .
[ firstpage ] star formation extinction
ism : clouds ism : dust ism : molecules |
this article deals with the relationship between zeros and critical points of random polynomials in one complex variable . recall that a _ critical point _ of a polynomial @xmath10 is a root of its derivative @xmath11
there are many results concerning the location of critical points of polynomials whose roots are known .
one of the most famous examples is the gauss
lucas theorem , which offers a geometric connection between the roots of a polynomial and the roots of its derivative .
[ thm : gauss ] if @xmath10 is a non - constant polynomial with complex coefficients , then all zeros of @xmath11 belong to the convex hull of the set of zeros of @xmath10 .
there are many refinements of theorem [ thm : gauss ] ; we refer the reader to @xcite and references therein . a probabilistic version of the problem was first studied by pemantle and rivin @xcite .
specifically , pemantle and rivin raised the following question . for a random polynomial @xmath10 , when are the zeros of @xmath11 stochastically similar to the zeros of @xmath10 . before introducing their results
, we fix the following notation . for a polynomial @xmath10 of degree @xmath1
, we define the empirical measure constructed from the roots of @xmath10 as @xmath12 where each root in the sum is counted with multiplicity and @xmath13 is the unit point mass at @xmath14 . in particular , when @xmath10 is a random polynomial , @xmath15 becomes a random probability measure .
for the critical points of @xmath10 , we introduce the notation @xmath16 in other words , @xmath17 is the empirical measure constructed from the critical points of @xmath10 .
let @xmath18 be independent and identically distributed ( iid ) random variables taking values in @xmath19 , and let @xmath2 be their common probability distribution .
for each @xmath20 , consider the polynomial @xmath21 under the assumption that @xmath2 has finite one - dimensional energy , pemantle and rivin @xcite show that @xmath22 converges weakly to @xmath2 as @xmath1 tends to infinity .
let us recall what it means for a sequence of random probability measures to converge weakly .
let @xmath23 be a topological space ( such as @xmath24 or @xmath19 ) , and let @xmath25 be its borel @xmath26-field .
let @xmath27 be a sequence of random probability measures on @xmath28 , and let @xmath2 be a probability measure on @xmath28 .
@xmath29 converges weakly to @xmath2 in probability _ as @xmath30 ( and write @xmath31 in probability ) if for all bounded continuous @xmath32 and any @xmath33 , @xmath34 in other words , @xmath31 in probability as @xmath30 if and only if @xmath35 in probability for all bounded continuous @xmath36 .
similarly , we say _
@xmath29 converges weakly to @xmath2 almost surely _ as @xmath30 ( and write @xmath31 almost surely ) if for all bounded continuous @xmath32 , @xmath37 almost surely .
kabluchko @xcite generalized the results of pemantle and rivin to the following .
[ thm : pmk ] let @xmath2 be an arbitrary probability measure on @xmath19 , and let @xmath38 be a sequence of iid random variables with distribution @xmath2 . for each @xmath20 ,
let @xmath0 be the degree @xmath1 polynomial given in .
then @xmath22 converges weakly to @xmath2 in probability as @xmath30 .
subramanian , in @xcite , verified a special case of theorem [ thm : pmk ] when @xmath2 is supported on the unit circle in the complex plane .
naturally , one may ask whether the assumptions in theorem [ thm : pmk ] ( such as the roots @xmath38 being independent ) can be relaxed . in @xcite ,
the first author managed to prove a version of theorem [ thm : pmk ] for random polynomials with dependent roots provided the roots lie on the unit circle and satisfy a number of technical conditions .
in particular , the results in @xcite apply to characteristic polynomials of random unitary matrices and other matrices from the classical compact groups ( the eigenvalues of such matrices are known to not be independent ) .
similar results for characteristic polynomials of nearly hermitian matrices were studied in ( * ? ? ?
* section 2.5 ) . in @xcite
, reddy considers polynomials whose zeros are chosen randomly from two deterministic sequences of complex numbers in which the empirical measures for both sequences converge to the same limit .
it is shown that the limiting empirical measure of the zeros and critical points agree for these polynomials , yielding a version of theorem [ thm : pmk ] where the randomness can be reduced and independence still remains . however , as the following example shows , the randomness in theorem [ thm : pmk ] can not be completely eliminated ( i.e. , the theorem does not always hold for sequences of deterministic polynomials ) .
let @xmath39 .
then the roots of @xmath0 are the @xmath1-th roots of unity , and so @xmath40 converges weakly to the uniform measure on the unit circle as @xmath1 tends to infinity .
however , all @xmath41 critical points of @xmath0 are located at the origin . hence , @xmath42 for all @xmath1 .
we use asymptotic notation ( such as @xmath43 ) under the assumption that @xmath30 .
in particular , @xmath44 , @xmath45 , @xmath46 , and @xmath47 denote the estimate @xmath48 , for some constant @xmath49 independent of @xmath1 and for all @xmath50 .
if we need the constant @xmath51 to depend on another constant , e.g. @xmath52 , we indicate this with subscripts , e.g. @xmath53 , @xmath54 , @xmath55 , and @xmath56 .
we write @xmath57 if @xmath58 for some @xmath59 that goes to zero as @xmath30 .
specifically , @xmath60 denotes a term which tends to zero as @xmath30 .
to introduce our results , we first consider a special case of the polynomial @xmath0 , defined in , when @xmath2 is the uniform probability measure on the unit circle centered at the origin . in this case ,
theorem [ thm : pmk ] implies that @xmath61 converges weakly in probability to @xmath2 as @xmath30 .
a numerical simulation of this result is shown in figure [ fig : iidcirc ] ; as can be seen , all critical points of @xmath62 lie very close to the unit circle . on the other hand ,
if we consider the polynomial @xmath63 for some deterministic point @xmath3 outside the unit circle , we see in figure [ fig : iidcircdetout ] that one of the critical points leaves the unit disk and lies very close to @xmath3 . however , the remaining critical points still lie close to the unit circle .
the goal of this note is to describe the pairing between the root @xmath3 and the nearby critical point .
more generally , we consider the case when several deterministic zeros are appended to the random polynomial @xmath0 and when @xmath2 is an arbitrary measure in the complex plane with compact support ( not just the uniform distribution on the unit circle ) .
see , for example , figures [ fig : iidcirc3detout ] and [ fig : distinguished : iidblob ] .
let us mention that this pairing phenomenon between roots and critical points has been observed previously for random polynomials .
hanin @xcite proves a similar pairing result when a number of deterministic roots are appended to a random polynomial whose roots are chosen independently from a probability measure @xmath2 on the riemann sphere .
hanin s proof is guided by an intuitive electrostatic interpretation of the zeros and critical points .
unlike the results in @xcite however , our results do not require the measure @xmath2 to have bounded density or require the deterministic roots to satisfy a separation condition .
in addition , our methods are significantly different than those used in @xcite and allow us to describe the exact number of critical points lying in a region outside the support of @xmath2 . in a separate paper @xcite , hanin considers the joint distribution of roots and critical points for a class of gaussian random polynomials .
however , the polynomials considered in @xcite are quite different than the model considered in this paper .
finally , let us mention the work of dennis and hannay @xcite from the physics literature , which gives an electrostatic explanation for the pairing of critical points and zeros of random polynomials and characteristic polynomials of random matrices . to begin , we first consider the analogue of theorem [ thm : pmk ] when @xmath64 deterministic zeros are appended to the random polynomial @xmath0 in .
[ thm : genkabluchko ] let @xmath2 be an arbitrary probability measure on @xmath19 , and suppose @xmath38 are iid random variables with distribution @xmath2 . for each @xmath20 , let @xmath65 be a deterministic non - negative integer no larger than @xmath1 such that @xmath66 .
in addition , let @xmath67 be a deterministic triangular array of complex values , and let @xmath68 then @xmath61 converges weakly to @xmath2 in probability as @xmath30 .
theorem [ thm : genkabluchko ] is a generalization of theorem [ thm : pmk ] . indeed ,
theorem [ thm : pmk ] can be recovered from theorem [ thm : genkabluchko ] by taking @xmath69 . unsurprisingly , we prove theorem [ thm : genkabluchko ] in appendix [ sec : genkabluchko ] by slightly generalizing the methods developed by kabluchko in @xcite .
let us discuss the intuition behind theorem [ thm : genkabluchko ] .
to do so , we must begin with theorem [ thm : pmk ] . roughly speaking , theorem
[ thm : pmk ] describes the phenomenon that if @xmath0 is a degree @xmath1 random polynomial , then @xmath70 in probability as @xmath30 . in other words ,
the limiting behavior of the critical points is the same as the limiting behavior of the roots . while theorem [ thm : pmk ] only applies to random polynomials with iid roots ,
the same phenomenon has been observed for other ensembles of random polynomials @xcite , and numerical simulations show that it should be true for many other models .
stated another way , the behavior in appears to be universal among random polynomials .
let us now consider the polynomial @xmath0 from theorem [ thm : genkabluchko ] .
it follows from the law of large numbers that @xmath71 weakly almost surely as @xmath30 since @xmath72 . therefore ,
if the convergence in applies to the polynomial @xmath0 , the triangle inequality would immediately imply that @xmath61 also converges weakly to @xmath2 in probability .
this heuristic is the basis for our proof of theorem [ thm : genkabluchko ] .
the above heuristic also hints that the condition @xmath66 in theorem [ thm : genkabluchko ] is sharp . indeed ,
if @xmath73 deterministic roots were to be appended , the limiting distribution is , in general , not @xmath2 as shown by the following example .
let @xmath74 and @xmath75 .
define @xmath76 where @xmath38 are iid random variables uniformly distributed on the unit circle centered at the origin in the complex plane . then , by theorem [ thm : pmk ] , @xmath22 converges weakly to the uniform measure on the unit circle in probability as @xmath30 .
however , the polynomial @xmath77 has at least @xmath78 critical points at the origin . in particular
, @xmath79 for @xmath1 sufficiently large . among other things
, this implies that @xmath80 does not converge weakly to the uniform probability measure on the unit circle as @xmath30 .
while theorem [ thm : genkabluchko ] shows that the global behavior of the critical points is unchanged by the addition of @xmath64 deterministic roots , the addition of one or more deterministic roots can create a number of outlying critical points as illustrated in figures [ fig : iidcircdetout ] and [ fig : iidcirc3detout ] .
one way of viewing this phenomenon is to view the deterministic roots as a small perturbation of the original polynomial .
this small perturbation is not enough to change the global distribution of the critical points ; it may , however , as observed in the figures above , create a small number of outlying critical points .
our main results below describe these outliers . before we consider the perturbed model , we first consider the case when there are no deterministic roots . in this initial case ,
we want to determine exactly where the critical points of the random polynomial @xmath0 , defined in , are located .
this way , when we do append the small perturbation of deterministic roots , we will be able to tell exactly what effect the perturbation has had .
let @xmath2 be a probability measure on @xmath19 , and suppose @xmath81 are iid random variables with distribution @xmath2 . in view of the gauss
lucas theorem ( theorem [ thm : gauss ] ) , the roots of @xmath82 , must lie in @xmath83 , the convex hull of the support of @xmath2 .
however , as we discussed above in the case when @xmath2 is supported on the unit circle ( shown in figure [ fig : iidcirc ] ) , nearly all of the critical points appear near the support of @xmath2 , which is only a small subset of the convex hull .
thus , our goal is to determine the exact subset of @xmath83 where the critical points will lie , with high probability .
we do so in the theorem below . to define this set where the critical points are located
, we will first need to introduce the cauchy
stieltjes transform .
let @xmath2 be a probability measure on @xmath19 , and let @xmath84 be the cauchy
stieltjes transform of @xmath2 defined by @xmath85 also , define @xmath86 to be the set of zeros of @xmath84 . if @xmath2 has compact support , it turns out that @xmath87 ; see proposition [ prop : mmu ] for details . for @xmath33
, we also define the set @xmath88 to be the @xmath89-neighborhood of @xmath90 . here , @xmath91 is the distance from @xmath92 to a set @xmath93 .
the following theorem shows that all critical points of @xmath0 must lie inside @xmath94 with high probability .
[ thm : nooutliers ] let @xmath2 be a probability measure on @xmath19 with compact support , and suppose @xmath81 are iid random variables with distribution @xmath2 .
then , for every @xmath33 , there exists @xmath95 ( depending only on @xmath2 and @xmath89 ) such that , with probability at least @xmath96 , the polynomial @xmath97 has no critical points outside @xmath98 . by the gauss
lucas theorem ( theorem [ thm : gauss ] ) , the critical points of @xmath0 must lie inside @xmath83 .
thus , theorem [ thm : nooutliers ] actually reveals that , with high probability , @xmath0 has no critical points outside @xmath99 .
we now justify our choice of the set @xmath94 as the correct location of the critical points .
first , in the case that @xmath2 is degenerate , @xmath100 for some @xmath101 , which has critical point @xmath102 with multiplicity @xmath41 .
this example shows that clearly the critical points of @xmath0 may lie in @xmath103 .
the next example shows that the critical points can also be in a neighborhood of the zero set @xmath104 .
let @xmath105 for some @xmath106 with @xmath107 and @xmath108 , and assume @xmath38 are iid random variables with distribution @xmath2 .
then @xmath109 for some non - negative integers @xmath110 with @xmath111 .
almost surely , for @xmath1 sufficiently large , @xmath112 , and , in this case , @xmath113 thus , by the law of large numbers , @xmath0 has a critical point at @xmath114 almost surely . on the other hand
, @xmath115 has exactly one zero located at @xmath116 . by the borel
cantelli lemma , theorem [ thm : nooutliers ] immediately implies the following corollary .
let @xmath2 be a probability measure on @xmath19 with compact support , and suppose @xmath38 are iid random variables with distribution @xmath2 .
fix @xmath33 .
then , almost surely , for @xmath1 sufficiently large , the polynomial @xmath97 has no critical points outside @xmath98 .
we now consider the outlying critical points depicted in the figures above .
to do so , we will need the following notation . for a polynomial @xmath117 of degree @xmath1 , we let @xmath118 be the critical points of @xmath117 counted with multiplicity . [
thm : simple ] let @xmath2 be a probability measure on @xmath19 with compact support , and suppose @xmath38 are iid random variables with distribution @xmath2 .
let @xmath119 , and assume @xmath7 are deterministic complex numbers ( which do not depend on @xmath1 ) ; in addition , suppose there are @xmath120 values @xmath121 not in @xmath90 .
then , there exists @xmath122 such that the following holds for any fixed @xmath123 . almost surely , for @xmath1 sufficiently large , there are exactly @xmath120 critical points ( counted with multiplicity ) of the polynomial @xmath124 outside @xmath98 , and after labeling these critical points correctly , @xmath125 for each @xmath126 .
theorem [ thm : simple ] describes exactly the phenomenon we observe in the figures above .
in particular , this theorem shows that each deterministic root outside @xmath90 creates one outlying critical point , which is asymptotically close to the deterministic root . in figure [ fig : iidcircdetin ] , we give the results of a numerical simulation when @xmath2 is the uniform probability measure on the unit circle centered at the origin , @xmath127 , and @xmath128 . as predicted by theorem [ thm : simple ] , the figure shows that there is exactly one critical point near the outlier @xmath129 .
for comparison , we provide the following example which shows that theorem [ thm : simple ] fails for deterministic polynomials .
let @xmath130 and @xmath131 .
then the roots of @xmath9 are @xmath132-th roots of unity with an outlier at @xmath133 .
however , we will show that @xmath9 has no critical points near @xmath133 .
indeed , @xmath134 and so the critical points are the solutions of @xmath135 for @xmath136 , we have @xmath137 for @xmath1 sufficiently large .
this implies that @xmath138 for every @xmath92 with @xmath136 .
hence , for @xmath1 sufficiently large , there are no critical points of @xmath9 in the disk @xmath139 .
more generally , this argument shows that for a fixed @xmath140 , there are no critical points of @xmath9 in the disk @xmath141 for sufficiently large @xmath1 .
we next state two generalizations of theorem [ thm : simple ] .
both results deal with the case when the deterministic points @xmath7 ( as well as the integer @xmath142 ) are allowed to depend on @xmath1 . because the points can now depend on @xmath1 , some additional technical assumptions are required .
these technical assumptions are trivially satisfied when @xmath7 do not depend on @xmath1 . as such , theorem [ thm : simple ]
is actually a corollary of the following more general result .
[ thm : main ] let @xmath2 be a probability measure on @xmath19 with compact support , and suppose @xmath38 are iid random variables with distribution @xmath2 . for each @xmath20 ,
let @xmath67 be a triangular array of deterministic complex numbers with @xmath143 , and assume @xmath144 fix @xmath33 , and suppose that for all sufficiently large @xmath1 , there are no values of @xmath67 in @xmath145 and there are @xmath120 values @xmath146 outside @xmath147 .
then , almost surely , for @xmath1 sufficiently large , there are exactly @xmath120 critical points ( counted with multiplicity ) of the polynomial @xmath148 outside @xmath149 , and after labeling these critical points correctly , @xmath150 for each @xmath126 .
the @xmath151-magnitude assumption in is required for our proof . however , we conjecture that this condition is not needed .
in fact , in the case when @xmath152 , we can remove this assumption , and we obtain the following stronger result .
[ thm : distinguished ] let @xmath2 be a probability measure on @xmath19 with compact support , and suppose @xmath38 are iid random variables with distribution @xmath2 .
for each @xmath20 , let @xmath67 be a triangular array of deterministic complex numbers with @xmath143 .
fix @xmath33 , and suppose that for all sufficiently large @xmath1 , there are no values of @xmath67 in @xmath145 and there is one value @xmath153 outside @xmath147 . then , almost surely , for @xmath1 sufficiently large , there is exactly one critical point of the polynomial @xmath148 outside @xmath149 , and after labeling the critical points correctly , @xmath154 if @xmath155 , then implies that , almost surely , @xmath156 more generally , if @xmath157 , theorem [ thm : distinguished ] yields that , almost surely , @xmath158 in other words , the location of the outlying critical point @xmath159 is asymptotically close to the outlying root @xmath153 .
we present a numerical simulation of theorem [ thm : distinguished ] in figure [ fig : distinguished : iidblob ] .
the rest of the paper is devoted to the proof of our main results . in section [ sec : tools ] , we develop several tools we will need for the proofs .
the proof of theorem [ thm : nooutliers ] is presented in section [ sec : proof : nooutliers ] .
we prove theorems [ thm : simple ] , [ thm : main ] , and [ thm : distinguished ] in section [ sec : proofs ] .
finally , the proof of theorem [ thm : genkabluchko ] is given in appendix [ sec : genkabluchko ] .
we present here some tools we will need to prove our main results . we will need the following complex - valued version of hoeffding s inequality .
[ lemma : hoeffding ] let @xmath160 be iid complex - valued random variables which satisfy @xmath161 almost surely for some @xmath162 . then there exist absolute constants @xmath163 such that @xmath164 \right| \geq t \right ) \leq c \exp \left ( - c n t^2
/ k^2 \right)\ ] ] for every @xmath165 .
let @xmath166.\ ] ] if @xmath167 , then @xmath168 or @xmath169 .
so , we have @xmath170 the claim now follows from the classic ( real - valued ) version of hoeffding s inequality ( see @xcite ) since @xmath171 and @xmath172 .
we introduce @xmath173-nets as a convenient way to discretize a compact set .
let @xmath174 be a subset of @xmath19 , and @xmath33 .
a subset @xmath175 of @xmath174 is called an @xmath89-net of @xmath174 if every point @xmath176 can be approximated within @xmath89 by some point @xmath177 , i.e. so that @xmath178 . for a finite set @xmath175
, we let @xmath179 denote the cardinality of @xmath175 . we will need the following estimate for the size of an @xmath89-net .
[ lemma : net ] let @xmath180 be a compact subset of @xmath181 for some @xmath182 . then
, for every @xmath33 , there is an @xmath89-net @xmath175 of @xmath180 such that @xmath183 let @xmath184 be a maximal @xmath185-separated subset of @xmath186 . in other words , @xmath184 is such that @xmath187 for all @xmath188 with @xmath189 , and no subset of @xmath190 containing @xmath184 has this property .
such a set can always be constructed by starting with an arbitrary point in @xmath190 and at each step selecting a point that is at least @xmath185 distance away from those already selected .
since @xmath190 is compact , this procedure will terminate after a finite number of steps .
the maximality property implies that @xmath184 is an @xmath185-net of @xmath190 .
indeed , otherwise there would exist @xmath191 that is at least @xmath185-far from all points in @xmath184 .
so @xmath192 would still be an @xmath185-separated set , contradicting the maximality property above .
moreover , the separation property implies that the balls of radii @xmath193 centered at the points in @xmath184 are disjoint .
in addition , all such balls lie in the ball of radius @xmath194 centered at the origin .
comparing areas gives @xmath195 and hence @xmath196 we now use @xmath184 to construct an @xmath89-net of @xmath180 .
indeed , we construct @xmath175 iteratively using the following procedure .
let @xmath197 be an enumeration of the points in @xmath184 , and set @xmath198 . given @xmath199 for @xmath200 ,
we construct @xmath201 as follows : 1 .
if the ball of radius @xmath185 centered at @xmath202 does not intersect @xmath180 , then let @xmath203 .
2 . if the ball of radius @xmath185 centered at @xmath202 does intersect @xmath180 , let @xmath204 be an element of the intersection and set @xmath205
. now take @xmath206 . by the procedure above , it follows that @xmath207 .
it remains to show that @xmath175 is an @xmath89-net of @xmath180 .
let @xmath208 .
since @xmath209 , there exists @xmath210 such that @xmath211 .
this means that the ball of radius @xmath185 centered at @xmath212 intersects @xmath180 .
thus , from the procedure above , there exists @xmath177 such that @xmath213 .
therefore , by the triangle inequality , @xmath214 .
we will need the following companion matrix result , which describes a matrix whose eigenvalues are the critical points of a given polynomial .
this result appears to have originally been developed in @xcite ( see ( * ? ? ?
* lemma 5.7 ) ) .
however , the same result was later rediscovered and significantly generalized by cheung and ng @xcite .
[ thm : companion ] let @xmath215 for some complex numbers @xmath216 , and let @xmath180 be the diagonal matrix @xmath217 . then @xmath218 where @xmath219 is the @xmath220 identity matrix and @xmath221 is the @xmath220 all - one matrix .
theorem [ thm : companion ] allows us to translate the problem of studying critical points to a problem involving the eigenvalues of certain matrices . for studying the eigenvalues of such matrices
, we will need the following lemmata .
[ lemma : block ] suppose @xmath222 , and @xmath180 are matrices of dimension @xmath220 , @xmath223 , @xmath224 and @xmath225 , respectively .
if @xmath226 is invertible , then @xmath227 the conclusion follows immediately from the decomposition @xmath228 where @xmath229 and @xmath230 are the identity matrices of dimension @xmath220 and @xmath225 , respectively . a similar proof is given in ( * ? ?
* section 0.8.5 ) .
[ lemma : sm ] suppose @xmath226 is an invertible matrix and @xmath231 are column vectors . if @xmath232 , then @xmath233 lemma [ lemma : sm ] can be found in @xcite ; see also ( * ? ? ?
* section 0.7.4 ) for a more general version of this identity known as the sherman morrison
woodbury formula .
we will also require the following bound involving the difference of two determinants . for a matrix @xmath226 , we let @xmath234 denote the spectral norm of @xmath226 , i.e. , @xmath234 is the largest singular value of @xmath226 .
[ lemma : detdiff ] let @xmath226 and @xmath235 be @xmath236 matrices . if @xmath237 , then @xmath238 by the leibniz formula for the determinant , it follows that @xmath239 where the sums range over all permutations @xmath26 of @xmath240 and @xmath241 is the sign of the permutation @xmath26 .
we now take advantage of the fact that the spectral norm of a matrix bounds the magnitude of each entry . in particular ,
@xmath242 and @xmath243 thus , by multiple applications of the triangle inequality , we obtain @xmath244 uniformly in @xmath26 . combining this bound with completes the proof .
we collect here some additional tools and facts we will need .
first , we note that if @xmath2 has compact support , then the convex hull of the support of @xmath2 is also a compact set ; see ( * ? ? ?
* corollary 5.33 ) for details .
the following proposition shows that the zero set of the cauchy stieltjes transform of @xmath2 must lie inside the convex hull of the support of @xmath2 .
[ prop : mmu ] let @xmath2 be a probability measure on @xmath19 with compact support . if @xmath245 for some @xmath246 , then @xmath247 .
let @xmath248 , and define @xmath249 suppose @xmath250 .
then @xmath251 for any @xmath252 .
since @xmath103 is compact , it follows from ( * ? ? ?
* corollary 5.33 ) that @xmath253 is also compact .
thus , by the hyperplane separation theorem , there exists a pair of parallel lines , separated by a gap @xmath33 , separating @xmath253 and @xmath254 .
let @xmath255 be the angle these lines make with the real axis ( if they do not meet the real axis take @xmath256 ) .
then @xmath257 is of the same sign for all @xmath258 and @xmath259 for all @xmath258 .
thus , we obtain @xmath260 as @xmath103 is compact , there exists @xmath261 such that @xmath262 for all @xmath258 .
hence , we conclude that @xmath263 and the proof is complete .
we will also need the following observation concerning the translation of roots and critical points .
[ prop : translate ] let @xmath117 be a monic polynomial of degree @xmath1 , and suppose @xmath264 are the critical points of @xmath117 counted with multiplicity .
then , for any @xmath101 , the critical points of @xmath265 are @xmath266 . since @xmath117 is a monic polynomial of degree @xmath1 , @xmath267 thus , @xmath268 and the claim follows .
this section is devoted to the proof of theorem [ thm : nooutliers ] . for @xmath33 , define @xmath269 to be the @xmath89-neighborhood of the support of @xmath2 .
we begin with the following concentration inequality .
[ lemma : lln ] let @xmath2 be a probability measure on @xmath19 with compact support , and suppose @xmath81 are iid random variables with distribution @xmath2 . then , for every @xmath270 , @xmath271 for some absolute constants @xmath272 .
let @xmath270 , and define @xmath273 we assume @xmath180 is nonempty as the conclusion is trivial otherwise . let @xmath175 be an @xmath274-net of @xmath180 .
by lemma [ lemma : net ] , @xmath175 can be chosen so that @xmath275 we observe that @xmath276 almost surely for every @xmath277 .
thus , almost surely , for @xmath208 , @xmath278 hence , for @xmath279 , @xmath280 in other words , the function @xmath281 is almost surely lipschitz continuous on @xmath180 with lipschitz constant @xmath282 .
similarly , for @xmath283 , @xmath284 suppose @xmath285 .
as @xmath286 and @xmath84 are both continuous on the compact set @xmath180 , there exists @xmath208 such that @xmath287 . since @xmath175 is an @xmath274-net of @xmath180 , there exists @xmath288 such that @xmath289 .
so , by the reverse triangle inequality and the fact that the @xmath286 and @xmath84 are lipschitz continuous , we have @xmath290 therefore , by the union bound , we conclude that @xmath291 as @xmath292 for @xmath208 , hoeffding s inequality ( lemma [ lemma : hoeffding ] ) and the bound in imply that @xmath293 for some absolute constants @xmath95 .
thus , combining , , and yields @xmath294 as desired .
let @xmath33 .
with probability one , @xmath276 for each @xmath277 .
thus , the zeros of @xmath295 outside of @xmath94 are exactly the critical points of @xmath0 outside of @xmath94
. we will show that @xmath296 has no zeros in @xmath297 .
the claim then follows immediately since , by the gauss
lucas theorem ( theorem [ thm : gauss ] ) , all the critical points of @xmath0 lie in @xmath83 .
since @xmath2 has compact support , @xmath83 is also a compact set ( see ( * ? ? ?
* corollary 5.33 ) ) , and hence @xmath180 is compact . as @xmath84 is a continuous function on @xmath180 , @xmath298 achieves its minimum on @xmath180 , which , by definition of @xmath94 can not be zero ( since @xmath94 contains the zero set @xmath299 ) .
thus , there exists @xmath300 such that @xmath301 since @xmath180 is compact , there exists @xmath261 ( depending only on @xmath103 ) such that @xmath302 for all @xmath208 .
thus , by lemma [ lemma : lln ] ( taking @xmath303 ) , we obtain @xmath304 for some absolute constants @xmath305 . hence , on the complementary event , we have @xmath306 for all @xmath208 . since the constants @xmath307 and @xmath308 only depend on @xmath2 and @xmath89 , the proof is complete .
the proof of theorem [ thm : genkabluchko ] presented here is modeled after kabluchko s proof of ( * ? ? ?
* theorem 1.1 ) .
we note that theorem [ thm : genkabluchko ] does not follow from the results in @xcite , and the notable difference between our proof and the one given in @xcite is that we must control the additional contribution coming from the deterministic triangular array . for convenience ,
we use @xmath29 and @xmath419 to mean @xmath40 and @xmath420 , respectively and define @xmath421 to be the collection of values present in the deterministic triangular array .
we let @xmath422 represent lebesgue measure on @xmath423 , and we denote the positive and negative parts of the real logarithm by @xmath424 for @xmath425 .
we use the convention that @xmath426 so that @xmath427 is a function taking values in the extended real line . under the same hypotheses as in theorem [ thm : genkabluchko ] and with the additional assumption that there is a set @xmath428 of lebesgue measure zero for which @xmath429 implies @xmath430 it follows that @xmath431 converges weakly to @xmath2 in probability as @xmath30 .
[ thm : kabl : detcond ] unfortunately , we can not always guarantee that the deterministic array satisfies condition . to get around this issue
, we will work on subsequences where the condition does hold ; specifically , the proof of theorem [ thm : genkabluchko ] will require the following corollary of theorem [ thm : kabl : detcond ] .
assume the same hypotheses as in theorem [ thm : genkabluchko ] and , in addition , suppose @xmath432 is a subsequence of @xmath29 for which there is a set @xmath433 of zero lebesgue measure such that @xmath429 implies @xmath434 then @xmath435 converges weakly to @xmath2 in probability as @xmath30 .
[ cor : kabl : subseq ] we show that @xmath432 is a subsequence of a new sequence of random measures ( modified from @xmath29 ) for which condition does hold . to this end , define the sequence @xmath436 by @xmath437 and the random polynomial @xmath438 also let @xmath439 and @xmath440 denote @xmath441 and @xmath442 , respectively . by construction , @xmath432 and @xmath435 are subsequences of @xmath439 and @xmath440 , respectively .
now , @xmath443 , and for @xmath429 , @xmath444 thus , theorem [ thm : kabl : detcond ] implies that @xmath440 converges weakly to @xmath2 in probability as @xmath30 .
it follows that the subsequence @xmath435 also converges to @xmath2 weakly in probability as @xmath445 .
let @xmath29 be a sequence of random probability measures on @xmath423 , and suppose @xmath2 is a deterministic probability measure on @xmath423 .
then , @xmath29 converges weakly to @xmath2 in probability if and only if each subsequence of @xmath29 contains a further subsequence that converges weakly to @xmath2 in probability .
[ lma : kabl : subcriterion ] observe that , for each bounded and continuous function @xmath446 , the sequence @xmath447 is a sequence of complex - valued random variables whose subsequences are of the form @xmath448 , where @xmath432 is a subsequence of @xmath29 . in addition
, @xmath449 is a constant .
thus , the claim follows by applying theorem 2.6 on page 20 of @xcite to the random variables @xmath447 .
we now prove theorem [ thm : genkabluchko ] by way of corollary [ cor : kabl : subseq ] and lemma [ lma : kabl : subcriterion ] .
the proof of theorem [ thm : kabl : detcond ] is delayed until section [ sec : kabl : proofdetcond ] .
fix a subsequence @xmath435 of @xmath419 .
we will show that there exists a further subsequence that converges weakly to @xmath2 in probability , which , by lemma [ lma : kabl : subcriterion ] would complete the proof of theorem [ thm : genkabluchko ] .
clearly , @xmath432 is a subsequence of @xmath29 .
if @xmath422 denotes lebesgue measure on @xmath423 , then markov s inequality implies that , for any @xmath33 , @xmath450 the last expression tends to zero as @xmath445 by the local integrability of the logarithm and the fact that @xmath66 .
thus , the sequence of functions @xmath451converges to zero in measure as @xmath445 . among other things
, this implies that there exists a subsequence of this sequence that converges to zero for almost every @xmath452 ( see , for instance , theorem 2.30 on page 61 of @xcite for details ) .
let @xmath453 denote the corresponding subsequence of random measures . by corollary [ cor : kabl : subseq ] , we have that @xmath454 converges weakly to @xmath2 in probability as @xmath455 , completing the proof .
it remains to prove theorem [ thm : kabl : detcond ] .
the proof presented here is modeled after the arguments given in @xcite .
the case where @xmath2 is degenerate is straightforward to establish by computing @xmath431 explicitly and directly verifying that @xmath456 almost surely as @xmath30 for any bounded and continuous function @xmath446 .
we now consider the case that @xmath2 is non - degenerate .
the proof of theorem [ thm : kabl : detcond ] will reduce to studying the logarithmic derivative @xmath457 of @xmath0 defined by the formula @xmath458 specifically , theorem [ thm : kabl : detcond ] will follow from lemma [ lma : kabl : intsmall ] below .
we also now state a related lemma ( lemma [ lma : kabl : lnsmall ] ) , which we will need later .
note that these two lemmas are very similar to ( * ? ? ?
* lemmas 2.1 and 2.2 ) ; however , neither lemma follows directly from the results in @xcite because of the deterministic contribution to @xmath457 . under the assumptions of theorem [ thm : kabl : detcond ] , for any continuous , compactly supported function @xmath462 , we have @xmath463 in probability as @xmath30 .
( recall that @xmath422 denotes lebesgue measure on @xmath19 . )
[ lma : kabl : intsmall ] we now prove theorem [ thm : kabl : detcond ] assuming lemma [ lma : kabl : intsmall ] .
the key idea is the following formula ( see , for instance , ( * ? ? ?
* section 2.4.1 ) ) , which relates the integral in to the measures @xmath29 and @xmath419 . for any polynomial @xmath10 that is not identically zero , @xmath464 in the distributional sense , where each root in the sum is counted with multiplicity .
in other words , for any compactly supported , smooth function @xmath462 , we have @xmath465 from this relationship we obtain that , for any smooth , compactly supported function @xmath462 ,
@xmath466 in view of lemma [ lma : kabl : intsmall ] , the integral on the right tends to zero in probability as @xmath467 .
in addition , by the law of large numbers and the fact that @xmath66 , @xmath468 almost surely as @xmath30 . hence , for any smooth , compactly supported function @xmath462 @xmath469 in probability as @xmath30 . since @xmath2 is a probability measure , we conclude from a simple approximation argument that @xmath419 converges weakly to @xmath2 in probability .
this completes the proof of theorem [ thm : kabl : detcond ] .
we now turn our attention to proving lemmas [ lma : kabl : lnsmall ] and [ lma : kabl : intsmall ] .
we begin with lemma [ lma : kabl : lnsmall ] , which we will need to prove lemma [ lma : kabl : intsmall ] .
first , we construct the exceptional set @xmath470 described in lemma [ lma : kabl : lnsmall ] from several smaller subsets .
the first of these , @xmath471 , contains points where @xmath2 misbehaves , while another , @xmath472 , includes values too close to the deterministic array .
define the set @xmath471 by @xmath473 @xmath471 has lebesgue measure zero since @xmath474 by the fubini
tonelli theorem .
we now construct the subset @xmath472 by applying the borel cantelli lemma .
recall that the set @xmath475 , defined in , is at most countable , and hence @xmath476 .
thus , for a fixed @xmath477 and @xmath478 , @xmath479 by markov s inequality , where @xmath49 is an absolute constant equal to the integral of @xmath480 over @xmath423 .
thus , we obtain @xmath481 since @xmath482 .
it follows by the borel
cantelli lemma and the fact that @xmath475 is countable that there exists a set @xmath483 of lebesgue measure zero such that , for every @xmath484 , @xmath485 for all but finitely many pairs @xmath486 .
we conclude that , for @xmath487 , @xmath488 where the asymptotic notation @xmath489 means the implicit constant is allowed to depend on @xmath14 . if we define @xmath470 to be @xmath490 then @xmath470 has lebesgue measure zero and , as we shall see , satisfies the requirements of lemma [ lma : kabl : lnsmall ] .
( recall the definition of @xmath428 from the statement of theorem [ thm : kabl : detcond ] above . )
notice that @xmath470 contains the atoms of @xmath2 and the values in the deterministic triangular array .
fix @xmath460 , and let @xmath33 be given . by markov s inequality , for any @xmath477 , we have @xmath492}{{\varepsilon}^2 n^2}\\ & = \frac{1}{{\varepsilon}^2 n^2 } \int_{\mathbb{c}}\log^2_{-}{\left\vertz - y\right\vert}\,d\mu(y)\\ & = \frac{c_1}{{\varepsilon}^2 n^2},\end{aligned}\ ] ] for a non - negative constant @xmath493 since @xmath494 .
hence , @xmath495 so the borel cantelli lemma applies . in particular , almost surely @xmath496 for all but finitely many @xmath1 .
furthermore , @xmath14 is not an atom of @xmath2 , so we have almost surely that , for all @xmath1 , @xmath497 where @xmath498 is an almost surely finite random variable .
now , since @xmath484 , the bound in implies that , for @xmath1 sufficiently large , @xmath499 for a positive constant @xmath500 ( depending on @xmath14 ) .
it follows that @xmath501 almost surely .
since @xmath33 was arbitrary , the proof is complete .
the reverse inequality in lemma [ lma : kabl : lnsmall ] requires an anti - concentration result that can be found , for example , in ( * ? ? ?
* theorem 2.22 on page 76 ) . before stating the lemma
, we define the lvy concentration function of a complex - valued random variable .
suppose that @xmath506 are iid , non - degenerate , complex - valued random variables .
then , there is a positive constant @xmath51 ( depending only on the distribution of @xmath507 ) , so that , for any @xmath504 , @xmath508 for all @xmath20 .
[ lma : anticoncentration ] theorem 2.22 on page 76 in @xcite implies that equation holds when @xmath509 are iid real - valued random variables and the supremum in the concentration function is taken over real numbers ( see also ( * ? ? ?
* corollary 6.8 ) for a more general version of this inequality ) .
we extend this to the complex case in the following way . by assumption ,
@xmath509 are iid and non - degenerate , so at least one of the real - valued random variables @xmath510 or @xmath511 is non - degenerate .
without loss of generality , assume @xmath510 is non - degenerate .
then @xmath512 the last expression is bounded by @xmath513 , for some constant @xmath51 that depends only on the distribution of @xmath510 by the previously mentioned result in @xcite .
a nearly identical argument applies if @xmath510 is degenerate and @xmath511 is non - degenerate . since @xmath66 , we assume @xmath1 is sufficiently large so that @xmath515 .
fix @xmath460 , and let @xmath33 be given .
since @xmath2 is non - degenerate and @xmath14 is not an atom of @xmath2 , it follows that @xmath516 are iid , non - degenerate , complex - valued random variables satisfying the hypotheses of lemma [ lma : anticoncentration ] . by absorbing the contribution of @xmath517 into the complex number @xmath518 in the definition of the concentration function , we conclude from lemma [ lma : anticoncentration ] that @xmath519 for a positive constant @xmath51 depending only on the distribution of @xmath520 . as @xmath30
, the right - hand side goes to zero ( since @xmath66 ) , which completes the proof .
in order to prove lemma [ lma : kabl : intsmall ] , we will apply lemma [ lma : kabl : tauvu ] to the random functions @xmath530 , where @xmath531 is a continuous function with compact support .
lemma [ lma : kabl : lnsmall ] establishes the first condition , and the tightness condition ( with @xmath532 ) follows from the next lemma . for the remainder of the paper ,
we let @xmath533 denote the open disk of radius @xmath534 centered about the origin .
fix @xmath535 such that the support of @xmath531 is contained in the open disk @xmath536 .
we will occasionally use @xmath537 to denote the indicator function of the set @xmath536 . in view of lemma [ lma :
kabl : tauvu ] , the proof of lemma [ lma : kabl : intsmall ] reduces to establishing lemma [ lma : kabl : inttight ] .
we bound the integral in lemma [ lma : kabl : inttight ] by employing the poisson
jensen formula as in @xcite .
in order to do so , we will need a uniform bound on @xmath539 for @xmath14 of certain magnitudes , which is the content of the following lemma .
the proof is similar in spirit to that of lemma [ lma : kabl : logbelowzero ] .
we first claim that @xmath543 for any @xmath544 , @xmath545 , and @xmath546 .
this equivalence will allow us to employ the method of lemma [ lma : kabl : logbelowzero ] and control the behavior of @xmath547 . to establish the forward direction of ,
observe that @xmath548 for any @xmath14 satisfying @xmath549 .
hence , @xmath550 implies @xmath551 . on the other hand , if @xmath551 , write @xmath552 in polar coordinates , and note that @xmath553 has modulus @xmath554 and satisfies @xmath555 the fact that @xmath550 follows .
we are ready to construct @xmath556 from two exceptional sets @xmath557 and @xmath558 .
define @xmath559 it follows from the fubini
tonelli theorem that @xmath557 has lebesgue measure zero since @xmath560 we now construct @xmath558 .
let @xmath561 denote lebesgue measure on the real line , and let @xmath562 clearly , @xmath563 .
equivalence and markov s inequality imply that for a fixed @xmath477 and @xmath478 , @xmath564 where @xmath49 is an absolute constant .
it follows that @xmath565 so the borel
cantelli lemma and the countability of @xmath566 show that outside of a set @xmath567 of lebesgue measure zero , @xmath568 for all but finitely many pairs @xmath486 .
hence , for @xmath569 , @xmath570 where @xmath571 is a positive constant depending on @xmath572 .
( note that since @xmath573 , @xmath574 for each pair @xmath486 ) .
if we define @xmath575 , then , @xmath576 has lebesgue measure zero , and for @xmath541 , we have that , for any @xmath577 and any @xmath33 , @xmath578 \\ & = \frac{c_r'}{n^2},\end{aligned}\ ] ] where we used in the first step and markov s inequality in the second . here , @xmath579 is a positive constant depending only on @xmath572 and @xmath2 . by the borel
cantelli lemma , it follows that almost surely , @xmath580 for all but finitely many @xmath1 .
this guarantees that for @xmath581 , there is an almost surely bounded , real - valued random variable @xmath582 for which @xmath583 almost surely .
( note that @xmath584 for all @xmath541 by the definition of the set @xmath557 . )
the last inequality holds for all sufficiently large @xmath1 by . as @xmath33 was arbitrary , now follows .
we now use the poisson
jensen formula to re - write @xmath585 . for
any @xmath586 and @xmath477 , let @xmath587 be the roots and critical points , respectively , of @xmath0 that are located in the open disk @xmath588 .
the poisson
jensen formula ( see , for example , ( * ? ? ?
* chapter ii.8 ) ) implies that for any @xmath589 which is not a zero or pole of @xmath457 , @xmath590 where @xmath591 and @xmath592 denotes the poisson kernel @xmath593,\ \alpha \in [ 0,2\pi ] .
\label{eqn : kabl : pker}\ ] ] fix @xmath596 .
then , for any @xmath597 $ ] and @xmath598 , we have @xmath599 the last inequality follows from the fact that @xmath600 and from the equivalence @xmath601 which holds for all @xmath602 .
consequently , for any @xmath596 and @xmath603 , @xmath604 therefore , we obtain @xmath605 the desired result now follows by applying lemma [ lma : kabl : unifzbound ] to . in particular , since the exceptional set @xmath540 of lemma [ lma : kabl : unifzbound ] has measure zero , we can choose @xmath594 so that holds almost surely .
next , we show that @xmath606 is bounded below uniformly for @xmath596 .
we assume that @xmath607 , and we first consider the case when @xmath608 .
there is no loss of generality in assuming @xmath607 , for if @xmath609 , we can choose a different point @xmath610 and prove theorem [ thm : kabl : detcond ] for the random variables @xmath611 and the deterministic array @xmath612 .
this follows since the translation of the roots of @xmath0 by @xmath613 simply translates the critical points by @xmath613 ( see proposition [ prop : translate ] ) .
since @xmath607 , we have @xmath615 almost surely ; in other words , @xmath529 is almost surely not a pole of @xmath457 .
furthermore , by lemma [ lma : kabl : logaboveneg ] , it follows that @xmath529 is not a zero of @xmath457 with probability @xmath616 .
consequently , on the same event , the poisson jensen formula applies to @xmath617 , and we obtain @xmath618 the inequality comes from eliminating @xmath619 we bound the remaining two terms in probability .
a bound for the first term follows from lemma [ lma : kabl : logaboveneg ] .
it remains to find a lower bound ( in probability ) for the last term in .
let @xmath620 be the random and deterministic roots , respectively , of @xmath0 that are contained in @xmath588 .
( note that @xmath621 . )
the law of large numbers implies that @xmath622 almost surely as @xmath30 .
the expectation on the right - hand side is finite since @xmath623 due to the assumption @xmath607 and by the bounds @xmath624 & = -{\mathbb{e}}\left[\log_{-}{\left\vert\frac{x_1}{r}\right\vert } - \log_{-}{\left\vertx_1\right\vert } + \log_{-}{\left\vertx_1\right\vert}\right]\\ & \geq-{\mathbb{e}}\left[\log_{-}\left(\frac{1}{r}\right)\right ] - { \mathbb{e}}\left[\log_{-}{\left\vertx_1\right\vert}\right ] \\ & \geq-\log(r ) - { \mathbb{e}}\left[\log_{-}{\left\vertx_1\right\vert}\right],\end{aligned}\ ] ] which follow from the fact that @xmath625 . since @xmath72
, it follows that @xmath626 \geq -\log(r ) - { \mathbb{e}}\log_{-}{\left\vertx_1\right\vert}\ ] ] almost surely as @xmath30 , and as a consequence , we have almost surely @xmath627 for some non - negative constant @xmath628 ( depending on @xmath572 ) .
similarly , as @xmath625 , we have @xmath629 by condition and the fact that @xmath66 , we obtain @xmath630 ( recall that @xmath607 , and hence @xmath631 . ) together , and imply the desired conclusion .
the proof presented here closely follows the arguments in @xcite . for simplicity , define @xmath633 for @xmath634 $ ] . by the definition of the poisson kernel and reasoning similar to that used to derive the bounds in
, we have @xmath635 for all @xmath596 and @xmath634 $ ] .
notice that @xmath636 for all @xmath637 $ ] , so we have @xmath638 it follows that , for any @xmath477 and any @xmath596 , @xmath639 in the case where @xmath640 for all @xmath634 $ ] , we obtain the bound @xmath641 otherwise , @xmath642 for all @xmath634 $ ] , and continuing from above , @xmath643 in either case , taking the infimum over all @xmath596 and applying the results of lemmas [ lma : kabl : unifzbound ] and [ lma : kabl : in0bd ] gives the desired conclusion .
we complete the proof of lemma [ lma : kabl : inttight ] by applying lemma [ lma : kabl : bdpint ] and lemma [ lma : kabl : bdpintbelow ] to .
let @xmath594 be as in lemma [ lma : kabl : bdpint ] . from
, we apply the cauchy
schwarz inequality twice to obtain @xmath644 for @xmath645 that is not a zero or pole of @xmath457 .
since there are finitely many zeros and poles of @xmath457 for a fixed @xmath1 and a fixed realization of @xmath457 , implies @xmath646 almost surely .
lemmas [ lma : kabl : bdpint ] and [ lma : kabl : bdpintbelow ] establish that @xmath647 for some constant @xmath49 , and hence the sequence @xmath648 is tight .
the remaining two terms of are bounded almost surely . indeed , for @xmath596 and @xmath649 , we have @xmath650 and hence @xmath651 by a simple change of variables , we obtain @xmath652 and similarly @xmath653 thus , by the local integrability of the squared logarithm , @xmath654 almost surely for all @xmath655 , where @xmath656 is a constant that depends only on @xmath572 and @xmath657 , and , in the last inequality , we used the fact that @xmath658 .
a similar argument applies to the integral of the sum in involving the critical points @xmath659 ; we omit the details .
r. pemantle , i. rivin , _ the distribution of zeros of the derivative of a random polynomial _ , advances in combinatorics .
waterloo workshop in computer algebra 2011 , i. kotsireas and e. v. zima , editors , springer , new york , 2013 .
t. r. reddy , _ on critical points of random polynomials and spectrum of certain products of random matrices _ ,
ph.d . thesis submitted in july , 2015 at indian institute of science , bangalore .
available at arxiv:1602.05298 .
we now prove theorem [ thm : simple ] using theorem [ thm : main ] .
indeed , let @xmath7 satisfy the assumptions of theorem [ thm : simple ] .
since @xmath7 do not depend on @xmath1 , there exists @xmath122 such that , for any @xmath123 , [ thm : reduction ] let @xmath2 be a probability measure on @xmath19 with compact support , and suppose @xmath310 .
let @xmath38 be iid random variables with distribution @xmath2 . for each @xmath20 ,
let @xmath67 be a triangular array of deterministic complex numbers with @xmath143 , and assume @xmath311 .
fix @xmath33 , and suppose that for all sufficiently large @xmath1 , there are no values of @xmath67 in @xmath145 and there are @xmath120 values @xmath146 outside @xmath147 . then , almost surely , for @xmath1 sufficiently large , there are exactly @xmath120 critical points ( counted with multiplicity ) of the polynomial @xmath148 outside @xmath149 , and after labeling these critical points correctly , @xmath150 for each @xmath126 .
the only difference between this theorem and theorem [ thm : main ] is that theorem [ thm : reduction ] assumes @xmath310 . using theorem [ thm : reduction ] , we prove theorem [ thm : main ] by applying proposition [ prop : translate ] .
let @xmath2 have compact support . since @xmath103 is nonempty , choose @xmath312 .
we now consider the polynomial @xmath313 where @xmath314 .
let @xmath315 be the distribution of @xmath316 .
then @xmath315 has compact support and @xmath317 . in addition , the sets @xmath318 and @xmath319 are translates by @xmath320 of the sets @xmath299 and @xmath103 , respectively .
thus , by assumption , there are no values of @xmath321 in @xmath322 and there are @xmath120 values @xmath323 outside @xmath324 .
therefore , by theorem [ thm : reduction ] and proposition [ prop : translate ] , we conclude that almost surely , for @xmath1 sufficiently large , there are exactly @xmath120 critical points of @xmath0 outside @xmath149 and after labeling correctly , @xmath325 for @xmath126 .
adding @xmath326 to both sides completes the proof .
[ thm : reductiondist ] let @xmath2 be a probability measure on @xmath19 with compact support , and suppose @xmath310 .
let @xmath38 be iid random variables with distribution @xmath2 . for each @xmath20 ,
let @xmath67 be a triangular array of deterministic complex numbers with @xmath143 .
fix @xmath33 , and suppose that for all sufficiently large @xmath1 , there are no values of @xmath67 in @xmath145 and there is one value @xmath153 outside @xmath147 . then , almost surely , for @xmath1 sufficiently large , there is exactly one critical point of the polynomial @xmath148 outside @xmath149 , and after labeling the critical points correctly , @xmath327 the proof of theorem [ thm : distinguished ] using theorem [ thm : reductiondist ] is nearly identical to the proof of theorem [ thm : main ] above ; we omit the details .
it remains to prove theorems [ thm : reduction ] and [ thm : reductiondist ] .
we prove theorems [ thm : reduction ] and [ thm : reductiondist ] simultaneously .
indeed , for the first part of the proof , we continue to use the notation of theorem [ thm : reduction ]
. however , the same argument applies to theorem [ thm : reductiondist ] by simply taking @xmath152 .
the conclusion of the proof will require us to consider the conditions of both theorems separately .
in fact , the conclusion of the proof is the only place where we require condition . for notational convenience , throughout the proof we allow the implicit constants and rates of convergence in our asymptotic notation ( such as @xmath328 ) to depend on the parameter @xmath89 without notating this dependence . for @xmath1 sufficiently large , we decompose @xmath329 where , by assumption
, @xmath330 are outside @xmath147 and @xmath331 are in @xmath94 .
in addition , @xmath332 are in @xmath333 with probability @xmath334 .
let @xmath180 be the diagonal matrix @xmath335 where @xmath336 and @xmath337 here , the subscripts `` @xmath338 '' and `` @xmath339 '' refer to the roots inside and outside @xmath94 , respectively .
of course , @xmath180 , @xmath340 , and @xmath341 all depend on @xmath1 , but we do not denote this dependence in our notation . by theorem [ thm : companion ] , it follows that @xmath342 , \nonumber\end{aligned}\ ] ] where @xmath219 is the identity matrix and @xmath343 is the @xmath220 all - one matrix .
we decompose , @xmath344 where @xmath345 denotes the @xmath346 all - one matrix .
thus , we conclude that @xmath347 we will eventually apply lemma [ lemma : block ] to compute this determinant , but first we will need to consider the upper - left block @xmath348 let @xmath349 denote the all - one @xmath1-vector ; we will often drop the subscript ( and just write @xmath350 ) when its size can be deduced from context .
we will make use of the following lemma .
[ lemma : technical ] under the assumptions of theorem [ thm : reduction ] ( alternatively , theorem [ thm : reductiondist ] ) , almost surely , for @xmath1 sufficiently large , the matrix @xmath351 is invertible for every @xmath352 and the function @xmath353 is analytic outside @xmath354 .
in addition , almost surely @xmath355 recall that the entries of the diagonal matrix @xmath340 are contained in @xmath94 .
thus , for @xmath352 , the matrix @xmath356 is invertible .
in addition , since @xmath357 is a diagonal matrix , we obtain @xmath358 \nonumber \\ & = \frac{1}{n } \sum_{j=1}^{n - k_n } \frac{x_j}{z - x_j } + \frac{1}{n } \sum_{l = s+1}^{k_n } \frac{\xi_l^{(n ) } } { z - \xi_l^{(n)}}. \label{eq : trpartial}\end{aligned}\ ] ] among other things , this implies that the function @xmath359 is analytic outside @xmath354 ; we will use this fact later to show that the function in is analytic on the same set .
since @xmath103 is compact , it follows from proposition [ prop : mmu ] that @xmath94 is bounded .
let @xmath360 be such that @xmath361 for all @xmath362 .
let @xmath363 .
then for @xmath364 , we have @xmath365 for @xmath366 and similarly @xmath367 for each @xmath368 .
thus , @xmath369 in particular , this bound implies that @xmath370 for all @xmath364 .
thus , we can apply lemma [ lemma : sm ] to conclude that the matrix in is invertible for every @xmath364 . indeed , since @xmath371 is at most rank one is rank one , and so the product @xmath372 is either rank one or rank zero . in fact , a simple computation reveals that the product is rank zero if and only if @xmath340 is the zero matrix . ]
, it follows from lemma [ lemma : sm ] ( taking @xmath373 and @xmath374 ) that @xmath375 hence , by the bound in , we have , with probability one , @xmath376 in addition , the right - hand side of is analytic in the region @xmath364 , which implies that the function on the left - hand side is also analytic in the same region .
let @xmath377 be the compact set @xmath378 .
it remains to show that , almost surely , for @xmath1 sufficiently large , the matrix in is invertible for every @xmath379 , the function in is analytic in @xmath377 , and @xmath380 to establish these results we will again apply lemma [ lemma : sm ] .
however , in this case , we will need more precise estimates than those established above . indeed , returning to , we find that @xmath381 since @xmath331 are contained in @xmath94 , it follows from the triangle inequality that @xmath382 in addition , by lemma [ lemma : lln ] and the borel
cantelli lemma , we have , almost surely @xmath383 as @xmath377 is compact and @xmath84 can not vanish on @xmath377 ( since @xmath384 ) , there exists @xmath95 such that @xmath385 for all @xmath379 .
specifically , by the assumption that @xmath310 , it follows that @xmath386 therefore , by , , and , we conclude from that , almost surely , for @xmath1 sufficiently large , @xmath387 and @xmath388 hence , by lemma [ lemma : sm ] , we obtain for @xmath379 which , combined with the bounds above , yields @xmath389 almost surely . as before , also implies that the function in is analytic on @xmath377 .
the proof of the lemma is complete .
let us dispatch the simplest case of theorem [ thm : reduction ] : when @xmath390 .
indeed , if @xmath390 , then @xmath391 . in this case , and the invertibility of imply that @xmath0 has no critical points outside @xmath149 , completing the proof .
thus , for the remainder of the proof , we assume @xmath392 .
we return to the block determinant in . by lemma [ lemma : technical ] ,
almost surely , for @xmath1 sufficiently large , the upper - left block is invertible for all @xmath352 .
thus , by lemma [ lemma : block ] , we conclude that almost surely @xmath393 for all @xmath352 , where @xmath394 in other words , the zeros of @xmath62 outside of @xmath354 ( counted with multiplicity ) are precisely the zeros of @xmath395 outside of @xmath354 ( counted with multiplicity ) .
notice that this is the determinant of an @xmath396 matrix , and @xmath397 .
we have thus reduced the problem of studying an @xmath220 matrix to an @xmath396 matrix .
this reduction greatly simplifies the forthcoming analysis .
before we conclude the proof , we make one final observation : since @xmath398 and @xmath399 , we can rewrite the determinant in as @xmath400 we now conclude the proof of theorems [ thm : reduction ] and [ thm : reductiondist ] separately .
let us begin with theorem [ thm : reduction ] .
indeed , under the assumptions of theorem [ thm : reduction ] , @xmath401 ( recall that @xmath402 denotes the spectral norm of the matrix @xmath341 . ) thus , by lemma [ lemma : detdiff ] and lemma [ lemma : technical ] , we have , almost surely @xmath403 because @xmath404 .
notice that the zeros of @xmath405 are precisely the values @xmath406 . in view of rouch
s theorem ( since both determinants are analytic outside @xmath149 due to lemma [ lemma : technical ] ) , we conclude that , almost surely , for @xmath1 sufficiently large , @xmath0 has exactly @xmath120 critical points outside @xmath407 , and after correctly labeling the critical points , @xmath408 for each @xmath126 .
this completes the proof of theorem [ thm : reduction ] .
we now turn to the proof of theorem [ thm : reductiondist ] .
recall that , in this case , @xmath409 .
thus , the matrix in is just a @xmath410 matrix , and hence the the zeros of @xmath62 outside of @xmath149 are precisely the solutions of @xmath411 outside @xmath149 . by lemma [ lemma : technical ] , we have , almost surely , @xmath412 for some constant @xmath49
. since both these terms are analytic outside @xmath354 due to lemma [ lemma : technical ] , we can again apply rouch s theorem .
however , since @xmath413 does not necessarily converge to zero , we have to be slightly more careful .
let @xmath414 be any simple closed contour outside @xmath354 which satisfies @xmath415 for all @xmath416 .
then , by the estimate in , rouch s theorem implies that the number of solutions to inside @xmath414 is the same as the number of zeros of @xmath417 inside @xmath414 .
hence , we conclude that almost surely , for @xmath1 sufficiently large , there is exactly one critical point of @xmath0 outside @xmath354 and that critical point takes the value @xmath418 .
the proof of theorem [ thm : reductiondist ] is complete . | let @xmath0 be a random , degree @xmath1 polynomial whose roots are chosen independently according to the probability measure @xmath2 on the complex plane . for a deterministic point @xmath3 lying outside the support of @xmath2 , we show that almost surely the polynomial @xmath4 has a critical point at distance @xmath5 from @xmath3 . in other words ,
conditioning the random polynomials @xmath0 to have a root at @xmath3 , almost surely forces a critical point near @xmath3 .
more generally , we prove an analogous result for the critical points of @xmath6 , where @xmath7 are deterministic .
in addition , when @xmath8 , we show that the empirical distribution constructed from the critical points of @xmath9 converges to @xmath2 in probability as the degree tends to infinity , extending a recent result of kabluchko @xcite . |
after over two decades of development , the advanced laser interferometer gravitational - wave observatory ( aligo ) has directly detected gravitational waves from an inspiralling black hole - black hole ( bh - bh ) binary @xcite . through the precise measurement of the gravitational waves , aligo is capable of characterizing many properties of inspiralling binaries , including the total mass of the binary , @xmath0 , the mass ratio , @xmath7 , and the black holes spins .
the first detected bh - bh binary , gw150914 , had unusually high component masses @xmath8 in comparison to bh masses inferred for star - bh x - ray binaries @xcite . a second , less significant event , lvt151012 , also had high inferred masses @xmath9 @xcite .
aligo has finally opened a new window to our universe . along with other upcoming instruments virgo and kagra
, aligo will allow us to probe the demographics of potentially hundreds of bh - bh binaries @xcite .
there are three primary pathways that lead to bh - bh binaries that can merge within the age of the universe , through binary evolution , through gas dynamics @xcite , and through stellar dynamics ( see * ? ? ?
* for a review ) .
first , such binaries can form through the evolution of isolated , massive binary stars . a major bottleneck in our understanding of this channel
is the complex tidal @xcite and common envelope ( e.g * ? ? ?
* ) evolution such a binary must go through in order to produce two bhs that can merge within a hubble time from the loss of gravitational waves .
this is in addition to uncertainties in the details of massive star evolution , supernova explosions , and the birth kicks of black holes .
nevertheless , sophisticated population synthesis routines have been developed that incorporate many of these uncertainties to make predictions about the properties of the first gravitational wave sources .
a second possibility to get bh binary mergers is through gas assisted mergers @xcite , however simple rate estimates suggest that gas assisted stellar bh mergers are probably relatively uncommon @xcite .
dynamical interactions of bhs in dense stellar environments , such as globular clusters , present another promising method to produce tight bh - bh binaries whether through exchange interactions @xcite , three body dynamics @xcite , or direct dynamical capture @xcite . in these scenarios , the black holes that form at the end stage of stellar evolution collect near the center of the cluster through dynamical friction @xcite .
because of gravitational focusing , the most massive bhs are preferentially involved in subsequent gravitational encounters and form bh - bh binaries .
these binaries may merge within the cluster @xcite or be ejected from the cluster and merge on much longer timescales @xcite .
such models have presented their own theoretical obstacles , the initial mass function of bhs perhaps the largest , but ever sophisticated simulations over nearly a decade have generally found similar estimates for the expected merger rate of the binaries as well as their characteristics @xcite .
these results remain even in simulations that have shown a substantial fraction of bhs remain in the cluster ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) in this work , we present a number of observational signatures of the dynamical formation of black hole binaries .
in particular , we focus on signatures that are independent of the poorly known black hole initial mass function ( imf ) @xcite . a number of studies have qualitatively discussed that dynamical interactions preferentially form binaries with the most massive components in the cluster ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . yet
few studies focused on the expected mass ratio distribution of the bh - bh binaries that merge . in this work , we use the original monte carlo results of @xcite , a series of new monte carlo simulations , as well as a series of new direct @xmath10-body simulations to explore the expected mass distribution of the components of bh - bh binaries .
we argue that the mass distribution of the bh binaries found by aligo will present a unique signature of dynamically formed binaries and their underlying mass function .
after we have submitted our manuscript two papers appeared on the expected rates of stellar black hole mergers in globular clusters with independent methodologies , which confirm our findings @xcite
after the first supernovae , the more massive bhs collect near the center of the cluster owing to dynamical friction from the low mass stellar background .
in contrast to previous expectations @xcite , however , these bhs do not interact exclusively amongst themselves .
after the bh subcluster experiences core collapse and produces a binary , that binary acts through the bh population to heat the entire star cluster until it is ejected from the system ( @xcite ; cf .
nevertheless , the strongest dynamical interactions that eventually lead to binary ejection rarely include stars @xcite . in this work
, we are interested in the properties of the bh - bh binaries , and only follow the evolution of the bhs separately from the cluster .
although this does not follow the proper evolution of the entire bh cluster , it does capture the important dynamical interactions that lead to binary formation and , eventually , binary heating . in this work ,
we use two methods for following the evolution of the bhs in dense stellar clusters .
we use the monte carlo method presented in @xcite as well as a direct @xmath10-body method @xcite that follows the evolution of only the bhs .
both methods focus on the evolution of the bhs in isolation from the stellar cluster .
our monte carlo simulations of bhs in globular clusters follow directly from @xcite , and is based on the original method by @xcite .
globular clusters are treated as two component systems , with a constant density core , where all dynamical interactions take place , and a low - density halo .
bhs that are kicked into the halo from dynamical interactions remain there until dynamical friction returns the bhs to the cluster center .
the three and four - body interactions in the cluster are sampled from their likelihood distributions and directly integrated with fewbody @xcite .
binaries that are present in the simulation are evolved following @xcite equations which describes the inspiral due to gravitational wave ( gw ) emission . if two non - spinning bhs merge within the cluster their merger product receives a kick due to the asymmetric emission of gws with a maximum velocity of @xmath11 near @xmath12 @xcite .
equal mass mergers receive no kick as they are symmetric .
the bh mass function changes in the simulation due to bh mergers and ejections .
we also run a series of direct @xmath10-body simulations to follow the evolution of the bhs that form in a much larger stellar cluster .
we integrate the orbits of all the bhs directly using nbody6++ @xcite , and keep track of the dynamically - formed binary population .
future work will include the full post - newtonian treatment throughout the cluster evolution , as mergers in the cluster likely constitute about @xmath13 of all mergers @xcite .
these simulations , due to their initial conditions , produce too few bh - bh binaries that merge in a hubble time , so we include all binaries ejected from the system in our analysis to compare with the monte carlo results .
two model suites were produced .
first , the model that includes a potential that mimics the underlying stellar cluster using a static plummer sphere with half - mass radius @xmath14pc and mass @xmath15 and a bh subcluster made of 492 single particles and 10 `` primordial '' binary pairs with half - mass radius of 3.3pc .
we also simulate the cluster of bhs without an external potential .
this model has only 512 single particles taken from a plummer distribution with half - mass radius of 0.5pc .
the integration is performed to 10gyr in all models . in both monte carlo and @xmath10-body simulations ,
we generate our initial bh mass function from a power - law distribution @xmath16 between minimum and maximum bh mass of @xmath17 and @xmath18 with @xmath190 , 1 , 2 , 3 , and 4 . while this range of masses is larger than the inferred masses of bhs in _ x_-ray binaries @xcite , it is consistent with models of bhs that form with fallback accretion and direct collapse @xcite , and is minimally required to explain gw150914 .
we also run a set of simulations using the bh mass function found in the population synthesis studies of @xcite for low - metallicity stars .
this mass function has three peaks in its distribution , near @xmath20 , @xmath21 , and @xmath22 , and represents a more complicated and perhaps more realistic distribution of masses . in the monte carlo simulations , we assume that @xmath23 of black holes are in primordial binaries ( cf . the initial conditions for the @xmath10-body above ) , with a period distribution that is flat in the log of period .
these two methods are complimentary to each other , and both capture the strongest dynamical encounters between bhs .
nevertheless they have their limitations and benefits . in both series of simulations
, we do not directly follow the entire coupling between the main stellar population in the cluster and the black holes , which is important to determine the present day population of bhs in globular clusters ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , the latter simulations confirm that the dynamical interactions that lead to binary ejections are driven exclusively by the bh subcluster , and are dominated by bh - bh binaries . with the fast monte carlo simulations
, we can run a much larger number of simulations with different compositions and initial conditions than studies which follow the evolution of the entire stellar population , and we can turn important processes , such as three - body binary formation , off in order to understand which aspects of the dynamical interactions are most important . to leading ( 2.5 post - newtonian ) order ,
the root - mean - square ( rms ) detection signal to noise ratio for a circular inspiraling binary is given as @xmath24 where @xmath25 for an isotropic binary orientation relative to the detector , @xmath26 is the symmetric mass ratio , @xmath27 is the luminosity distance , and @xmath28 where @xmath29 is the gw frequency at the innermost stable circular orbit , @xmath30 is the lowest gw frequency that the instrument can detect , and @xmath31 is the one - sided noise spectral density @xcite .. to convert from mass to time units , one should multiply by @xmath32 .
if the source is at a cosmological distance , redshift @xmath33 , then @xmath0 should be replaced by @xmath34 . ]
we set @xmath35hz and @xmath36 using the calibrated sensitivity spectra of aligo - hanford on october 1 , 2015 @xcite . for a fixed detection threshold ( e.g. @xmath37 ) the maximum distance range @xmath38
may be obtained from eq .
( [ eq : snr ] ) . assuming that the source population is uniformly distributed in volume
, the detection rate is biased by approximately @xmath39 this function grows as @xmath40 between @xmath41@xmath42 , has a maximum at @xmath43 and decreases to zero at @xmath44 where @xmath45hz .
compared to @xmath46 and fixed @xmath7 , the detectable volume is a factor 27 larger for @xmath43 and a factor 4.5 smaller at @xmath47 .
dynamical interactions between bhs in stellar clusters are primarily driven by mass segregation , gravitational focusing , and multibody dynamics .
we therefore expect the most massive bhs to be preferentially found in merging bh - bh binaries . in figure
[ fig : massto ] , we compare the mass distribution of merged bh - bh binaries from our simulations to that expected from random pairings of bhs from the bh mass function . we find that dynamical interactions enhance the merger rate of bh binaries relative to random pairings in a star cluster by a boost factor that scales as a power - law , @xmath48 with @xmath2 and we assume that @xmath49 is independent of @xmath50 .
all of our monte carlo simulations have @xmath51 .
this is even true for the @xcite mass function , which is not monotonically decreasing as a function of mass , and has multiple peaks .
we have also found similar results for our monte carlo simulations with a variety of cluster models , with differing velocity dispersions and escape velocities .
although our @xmath10-body simulations with an external potential still had @xmath51 , we found that the binaries that form in the cluster without the external potential have @xmath52 .
we discuss this in more detail in section [ sec : conc ] .
+ the total event rate per star cluster is then @xmath53 which depends on the total binary mass as @xmath54 this integral may be evaluated analytically for @xmath55 using gauss s hypergeometric function @xmath56 as @xmath57 where @xmath58 this may be expressed with elementary functions for integer and half - integer @xmath59 .
plots of eq .
( [ eq : gamma ] ) show that the merger rate peaks at @xmath60 for @xmath61 and near @xmath62 for larger @xmath59 . accounting for the correlation between @xmath0 and @xmath7 seen in fig .
[ fig : massrat ] skews @xmath63 to even higher @xmath0 , discussed next .
we discuss how this relationship may be used to probe the underlying mass distribution in section [ sec : conc ] .
dynamical encounters not only preferably form high mass bh - bh binaries , but as @xmath0 approaches @xmath64 the binaries necessarily approach @xmath65 . in fig .
[ fig : massrat ] , we show two - dimensional ( @xmath0@xmath7 ) distribution functions for bh - bh mergers from the cluster for two bh mass functions using our monte carlo results . these distributions were recreated from the simulation results using a gaussian kernel density estimator to generate a smooth distribution .
the top panel shows the results from our power - law mass function @xmath66 and @xmath67 .
the bottom panel shows the results from the @xcite mass function .
both distributions peak near @xmath68 .
the peak is much more pronounced in mass functions with @xmath69 . for mass functions with @xmath70 ,
the @xmath7 distribution marginalized over @xmath0 rises rapidly near @xmath71 and is flat for higher @xmath7 until @xmath72 . since the maximum distance to which aligo may detect a binary is different for different @xmath0 and @xmath7 , the observed rate distribution of dynamically formed binaries will be biased relative to the merger rate , @xmath73 ( see eq .
[ eq : v ] ) .
@xcite has shown that for the planned aligo spectral noise density , the detection distance will be largest for @xmath0 between @xmath74 and @xmath75 for circular binaries ( see their figure 11 ) .
for the most recent observing run , o1 , aligo was most sensitive to bh mergers with @xmath76 .
further , the detection rate is biased toward equal - mass binaries as shown by eq .
( [ eq : v ] ) .
the overall impact of aligo s sensitivity can be seen by comparing the left and right panels of figure [ fig : massrat ] . in the right panel
, we show the distribution of mergers weighted by the observable volume of the universe respectively for different @xmath77 for the aligo o1 observing run .
all results have been normalized to peak at 1 . by comparing our results to the panels on the left
, we can see that aligo should expect to see high mass black hole binaries , with mass ratios peaking near 1 .
however , in these models the overall distribution is mostly determined by the underlying dynamics of bhs in star clusters , rather then aligo s intrinsic sensitivity .
this suggests that earth - based gw detectors will be a useful probe of the underlying demographics of bhs .
so far , we have only compared the binary merger distribution to the bh imf without regard for the age of the cluster . while young clusters may be an important contributor to the merger rate @xcite ,
it is not clear if they survive long enough to dominate the rate .
most studies therefore focus on an old population of clusters , such as globulars , which are long lived . in figure
[ fig : agemassrat ] , we show the merger and detection rate distributions from binaries during the last @xmath78gyr , roughly the observable horizon of aligo .
we find that mergers from old clusters is better fit by a broken power law . at high masses
@xmath79 , the merger rate is still enhanced by @xmath2 as in figure fig : massto . however , for @xmath80 , @xmath81 .
in dense stellar systems , the black holes that form at the end stage of stellar evolution collect near the center of the cluster through dynamical friction . gravitational focusing in subsequent multibody dynamical interactions naturally increase the interaction rate between the most massive black holes in the cluster , eventually leading them to merge . in this work
we have used a suite of numerical simulations to better understand the properties of bh - bh binaries that are formed in these clusters .
we have found that dynamical interactions between the bhs strongly enhance the rate of mergers among higher mass bhs , such that the merger rate of binaries is boosted by a factor @xmath1 , with @xmath2 , over a random selection of bh pairs from the cluster .
this relationship may be a useful tool to probe the underlying mass distribution of bhs in the aligo era with multiple detections .
the merger distribution can be estimated directly from the underlying bh mass function , assuming that aligo can distinguish the two primary channels of producing bh - bh binaries , or that bh - bh binaries formed during dynamical interactions dominate the detection rate . under this assumption
, the aligo measurement statistics will give physical meaning to the underlying mass function and dynamics .
in this work , we have minimally explored the underlying reason for @xmath2 . in the dynamical interactions considered here
, the process most sensitive to @xmath0 is the rate of binary formation through three - body encounters , which scales as @xmath82 @xcite .
we have run our monte carlo simulations without any primordial binaries , and indeed found @xmath83 . in these simulations ,
the binaries that form are sampled from the entire bh population and should follow this distribution ] .
subsequent interactions between the new binary do not appear to alter the distribution . in our @xmath10-body simulations
, we have found that @xmath51 , when we include a background potential .
however , we find @xmath52 when this potential was absent . we expect that this much higher boosting rate was caused by mass segregation in the core of the bh cluster , which is not present in our monte carlo simulations .
presently , our simulations do not include the impact of relaxation between the bh cluster and background stellar population . nevertheless , the results presented in @xcite , who used a similar initial mass function , appear consistent with our results . overall , we find the most likely bh binaries to be detected by aligo have total masses slightly less than @xmath84 when @xmath85 . for shallow mass functions , with @xmath86 ,
the effect is most prominent . for the case where @xmath51 and @xmath87
, the distribution of bh mergers is fairly broad between @xmath88 and @xmath64 .
mass segregation in the core of the bh cluster , however , can significantly boost the fraction of mergers near @xmath89 .
note that aligo is sensitive to inspirals with masses up to @xmath47 and it may measure the ringdown waveform for even higher @xmath0 @xcite .
thus ligo will be capable of measuring the value of @xmath90 over a wide range . a bayesian analysis of the distribution of ligo detections will allow one to put constraints on the underlying initial bh mass function and degree of mass segregation in these clusters .
looking at the distribution of mergers in figure [ fig : massrat ] , there is a small population of mergers with @xmath91 .
these binaries must have formed from at least one bh that was involved in a previous merger . for low - spinning bhs ,
as we consider here , we expect low merger kick velocities when @xmath92 , as these binaries emit gws symmetrically @xcite .
these merger remnants then remain in the cluster as the largest bh , and subsequently merge with another high mass bh in the cluster .
these mergers are therefore distributed around @xmath93 .
since bh spin is expected to reach @xmath94 in circular mergers of nonspinning bhs , the heavy binary component in this population may be expected to have such high spin while the other component is nonspinning , provided that high stellar - mass bhs form with nearly zero spin @xcite .
these subsequent mergers are infrequent , @xmath95 of mergers for the o1 science run of aligo , however they represent a smoking gun of dynamical interactions . at the final design specifications aligo
will be more sensitive at detecting this population of bhs , @xmath96 of all inspirals in dense stellar clusters will constitute subsequent mergers .
for an old population of clusters , the aligo detection rate of subsequent mergers is closer to @xmath97 of all detected inspirals . | the dynamical formation of stellar - mass black hole - black hole binaries has long been a promising source of gravitational waves for the laser interferometer gravitational - wave observatory ( ligo ) .
mass segregation , gravitational focusing , and multibody dynamical interactions naturally increase the interaction rate between the most massive black holes in dense stellar systems , eventually leading them to merge . we find that dynamical interactions , particularly three - body binary formation , enhance the merger rate of black hole binaries with total mass @xmath0 roughly as @xmath1 , with @xmath2 .
we find that this relation holds mostly independently of the initial mass function , but the exact value depends on the degree of mass segregation .
the detection rate of such massive black hole binaries is only further enhanced by ligo s greater sensitivity to massive black hole binaries with @xmath3 .
we find that for power - law bh mass functions @xmath4 with @xmath5 , ligo is most likely to detect black hole binaries with a mass twice that of the maximum initial black hole mass and a mass ratio near one .
repeated mergers of black holes inside the cluster result in about @xmath6 of mergers being observed between two and three times the maximum initial black hole mass .
using these relations , one may be able to invert the observed distribution to the initial mass function with multiple detections of merging black hole binaries . |
the results of sidis ( semi inclusive deep inelastic scattering ) experiments with polarised beams and targets enable the extraction of the spin dependent quark and gluon densities .
this powerful tool of studying the internal spin structure of the nucleon allows verification of sum rules .
one of them is the bjorken sum rule ( bsr ) @xcite , which refers to the first moment of the nonsinglet spin dependent structure function @xmath11 . because of @xmath12 flavour symmetry
, bsr is regarded as exact .
thus all of estimations of polarised parton distributions should be performed under the assumption that the bsr is valid .
determination of the sum rules requires knowledge of spin dependent structure functions over the entire region of @xmath13 .
the experimentally accessible @xmath0 range for the spin dependent dis is however limited ( @xmath14 for smc data @xcite ) and therefore one should extrapolate results to @xmath15 and @xmath16 .
the extrapolation to @xmath8 , where structure functions grow strongly , is much more important than the extrapolation to @xmath17 , where structure functions vanish . assuming that the bsr is valid
, one can determinate from existing experimental data the very small-@xmath0 contribution ( @xmath18 ) to the sum rule .
theoretical analysis of the small-@xmath0 behaviour of @xmath19 together with the broad @xmath0-range measurement data allow verification of the shape of the input parton distributions . in this way one can determinate the free parameters in these input distributions .
experimental data confirm the theoretical predictions of the singular small-@xmath0 behaviour of the polarised structure functions .
it is well known , that the low-@xmath0 behaviour of both unpolarised and polarised structure functions is controlled by the double logarithmic terms @xmath20 @xcite,@xcite . for the unpolarised case ,
this singular pqcd behaviour is however overridden by the leading regge contribution @xcite .
therefore , the double logarithmic approximation is very important particularly for the spin dependent structure function @xmath21 .
the resummation of the @xmath1 terms at low @xmath0 goes beyond the standard lo and nlo pqcd evolution of the parton densities .
the nonsinglet polarised structure function @xmath22 , governed by leading @xmath23 terms , is a convenient function both for theoretical analysis ( because of its simplicity ) and for the experimental bsr tests .
the small-@xmath0 behaviour of @xmath22 implied by double logarithmic approximation has a form @xmath24 with @xmath25 .
this or similar small-@xmath0 extrapolation of the spin dependent quark distributions have been assumed in recent input parametrisations e.g. in @xcite,@xcite . in our theoretical analysis within @xmath1 approach
we estimate the very small-@xmath0 contributions @xmath26 and @xmath27 ( @xmath28 ) to the bsr . using analytical solutions for the full and the truncated moments of the unintegrated structure function @xmath3 @xcite ,
@xcite we find the contributions @xmath27 for different input quark parametrisations : the regge nonsingular one and the singular ones .
we compare our results with the suitable experimental smc data for bsr . in the next section we recall some of the recent theoretical developments concerning the small-@xmath0 behaviour of the nonsinglet
polarised structure function @xmath22 .
section 3 is devoted to the presentation of the double logarithmic @xmath1 approximation , in which we calculate analytically the full and the truncated moments of the nonsinglet function @xmath3 .
section 4 contains our results for the very small-@xmath0 contributions to the bjorken sum rule @xmath27 ( @xmath29 ) .
we present our predictions using flat ( nonsingular ) @xmath30 and singular @xmath31 at small-@xmath0 parametrisations of the input structure function @xmath2 as well .
we compare our results with the smc data for the small-@xmath0 contribution to the bsr .
basing on the validity of the bsr we roughly estimate the slope @xmath5 controlling the small-@xmath0 behaviour of @xmath6 . in section 5
we discuss further possible improvement of our approach .
we formulate the unified equation for the truncated moments of the unintegrated function @xmath32 which incorporates @xmath1 resummation at low @xmath0 and lo dglap @xmath33 evolution as well .
we also discuss the role of the running coupling effects .
finally , section 6 contains a summary of our paper .
the small value of the bjorken parameter @xmath0 , specifying the longitudinal momentum fraction of a hadron carried by a parton , corresponds by definition to the regge limit ( @xmath8 ) .
therefore the small-@xmath0 behaviour of structure functions can be described using the regge pole exchange model @xcite . in this model
the spin dependent nonsinglet structure function @xmath34 in the low-@xmath0 region behave as : @xmath35 where @xmath36 is the intercept of the @xmath37 regge pole trajectory , corresponding to the axial vector meson and lies in the limits @xmath38 this low value of the intercept ( [ r2.2 ] ) implies the nonsingular , flat behaviour of the @xmath22 function at small-@xmath0 .
the nonperturbative contribution of the @xmath37 regge pole is however overridden by the perturbative qcd contributions , particularly by resummation of double logarithmic terms @xmath1 . in this way
the regge behaviour of the spin dependent structure functions is unstable against the perturbative qcd expectations , which at low-@xmath0 generate more singular @xmath0 dependence than that implied by ( [ r2.1])-([r2.2 ] ) .
nowadays it is well known that the small-@xmath0 behaviour of the nonsinglet polarised structure function @xmath22 is governed by the double logarithmic terms i.e. @xmath20 @xcite,@xcite .
effects of these @xmath1 approach go beyond the standard lo and even nlo @xmath33 evolution of the spin dependent parton distributions and significantly modify the regge pole model expectations for the structure functions .
from the recent theoretical analyses of the low-@xmath0 behaviour of the @xmath22 function @xcite one can find that resummation of the double logarithmic terms @xmath20 leads to the singular form : @xmath39 with @xmath25 .
this behaviour of @xmath22 is well confirmed by experimental data , after a low-@xmath0 extrapolation beyond the measured region @xcite,@xcite,@xcite .
perturbative qcd predicts a strong increase of the structure function @xmath11 with the decreasing parameter @xmath0 @xcite,@xcite what is confirmed by experimental data @xcite,@xcite,@xcite .
this growth is implied by resummation of @xmath1 terms in the perturbative expansion .
the double logarithmic effects come from the ladder diagram with quark and gluon exchanges along the chain . in this approximation
the unintegrated nonsinglet structure function @xmath3 satisfies the following integral evolution equation @xcite : @xmath40 where @xmath41 and @xmath42 is a nonperturbative contribution which has a form : @xmath43 the input parametrisation @xmath44 in the regge model of the low - lying trajectory @xmath37 exchange has a small-@xmath0 behaviour : @xmath45 more singular shape of the input function @xmath46 results from the recent experimental data and pqcd analyses : @xmath47 where @xmath5 has changed during last years obtaining the values from @xmath48 @xcite to recently 0.4 @xcite . in our calculations in the next section
we use different inputs @xmath46 : the flat one and the singular ones at low-@xmath0 as well .
the unintegrated distribution @xmath3 is related to the @xmath11 via @xmath49 relation ( [ r3.1 ] ) implies the following equation for the truncated mellin moment function @xmath50 @xcite : @xmath51 \nonumber \\\end{aligned}\ ] ] where @xmath52 for @xmath53 equation ( [ r3.8 ] ) reduces to that for the full mellin moment @xmath54 and in this case the analytical solution for fixed @xmath55 obtained in @xcite has a form @xmath56 where @xmath57\ ] ] @xmath58 and @xmath59 is obviously equal to @xmath60 at @xmath53 . using truncated moments
approach one can avoid uncertainty from the unmeasurable @xmath8 region and also obtain important theoretical results incorporating perturbative qcd effects at small @xmath0 , which could be verified experimentally .
truncated moments of parton distributions have been recently used in the lo and nlo dglap analysis @xcite . in the double logarithmic approach the analytical solution of the evolution equation ( [ r3.8 ] ) for fixed coupling @xmath55
has a form @xcite : @xmath61 where @xmath62 @xmath63 is given in ( [ r3.12 ] ) and @xmath64 is the inhomogeneous term , independent on @xmath33 : @xmath65 our purpose is to calculate the truncated moments of the nonsinglet polarised structure function @xmath22 @xmath66 using different input parametrisations @xmath46 .
this will allow estimation of the small-@xmath0 contribution to the bjorken sum rule and comparison of the results with suitable experimental data .
our predictions for @xmath67 will be presented in the forthcoming section .
dealing with truncated moments @xmath68 one can avoid uncertainties from the experimentally unavailable regions @xmath8 and @xmath17 . particularly important
is knowledge of the small-@xmath0 behaviour of the structure functions . in this limit
@xmath8 @xmath22 increases as @xmath69 @xcite , what is confirmed by a low-@xmath0 extrapolation of experimental data @xcite,@xcite,@xcite .
this extrapolation of @xmath22 ( @xmath70 and @xmath71 ) beyond the measured region of @xmath0 is necessary to compute its first moment @xmath72 and test the bsr .
the bsr is a fundamental rule and must be hold as a rigorous prediction of qcd in the limit of the infinite momentum transfer @xmath33 : @xmath73 where @xmath74 @xmath75 and @xmath76 is the neutron @xmath77-decay constant @xmath78 ) reads : @xmath79 the small-@xmath0 contribution to the bsr has a form : @xmath80 taking into account ( [ r3.15])-([r3.17 ] ) and also the relation @xmath81 one obtains immediately : @xmath82 where @xmath83 defined in ( [ r3.18 ] ) in a case of @xmath84 has a form : @xmath85 \nonumber\\\end{aligned}\ ] ] @xmath86 in the right - hand side of ( [ r4.9 ] ) is defined as @xmath87 and @xmath88 in table i we present our results for the low-@xmath0 contributions to the bsr ( [ r4.6 ] ) together with @xmath89 , which is defined by the following expression : @xmath90 \int\limits_{x_1}^{x_2 } dx g_1^{0ns}(x)\ ] ] in the last column we give the percentage value @xmath91 $ ] : @xmath92 .the small-@xmath0 contribution to the bsr ( [ r4.6 ] ) for different input parametrisations ( [ r4.14])-([r4.17 ] ) within @xmath1 approximation . [ cols="^,^,^,^,^",options="header " , ] the predictions have been found for four different input parametrisations @xmath93 , chosen at @xmath94 ( 1 , 2 , 3 inputs ) or at @xmath95 ( 4 input ) : @xmath96 @xmath97 @xmath98 \nonumber \\\end{aligned}\ ] ] @xmath99 \nonumber \\\end{aligned}\ ] ] input 1 is the simple regge form , constance as @xmath8 ; input 2 is a `` toy '' model , in which we have used the latest theoretical results concerning the small-@xmath0 behaviour @xmath69 of the nonsinglet function @xmath22 @xcite .
finally , input 3 @xcite and input 4 @xcite are parametrisations obtained lately within experimental data analysis .
they are quite different because there are nowadays still no direct experimental data from the low-@xmath0 region .
the extrapolation of @xmath22 to very small-@xmath0 region depends strongly on the assumption ( input parametrisation ) made for this extrapolation . in fig.1
we plot all used inputs @xmath93 in the small-@xmath0 region @xmath100 $ ] .
in fig.2 we present the small-@xmath0 contribution to the bsr @xmath101 as a function of @xmath0 and in fig.3 the value of @xmath102 also as a function of @xmath0 is shown .
numbers at each plot correspond to the suitable inputs @xmath103 .
( [ r4.14])-([r4.17 ] ) in the small-@xmath0 region.,width=340 ] contribution to the bsr @xmath101 ( [ r4.6 ] ) for different inputs @xmath46 within @xmath1 approach.,width=340 ] ( [ r4.12 ] ) in the low-@xmath0 region for different inputs @xmath46 within @xmath1 approach.,width=340 ] for all calculated moments @xmath104 and @xmath105 . from these results
one can read that the low-@xmath0 contribution to the bsr strongly depends on the input parametrisation @xmath46 .
for the flat regge form ( [ r4.14 ] ) @xmath106 is equal to around @xmath107 of the total @xmath108 , while for the singular inputs ( [ r4.15])-([r4.17 ] ) @xmath109 , @xmath110 and @xmath111 respectively .
the most singular parametrisation 4 ( [ r4.17 ] ) , predicting @xmath112 behaviour of @xmath22 as @xmath8 gives of course the largest small-@xmath0 contribution to the bsr in comparison to the rest of the inputs .
the value of @xmath89 , defined in ( [ r4.12 ] ) varies from @xmath113 for the regge input 1 to @xmath114 for the `` subsingular '' input 4 .
two similar as @xmath8 parametrisations 2 and 3 give @xmath89 about 0.2 .
it means that the double logarithmic @xmath1 effects are better visible in a case of nonsingular inputs . in a case of singular input parametrisations @xmath115 ( @xmath116 or so ) the growth of @xmath22 at small-@xmath0 , implied by the @xmath1 terms resummation , is hidden behind the singular behaviour of @xmath46 , which survives the qcd evolution . from the experimental smc data @xcite the low-@xmath0 contribution to
the bsr at @xmath104 is equal to @xmath117 the above result has been obtained via an extrapolation of @xmath22 to the unmeasured region of @xmath0 : @xmath8 .
forms of the polarised quark distributions have been fitted to smc semi - inclusive and inclusive asymmetries . in the fitting different parametrisations of the polarised quark distributions
@xcite @xcite have been used . in this way
present experimental data give only indirectly the estimation of the small-@xmath0 contribution to the moments of parton distributions .
the result ( [ r4.18 ] ) with a large statistical error and strongly fit - dependent can not be a final , crucial value .
nevertheless we would like to estimate the exponent @xmath5 in the low-@xmath0 behaviour of @xmath6 using the above smc result for the small-@xmath0 contribution to the bsr . assuming the validity of the bsr ( [ r4.5 ] ) at large @xmath104
, one can find : @xmath118 where @xmath119 is a very small value of the bjorken variable . taking into account the small-@xmath0 dependence of @xmath6 and the experimental data for @xmath120 one can obtain :
@xmath121 the constant @xmath122 can be eliminated from a low-@xmath0 smc data @xcite : @xmath123 taking different small-@xmath0 smc data , we have found @xmath124 ( @xmath125 ) ; @xmath126 ( @xmath127 ) ; @xmath128 ( @xmath129 ) . comparing our predictions for @xmath120 in table i with suitable smc data , one can read that the most probably small-@xmath0 behaviour of @xmath22 is @xmath130 this is consistent with latest theoretical analyses @xcite . the same value of @xmath131 was obtained in the semi - phenomenological estimation from bsr for lower @xmath33 @xcite .
small-@xmath0 contribution to the bjorken sum rule resulting from the indirect smc data analysis is equal to around @xmath132 of the total value of the sum .
similar value @xmath133 ( see table i ) gives our qcd approach , which incorporates the singular input parametrisation @xmath134 according to the double logarithmic resummation effects .
in the next point we will discuss further possible improvement of our approach .
in the previous section we have presented small-@xmath0 contribution to the bjorken sum rule , calculated within double logarithmic @xmath1 approximation . in our approach we have used a constance ( nonrunning ) @xmath136 .
this simplification allows the analytical analysis of the suitable evolution equations for truncated and full moments of the unintegrated structure function @xmath3 .
it has been however lately proved @xcite , that dealing with a very small-@xmath0 region one should use a prescription for the running coupling in a form @xmath137 .
this parametrisation is theoretically more justified than @xmath138 .
namely , the substitution @xmath138 is valid only for hard qcd processes , when @xmath139 .
however the evolution of dis structure functions at small-@xmath0 needs `` more running '' @xmath140 .
taking into account this running coupling constant effects we obtain the following modified equation for the @xmath3 : @xmath141 this modification should lead to smaller values of @xmath142 and hence @xmath143 than for the constance @xmath144 .
we will discuss this problem in the next paper .
besides , our present analysis is adequate only for a small-@xmath0 region , where the main role play the double logarithmic @xmath1 effects .
in the situation , when the present experimental data do not cover the whole region of @xmath13 , theoretical predictions for e.g. structure functions in the unmeasured low-@xmath0 region can not be directly verified .
thus theoretical analysis should concern both the small-@xmath0 physics and the available experimentally larger-@xmath0 area as well .
latest experimental smc @xcite,@xcite and hermes @xcite data provide results for the bsr from the region @xmath145 and
@xmath146 respectively . in the very small-@xmath0 region exist only indirect , extrapolated results with large uncertainties .
thus if one wants to compare theoretical predictions with suitable real measurements , one has to use an approach , which is proper for broader range of @xmath0 and @xmath33 .
the small-@xmath0 behaviour of the nonsinglet spin dependent structure function @xmath11 is governed by the double logarithmic @xmath1 terms .
this approximation is however inaccurate for qcd analysis at larger values of @xmath0
. therefore the double logarithmic approach should be completed by lo dglap @xmath33 evolution .
unified description of the polarised structure function @xmath3 incorporating dglap evolution and the double logarithmic @xmath1 effects at low-@xmath0 leads to the following equation for the truncated moments @xmath142 : @xmath147 \nonumber \\
+ \bar{\alpha_s}\int\limits_{q_0 ^ 2}^{q^2 } \frac{dk'^2}{k'^2 } \sum\limits_{p=0}^{m } c_{pn } \bar{f}^{ns}(x_0,n+p ,
k'^2 ) \nonumber \\\end{aligned}\ ] ] with fixed coupling constant @xmath135 .
matrix elements @xmath148 are given by @xmath149 \nonumber \\ + \sum\limits_{k = p}^{m}\frac{(-1)^p}{p!(k - p)!}[2\sum\limits_{i=1}^{n+1 } \frac{(i+k-1)!}{i!}x_0^i - \frac{(n+k-1)!}{n ! } ( x_0^n + \frac{n+k}{n+1 } x_0^{n+1 } ) ] \nonumber \\\end{aligned}\ ] ] where @xmath150 the unified description @xmath1 + lo dglap of the nonsinglet spin dependent structure function @xmath22 enables to determine the small-@xmath0 contribution to the bjorken sum rule via medium- and large-@xmath0 contribution of the unintegrated structure function @xmath151 .
because of the relation @xmath152 which is discussed e.g. in @xcite and @xcite , one can find immediately @xmath153 in this way , the small-@xmath0 part @xmath154 can be expressed by the well known from experimental analysis the larger-@xmath0 contribution @xmath155 . improved picture of pqcd behaviour of the truncated moments of @xmath22 incorporating the running coupling effects at small-@xmath0 and the unified @xmath1 + lo dglap approach will be a topic of our next paper .
in this paper we have estimated the contribution from the small-@xmath0 region to the bjorken sum rule .
we have used the analytical solutions for the full and truncated moments of the nonsinglet polarised structure function @xmath11 within double logarithmic @xmath1 approximation .
our predictions @xmath156 have been found for different input parametrisations @xmath157 with fixed coupling constant @xmath158 .
these four parametrisations describe different small-@xmath0 behaviour of @xmath159 at @xmath160 : @xmath6 .
we found that the low-@xmath0 contribution to the bsr strongly depends on the input parametrisation @xmath46 .
the percentage value @xmath161 of the total bsr @xmath162 varies from 5.5 for the flat regge input 1 ( @xmath163 ) to almost 28 for the most singular input 4 ( @xmath164 ) .
input parametrisation 2 ( @xmath131 ) , which incorporates latest theoretical knowledge about small-@xmath0 behaviour of @xmath22 driven by @xmath1 terms , gives this @xmath165 about @xmath166 of the total bsr .
comparing our results with the experimental smc data one can see good agreement , particularly for the input 2 , of the small-@xmath0 contribution @xmath167 to the bsr
. however it must be emphasized , that smc data for the low-@xmath0 region suffer from large uncertainties . using smc data for @xmath22 at small-@xmath0 ( @xmath168 , @xmath169 , @xmath170 )
we have also estimated the exponent @xmath5 which governs the low-@xmath0 behaviour of @xmath22 .
thus we have obtained @xmath171 with large uncertainties .
latest theoretical investigations suggest singular small-@xmath0 shape of polarised structure functions : @xmath172 for the nonsinglet case and even @xmath173 for the singlet one .
both these values are indirectly confirmed by fitted experimental hermes data .
basing on these results , similar extrapolations of the spin dependent quark distributions towards the very low-@xmath0 region have been assumed in several recent input parametrisations @xmath174 .
resummation of the @xmath1 terms generates correctly the leading small-@xmath0 behaviour of the polarised structure function but is inaccurate for larger values of @xmath0 .
in order to have reliable theoretical predictions for the polarised structure functions e.g. @xmath11 and to compare them with the real and the extrapolated recent experimental data , one should use unified theoretical description of the evolution of the structure functions .
such unified approach is the formalism , which contains the resummation of the @xmath1 and the lo dglap @xmath33 evolution as well . besides
, in each realistic analysis of the @xmath33 evolution of the structure functions one should take into account the running coupling , where @xmath175 .
latest theoretical investigations imply such introduction of the running coupling , where @xmath176 .
this is more justified in the small-@xmath0 region .
combined @xmath1+lo dglap analysis together with taken into account the running coupling effects should give a correct description of the polarised structure function e.g. @xmath22 in the whole region of @xmath0 .
it is very important because of lack of the experimental data from the very small-@xmath0 region ( @xmath177 ) .
agreement of the theoretical predictions e.g. for the bsr with real experimental data at medium and large @xmath0 may give hope , that for the very interesting small-@xmath0 region the suitable theoretical results are also reliable .
we thank boris ermolaev for constructive remarks and useful comments concerning the running coupling effects in the small-@xmath0 region .
j.d.bjorken , _ phys . rev . _ * 148 * , 1467 ( 1966 ) ; _ phys
. rev . _ * d1 * , 1376 ( 1970 ) .
smc collaboration : d.adams _ et al .
_ , _ phys .
rev . _ * d56 * , 5330 ( 1997 ) .
j.kwieciski , _ acta phys . pol . _
* b27 * , 893 ( 1996 ) .
j.bartels , b.i.ermolaev , m.g.ryskin , _ z. phys . _
* c70 * , 273 ( 1996 ) ; j.bartels , b.i.ermolaev , m.g.ryskin , _ z. phys . _ * c72 * , 627 ( 1996 ) .
p.d.b.collins , _ an introduction to regge theory and high energy physics _ , cambridge university press , cambridge 1977 .
m.glck , e.reya , m.stratmann , w.vogelsang , _ phys .
rev . _ * d63 * , 094005 ( 2001 ) .
g.altarelli , r.d.ball , s.forte , g.ridolfi , _ acta phys .
* b29 * , 1145 ( 1998 ) .
d.kotlorz , a.kotlorz , _ acta phys . pol .
_ * b35 * , 705 ( 2004 ) .
b.i.ermolaev , m.greco , s.i.troyan , _ nucl .
_ * b571 * , 137 ( 2000 ) ; hep - ph/0106317 ; _ nucl
. phys . _ * b594 * , 71 ( 2001 ) ; _ phys .
lett . _ * b522 * , 57 ( 2001 ) ; _ phys .
* b579 * , 321 ( 2004 ) ; hep - ph/0404267 .
smc collaboration : b.adeva _ et al .
_ , _ phys .
lett . _ * b420 * , 180 ( 1998 ) .
hermes collaboration : k.ackerstaff _ et al .
_ , _ phys .
lett . _ * b404 * , 383 ( 1997 ) ; a.airapetian _ et al . _ , _ phys
. lett . _ * b442 * , 484 ( 1998 ) ; k.ackerstaff _ et al . _ , _ phys .
* b464 * , 123 ( 1999 ) .
s.forte , l.magnea , _ phys .
lett . _ * b448 * , 295 ( 1999 ) .
_ , _ phys .
rev . _ * d62 * , 034017 ( 2000 ) .
j.blmlein , h.bttcher , _ nucl .
phys . _ * b636 * , 225 ( 2002 ) .
m.glck , e.reya , m.stratmann , w.vogelsang , _ phys . rev . _ * d53 * , 4775 ( 1996 ) .
t.gehrmann , w.j.stirling , _ phys .
rev . _ * d53 * , 6100 ( 1996 ) .
a.knauf , m.meyer-hermann , g.soff , _ phys .
lett . _ * b549 * , 109 ( 2002 ) .
b.badeek , j.kwieciski , _ phys .
lett . _ * b418 * , 229 ( 1998 ) .
b.ziaja , _ acta phys . pol .
_ * b32 * , 2863 ( 2001 ) . | the small-@xmath0 contributions to the bjorken sum rule within double logarithmic @xmath1 approximation for different input parametrisations @xmath2 are presented .
analytical solutions of the evolution equations for full and truncated moments of the unintegrated structure function @xmath3 are used .
theoretical predictions for @xmath4 are compared with the smc small-@xmath0 data .
rough estimation of the slope @xmath5 , controlling the small-@xmath0 behaviour of @xmath6 from the smc data is performed .
double logarithmic terms @xmath7 become leading when @xmath8 and imply the singular behaviour of @xmath9 .
this seems to be confirmed by recent experimental smc and hermes data .
advantages of the unified @xmath1+lo dglap approach and the crucial role of the running coupling @xmath10 at low-@xmath0 are also discussed .
pacs 12.38 bx |
high precision astrometry is a powerful tool to search for companions to nearby stars .
it also has the potential to discover significant numbers of extrasolar planetary systems . in this paper
we discuss the potential of astrometry as a probe of self - gravitating discs around pre - main sequence stars .
a gaseous disc with sound speed @xmath3 , surface density @xmath4 , and epicyclic frequency @xmath5 is described as self - gravitating if the toomre ( 1964 ) @xmath6 parameter , @xmath7 is of order unity . in discs where self - gravity is important
, the outcome can either be fragmentation into one or more bound objects , or a quasi - steady state in which gravitational instabilities lead to the outward transport of angular momentum .
local simulations suggest that the boundary between these possibilities is set by the ratio of the local dynamical time - scale @xmath8 to the time - scale on which the disc radiates thermal energy .
fragmentation occurs whenever the cooling time @xmath9 , while longer cooling times lead to stable angular momentum transport ( gammie 2001 ) .
circumstantial evidence suggests that self - gravity could play a role in protoplanetary discs as late as the optically visible classical t tauri phase , which lasts for several myr ( haisch , lada & lada 2001 ) .
evidence that relatively old disc may be self - gravitating comes , first , from models of fu orionis outbursts , which require a low efficiency of disc angular momentum transport to reproduce the observed @xmath10 time - scales . if the viscosity is parameterized using the shakura - sunyaev ( 1973 ) prescription , @xmath11 , where @xmath12 is the vertical scale height , fu orionis models suggest a quiescent @xmath13 ( bell & lin 1994 ; bell et al .
1995 ) . for a given accretion rate , small values of @xmath14
imply high surface densities , so that the disc would be self - gravitating at @xmath15 .
second , theory suggests that angular momentum transport ought to be suppressed in cool regions of the disc where the gas is poorly coupled to magnetic fields ( matsumoto & tajima 1995 ; gammie 1996 ; fleming , stone & hawley 2000 ; sano et al .
2000 ; reyes - ruiz 2001 ; sano & stone 2002 ) .
again , this suggests that self - gravity may set in at radii of a few au as the first significant non - magnetic source of angular momentum transport ( armitage , livio & pringle 2001 ) .
ascertaining when self - gravity is at work within the disc requires either the observation of spiral patterns using extremely high resolution imaging , or detection of the astrometric motion of the stellar photocentre induced by the self - gravitating disc .
it has been shown ( adams et al .
1989 ) that self - gravitating perturbations with @xmath16 , can force the central star to move from the centre of mass . in this paper
, we use numerical simulations to quantify the magnitude of the astrometric displacement .
this has previously been studied by boss ( 1998 ) , who simulated a disc with a mass of @xmath17 and found a large motion of the star , of the order of 0.1 au .
this corresponds to milliarcsecond displacements at the distance of the nearest star - forming regions , which would be easily detectable by any of the forthcoming high precision astrometry experiments .
the disc simulated by boss ( 1998 ) , however , was highly unstable , and subsequently fragmented with the formation of substellar objects .
although promising for giant planet formation ( armitage & hansen 1999 ; boss 2000 ) , prompt fragmentation implies that extremely fortuitous timing would be needed for the astrometric detection of self - gravitating discs .
we concentrate instead on marginally unstable discs , which are not vulnerable to fragmentation and could potentially exist around many classical t tauri stars .
the three - dimensional simulations presented here were performed using smooth particle hydrodynamics ( sph ) , a lagrangian hydrodynamics code ( e.g. , benz 1990 ; monaghan 1992 ) .
our implementation allows for the inclusion of point masses and for point mass creation ( bate et al .
1995 ) . in this simulation
the central star is modelled as a point mass onto which gas particles can accrete if they approach to within the sink radius .
although point mass creation is allowed , the discs considered here are stable against fragmentation and the density never reaches values high enough for the creation of a point mass within the disc itself . a great saving in computational time is made by using individual , particle time - steps ( bate et al .
1995 ; navarro & white 1993 ) .
the time - steps for each particle are limited by the courant condition and a force condition ( monaghan 1992 ) . both point
masses and gas use a tree to determine neighbours and to calculate gravitational forces ( benz et al .
1990 ) .
an advantage of using sph for this calculation is that it is not necessary to impose any outer boundary conditions , and the sph particles are free to move to radii greater than the initial disc radius .
the outer edge of the disc is therefore free to distort and modes with @xmath16 will not be affected or artificially driven by the outer boundary conditions ( heemskerk et al . , 1992 ) .
we consider a system comprising a central star , modelled as a point mass with mass @xmath18 , surrounded by a gaseous circumstellar disc with a mass of @xmath19 .
the disc temperature is taken to have an initial radial profile of @xmath20 ( e.g. yorke & bodenheimer 1999 ) and the toomre @xmath6 parameter is assumed to be initially constant with a value of @xmath21 .
a stable accretion disc where self - gravity leads to the steady outward transportation of angular momentum should have a near constant q throughout .
a constant @xmath6 together with equation ( 1 ) then gives a surface density profile of @xmath22 , and hydrostatic equilibrium in the vertical direction gives a central density profile of @xmath23 .
the disc is modelled using 250,000 sph particles , which are initially distributed according to the specified density profile between inner and outer radii of @xmath24 and @xmath25 respectively .
the actual particle positions are chosen randomly , subject to the constraint that the centre of mass ( and momentum ) of the disc is initially coincident with that of the star . to achieve this
, we simply add particles in point symmetric pairs about the central point mass .
the calculations performed here are essentially scale free . in code units , we take @xmath26 , @xmath27 and @xmath28 . to simplify the discusion
we will generally assume a physical mass scale of @xmath29 and a length scale of @xmath30 au .
the circumstellar disc therefore has a mass of @xmath31 , extends from 1 au to 25 au , and surrounds a star of mass @xmath29 .
previous work has shown that the outcome of gravitationally unstable discs depends critically on the treatment of the energy equation ( pickett et al .
2000 ; gammie 2001 ) .
our global simulations adopt the same approach as was used for local models by gammie ( 2001 ) .
we use an adiabatic equation of state , with adiabatic index @xmath32 , and allow the disc gas to heat up due to both pdv work and viscous dissipation . in the absence of cooling , viscous dissipation would act to heat up the disc , increase the toomre @xmath6 parameter , and drive the disc towards stability . cooling is implemented by adding a simple cooling term to the energy equation .
specifically , for a particle with internal energy per unit mass @xmath33 , @xmath34 where @xmath35 is a radially dependent parameter which we specify for each run .
although simple , this approach to the energetics of the disc can be related ( at least approximately ) to the real physics of an accretion disc . for an optically thick disc in equilibrium
the cooling time is the ratio of the thermal energy per unit area to the radiative losses per unit area .
it can be shown ( see e.g. pringle 1981 ) that in such a viscous accretion disc , the cooling time is given by @xmath36 where @xmath37 is the adiabatic index , @xmath38 is the angular frequency , and @xmath14 is the shakura & sunyaev ( 1973 ) viscosity parameter .
based on this , we assume radially dependent cooling times that scale as @xmath8 , @xmath39 with @xmath40 a constant . gammie ( 2001 ) has shown , using a local model , that cooling times of @xmath41 lead to fragmention , while longer cooling times lead to a quasi - stable state .
since we are interested in the possibility of detecting , through the displacement of the central star , the presence of gravitational instabilities in quasi - stable circumstellar discs , we will use @xmath40 values greater than 3 .
we consider a @xmath31 disc with an inner radius of 1 au and an outer radius of 25 au .
the disc is initially gravitationally stable ( @xmath42 at all radii ) and undergoes cooling with a radially dependent cooling time given by @xmath43 .
we consider cooling times of @xmath44 and @xmath45 which , according to gammie ( 2001 ) , should lead to quasi - stable discs that do not fragment .
the @xmath46 and @xmath47 simulations were run for 2.6 and 3.7 orbits of particles at the outer disc edge at @xmath48 .
these run times are short ( 326 years and 467 years respectively for our assumed scaling ) compared to the evolutionary time of the disc ( i.e the viscous time , @xmath49 ) but are sufficient to reach a quasi - steady state in which heating and cooling are locally in balance . for a cooling time of @xmath45 ( figure 1 )
there is no noticeable structure in the disc . for a cooling time of @xmath44 ( figure 2 )
there is significant structure in the disc but , as in gammie ( 2001 ) , there has been no fragmentation .
the filamentary spiral structure which we observe is consistent ( e.g. nelson et al .
1998 ) with previous simulations of discs with masses significantly less than that of the central star .
although it is likely that the @xmath16 mode is the dominant mode driving the stellar motion ( adams et al .
1989 ) , it is not an obviously dominant mode in the disc . figures 3 and 4 show the surface densities of the @xmath50 and @xmath51 simulations at the beginning and end of each run . because we have not attempted to model the inner boundary condition of the disc in any detail , there is rapid accretion and a drop in surface density close to the inner boundary . as discussed below , we find that the main contribution driving displacement of the star comes from material near the outer edge of the disc , so this numerical effect near the inner boundary is not a major concern . apart from these particles with radii between 1 and 2 being accreted onto the central star ,
the surface density profile does not change significantly during the course of the simulation .
figures 5 and 6 shows the toomre @xmath6 parameter at the beginning and end of each simulation .
for @xmath46 the value of @xmath6 at the end of the simulation is smaller than the initial value of @xmath21 but still generally greater than @xmath30 .
this disc is largely stable , although @xmath6 is small enough that some structure ( not noticeable in fig 1 ) may exist . for @xmath52
the final @xmath6 value is close to 1 for radii between 2 and 15 .
the disc is quasi - stable with heating through gravitational instabilities balancing the cooling to give @xmath53 .
figures 7 and 8 show the displacement of the central star from the centre of mass of the star - disc system as a function of time , for the runs with @xmath54 and @xmath55 respectively .
the displacement grows approximately exponentially at early times , before saturating ( laughlin et al .
1997 ) and reaching a plateau at a displacement which depends upon the assumed cooling time . for the more unstable disc , with @xmath55 ,
the saturation level of the stellar displacement is @xmath56 au , while for the more stable disc with @xmath54 the plateau occurs at a lower level of @xmath57 au . for a source at 100 pc
, the above displacements will produce angular displacements of 0.1 milliarcsec ( mas ) and 0.02 mas respectively . in both cases
the star is executing an approximately circular orbit around the centre of mass , with periods of approximately 50 yr ( for @xmath55 ) and 35 yr ( for @xmath54 ) .
instead of the disk extending from 1 - 25 au , the scale free nature of the simulation means that we could equally well assume , without changing the mass scale , that it extends from 4 - 100 au , closer to a more commonly observed disc size ( padgett et al .
the displacement of the central star from the center of mass and the angular displacement would both be 4 times greater , and the period of the orbit would be 8 times greater than that of a 25 au disk .
a planet producing the same displacement with the same period as that calculated here , would have a mass several times less than that of jupiter .
a jupiter mass planet producing the same displacement would be significantly closer to the central star and hence would have a correspondingly smaller period .
it therefore seems unlikely ( unless considering low mass planets at large radii ) that there could be much confusion between astrometric signals caused by planets and those due to disc instabilities .
in addition to a roughly circular motion , gravitational instability in the disc also generates lower amplitude , short - timescale motions .
this is illustrated in figure 9 , which shows a projection of the position of the central star onto the orbital plane for the last 57 years of the simulation with @xmath55 .
the center of mass is located at @xmath58 . as well as
the large orbit taking @xmath59 years there are a number of smaller orbital structures due to the presence of higher order modes in the disc . for our surface density profile ,
most of the mass in the disc lies at relatively large radius ( the enclosed mass scales as @xmath60 ) .
this behaviour is even stronger for the flatter surface density profiles ( typically @xmath61 or @xmath62 ) often considered as appropriate for protoplanetary discs . as expected given this mass distribution , the timescale of the dominant circular motion of the star is of the same order as the dynamical timescale in the _
outer _ disc . assuming this to be true more generally
, we can obtain estimates for the angular displacement @xmath63 and characteristic timescale @xmath64 as a function of the outer radius , @xmath65 , within which a disc is self - gravitating . for a star at distance @xmath66
we find , @xmath67 where we have used the numbers from the @xmath55 run for the estimate .
we expect the exact numbers to depend upon the actual surface density distribution in the disc , and on the mass , so these figures should be regarded as order of magnitude estimates for discs with masses around a tenth of the mass of the star .
we can also compare the displacement caused by a gravitionally unstable disc with that generated by an orbiting planet within the disc . for a planet
whose mass ratio to the central star is @xmath68 , in a circular orbit at radius @xmath69 , the displacement is , @xmath70 i.e. of similar magnitude to the result derived above .
the _ period _ of the stellar oscillation caused by the planet , however , is substantially shorter than that generated by the disc .
we conclude , therefore , that even in the worst case scenario where ( i ) the disc is gravitationally unstable , and ( ii ) the cooling time is within a factor of two of the fragmentation boundary , stellar displacements due to the disc are _ not _ likely to be confused with the signal from a jupiter mass planet . at substantially lower masses , of course ,
confusion would be possible . in this case ,
detection of the substructure in the orbit shown in figure 9 would be necessary to unambiguously determine whether an observed signal was of disc or planetary origin .
the disc mass of @xmath71 m@xmath72 is also quite large , with most t tauri stars having disc masses smaller than that used here ( beckwith et al .
not only will lower mass discs be less likely to be gravitationally unstable ( see equation ( 1 ) ) , the amplitude of the displacement of the central star is likely to be smaller than that obtained here .
this will further reduce the possibility of confusing stellar displacements due to the disc with that due to an orbiting planet .
we have used numerical simulations of discs around young stellar objects to quantify the astrometric displacement of the star caused by a self - gravitating disc . by modelling the energy balance of the disc using a cooling time formalism ( gammie 2001 )
, we have shown that the magnitude of the displacement depends upon how stable the disc is against fragmentation into bound substellar objects . for a cooling time of @xmath44 ( within a factor of two of the fragmentation boundary at @xmath73 ) , a disc mass of @xmath74 , and a self - gravitating disc radius of the order of 10 au , we obtain relatively large displacements in the @xmath2 range ( for a star at a distance of 100 pc ) .
this magnitude of astrometric signal is potentially observable with any of several upcoming high precision astrometric instruments , although the presence of a circumstellar disc is likely to complicate such observations , especially if time dependent perturbations are present within the disc .
although the gravitationally driven instabilities modelled in this work are intrinsically time dependent , their growth saturates ( laughlin et al . 1997 ) and their filamentary nature ( nelson et al . 1998 ) makes their structure approximately azimuthally symmetric .
their presence is therefore unlikely to further complicate observations .
discs with cooling times longer than @xmath44 , which are correspondingly more stable , have significantly smaller stellar motions . a detailed model for the angular momentum transport processes in the disc , together with an assessment of the heating and cooling processes at work ,
would be required to determine whether protoplanetary discs fall into the parameter regime that produces the largest displacements .
our estimate for the angular displacement of the star caused by a self - gravitating disc is an order of magnitude smaller than that found by boss ( 1998 ) .
although there are numerous differences in the initial conditions and numerical techniques used , we believe that the smaller displacement that we obtain primarily reflects our choice of a more stable disc model . by design ,
our disc is permanently stable against fragmentation into substellar objects , whereas the disc simulated by boss ( 1998 ) broke up towards the end of the simulation .
taken together , our results , plus those of boss ( 1998 ) , suggest that a relatively narrow window of self - gravitating disc conditions could lead to small but long - lived astrometric wobbles ( @xmath2 at 100 pc ) .
more dramatic displacements perhaps of the order of a mas at the same distance are also possible , but only as a precursor to fragmentation of the disc into substellar objects .
our simulations suggest that the disc induced wobble could mimic that of a planet a few times smaller than jupiter and orbiting at a large radius , relative to the disc size .
most t tauri stars have disc masses smaller than that used in this simulation and would consequently induce a correspondingly smaller motion of the central star , reducing the likelihood of confusion between stellar motion due to disc instabilities and that due to an orbiting planet .
the disc induced stellar motions are therefore not likely to be a serious obstacle to the astrometric detection of planets orbiting at small radii ( relative to the disc size ) . for planets orbiting at large radii ,
the planet mass which could be mimiced by a disc instability depends on the disc mass ( relative to the central star ) but is likely to be less than a jupiter mass for most t tauri disc systems .
the simulations reported in this paper made use of the uk astrophysical fluids facility ( ukaff ) . | we use high resolution numerical simulations to study whether gravitational instabilities within circumstellar discs can produce astrometrically detectable motion of the central star . for discs with masses of @xmath0 , which are permanantly stable against fragmentation , we find that the magnitude of the astrometric signal depends upon the efficiency of disc cooling .
short cooling times produce prominent filamentary spiral structures in the disc , and lead to stellar motions that are potentially observable with future high precision astrometric experiments . for a disc that is marginally unstable within radii of @xmath1 , we estimate astrometric displacements of @xmath2 on decade timescales for a star at a distance of 100 pc .
the predicted displacement is suppressed by a factor of several in more stable discs in which the cooling time exceeds the local dynamical time by an order of magnitude .
we find that the largest contribution comes from material in the outer regions of the disc and hence , in the most pessimistic scenario , the stellar motions caused by the disc could confuse astrometric searches for low mass planets orbiting at large radii .
they are , however , unlikely to present any complications in searches for embedded planets orbiting at small radii , relative to the disc size , or jupiter mass planets or greater orbiting at large radii .
psfig.sty accretion , accretion discs astrometry planetary systems : protoplanetary discs stars : formation stars : pre - main sequence |
linear diophantine equations and their systems are easy to solve .
there are three possibilities : either a system has no solution , one solution or infinitely many solutions . the case of one solution can be thought of as a special case of infinitely many solutions .
the method to distinguish the particular cases is known ( * ? ? ?
* ch . 1 ) and
the issue can often be decided by inspection ( by a heuristic search for at least one solution ) .
this can be contrasted with the case of a general diophantine equation , or its system , where the decision whether a solution exists belongs to hard problems .
focusing on the linear case from now on , if a system has infinitely many solutions it may be of an interest to investigate the total number of nonnegative solutions .
the answer is necessarily a finite number .
one such a system appeared in the author s recent work @xcite [ eq : diophantine ] @xmath1 where @xmath2 such that @xmath3 is even .
its importance comes from the fact that it is closely related to counting the number of feynman diagrams for a wide class of boson models in interacting quantum field theory .
the linear equations in ( [ eq : diophantine ] ) are one of those cases where for a given @xmath4 one can quickly find a solution and conclude that the number of solutions for @xmath5 is infinite .
it is only slightly less obvious to see whether the system itself has zero or infinitely many solutions ( any of the four equations in imposes a constraint on a solution for the remaining three equations ) . the closed expression for the total number of _ nonnegative _ solutions proved to be a pertinacious problem to pursue and we present its derivation .
we simplify the system by considering @xmath6 ( both even and odd ) but , as will become clear , our counting ( proof ) strategy can be used to count the solutions for different @xmath4 if there is a need for it .
also , by setting @xmath7 the number of nonnegative solutions of the resulting linear system is interesting on its own and turns out to be related to floyd s triangle http://oeis.org/a006003 [ ] and the row , column and diagonal sum of the normal magic square of order @xmath8 ( called the magic constant ) .
the problems related to linear diophantine equations and their systems often appear in the theory of integer programming @xcite , lattice counting @xcite and combinatorics @xcite .
typically , one is interested in finding the solutions of linear equations rather than counting them .
as argued in @xcite , that is not a problem here .
system is simple enough so that all nonnegative solutions can be systematically listed .
alternatively , one can easily cast the system into the smith normal form @xcite and get a generic expression for all solutions .
but the smith form does not seem to provide an easy way of counting the solutions .
there exist several algorithms for lattice point counting which can be used to obtain the same result we got here . for a single - variable problem ( @xmath9 ) one only needs to know the polynomial order and the first few solutions to find the polynomial for any @xmath8 by using , for example , the lagrange method . for multivariate problems , such as the original system
, one can use barvinok s algorithm @xcite or the approach by macmahon called partition analysis @xcite originally developed for other purposes .
these could be called ` black box ' methods or http://www.risc.jku.at/research/combinat/software/ergosum/risc/omega.html[omega ] finding the number of solutions instantaneously . ] and are not the methods used here .
the author s hope is that for the physically relevant problem of many variables @xmath10 ( and for a larger , but similar , system we briefly discuss in section [ sec : open ] ) we will be able to use the symmetries of as well as a certain invariant which were instrumental in finding the number of solutions here .
[ thm : main ] the number of nonnegative solutions of system is given by @xmath11 for @xmath12 and @xmath13 even and @xmath14 for @xmath13 odd . by remapping @xmath15 in and @xmath16 in
we get @xmath17 for @xmath18 and @xmath19 for @xmath20 , showing a certain similarity .
let us recall the definition of the square lattice and all other useful concepts we will use here used here stands for ` define ' . ] .
[ def : lattice ] the _ square lattice _ is the set @xmath21 as a subset of @xmath22 .
the nonnegative even quadrant is the set @xmath23 and the positive odd quadrant is defined as @xmath24 .
the elements of @xmath25 or @xmath26 are referred to as _ vertices _ and the path connecting two neighboring vertices is called a _
segment_. an affine hyperplane is defined as @xmath27 where @xmath28 .
a hyperplane is called _ reciprocal , horizontal _ and _ vertical _ by setting @xmath29 and @xmath30 , @xmath31 and @xmath32 ( in this order ) .
the length of any segment in the even and odd square lattice is two .
this is the reason for a frequent occurrence of the factor of one half in the upcoming lemmas where we count the number of vertices .
[ lem : diosimsystem ] considering @xmath33 in , let @xmath34 and @xmath35 .
then , for @xmath36 and @xmath37 , there exists a nonnegative solution for any @xmath38 and @xmath39 satisfying @xmath40 furthermore , @xmath41 classifies all nonnegative solutions according to whether @xmath42 or @xmath43 and the number of nonnegative solutions for @xmath44 equals the number od solutions for @xmath45 .
finally , any pair @xmath46 consistent with @xmath47 satisfying determines the total number of nonnegative solutions @xmath48 calculated from the following expression : @xmath49}]}.\ ] ] we rewrite as [ eq : diophantinesim ] @xmath50 and add and followed by subtraction from the sum of the last two lines of .
we get @xmath51 we are looking for nonnegative solutions and so the lower bound @xmath52 holds .
then , from we get @xmath53}+\d/2=\ell_{11}+\d/2\geq0 $ ] that becomes .
we also see that a solution of exists for any @xmath37 by rewriting as @xmath54 for @xmath29 , inserting the rhs to the middle expression in and setting @xmath55 .
we get @xmath56 .
since @xmath57}=\ell_{11}$ ] the worst - case scenario is @xmath58 and even in that case the equation can be satisfied by setting @xmath59 .
we also have an upper bound @xmath60}$ ] but there is no guarantee that @xmath61 can take on all the values . from it follows that any such value must be ` matched ' by @xmath53}+{\d/2}$ ] .
hence , we choose from the two competing quantities @xmath62}+{\d/2},\min{[\ell_{33},\ell_{44}]}]}$ ] and considering @xmath36 we arrive at @xmath61 to be upper bounded by .
when is expression minimized by the first argument ?
the question is when @xmath63}$ ] holds .
taking into account both possibilities , @xmath64 and @xmath65 , we obtain the inequality @xmath66 eq .
contains an important piece of information . by searching for all nonnegative solutions we are after all possible nonnegative six - tuples @xmath48 .
naturally , many of them contain the same pair @xmath46 and so we have to find the pairs multiplicities to count all the solutions .
due to eq .
the multiplicity of @xmath67 equals the multiplicity of @xmath61 which is determined by the value of @xmath61 itself .
. provides the greatest value @xmath61 can achieve and tells us where the two possibilities happen . but
@xmath61 is not the multiplicity itself . for an admissible @xmath61
there is @xmath68 of ways @xmath69 sums to @xmath70 in or equivalently in .
summing over all allowed @xmath61 we find the multiplicity factor to be a triangle number http://oeis.org/a000217 [ ] .
so far we considered separately @xmath43 and @xmath45 .
the latter is equivalent to @xmath71 and we can indeed afford to consider only these two cases . this is because linear system is invariant w.r.t .
the relabeling @xmath72 and @xmath73 and the permutation flips the sign of @xmath41 .
we will find the explicit expressions for the number of solutions for @xmath37 in lemma [ lem : countingevenell ] and [ lem : countingoddell ] .
it is convenient to depict the found inequalities in a nonnegative quadrant of a square lattice whose segment has length two as introduced in definition [ def : lattice ] .
the quadrant s axes are identified with @xmath38 and @xmath39 and inequalities and together with the upper bound on @xmath38 and @xmath39 demarcate a polygon whose boundary and interior contain all admissible pairs @xmath74 leading to the solutions of . an area given by inequalities , and will be referred to as a diagonal _ strip _ and it further splits the polygon into several regions .
different rules for calculating the multiplicities hold in different parts of the polygon and a special care will be taken for the degenerate strip when @xmath75 .
[ lem : countingevenell ] given the assumptions of lemma [ lem : diosimsystem ] and for @xmath8 even the number of nonnegative solutions of eqs .
is the sum of the following expressions : [ eq : noofsolseven ] @xmath76 where @xmath77 denote @xmath45 and @xmath43 , respectively , and the subscripts @xmath78 and @xmath79 distinguish between @xmath80 and @xmath75 .
the split into eight cases will become relevant in the proof of theorem [ thm : main ] .
for the same reason , there is no need to evaluate the sums at this point . for @xmath45
it is advantageous to distinguish between the following two cases : @xmath81 and @xmath82 .
the first inequality combined with implies @xmath83 .
since neither of @xmath84 can be greater than @xmath8 it follows that two polygon vertices lie on the quadrant axes ( connected by the line @xmath85 ) .
the same holds for @xmath86 where we separately investigate @xmath87 and @xmath88 .
let us consider @xmath80 first . to count the points in the strip we will use the reciprocal hyperplanes introduced in definition [ def : lattice ] . all points in the even square lattice lie on the reciprocal hyperplanes delimited by and which gives us a very convenient way of labeling _ and _ counting of the hyperplanes : @xmath90 .
inequalities and are saturated when @xmath91 and @xmath92 , respectively .
it follows that there is @xmath93 segments between the intersection points of these two lines with the axis @xmath38 or @xmath39 .
it also means that there is @xmath94 reciprocal hyperplanes .
there are two types of reciprocal hyperplanes .
one type intersects @xmath95 points and the other passes through @xmath96 points .
this can be seen in the following way .
the diagonal strip boundaries intersect @xmath92 at two points whose @xmath38 coordinates are @xmath97 and @xmath98 . their distance ( projected onto the @xmath38 or @xmath39 axis ) is @xmath99 but because these are odd coordinates no solution can lie on any vertical or horizontal line passing through them .
the closest ` even ' points inside the strip are one segment away ( from each ` odd ' point ) and that is how we got the @xmath100 points above .
thus , the neighboring reciprocal hyperplane passes through @xmath96 points . since we counted the number of hyperplanes to be @xmath101
( which is even ) there is @xmath102 of them for both types .
hence the strip contains @xmath103 solutions .
the @xmath104 parameter is set up such that it takes the corresponding values from the interval @xmath90 governing the multiplicity factor .
we will use the vertical hyperplanes to count the number of solutions for the rest of the polygon . in the subset where @xmath105 holds it
is ( see eq . ) @xmath38 according to which the multiplicities are calculated .
the upper diagonal strip boundary intersects the @xmath39 axis at @xmath99 and the line @xmath92 intersects the axis at @xmath106 .
so there is @xmath101 vertices with nonnegative solutions .
every time @xmath38 increases by two we get two points less and from the previous paragraph the maximal value of @xmath38 is @xmath107 .
hence there is @xmath108 vertical axes . for @xmath65
the situation is verbatim where instead of vertical hyperplanes we study horizontal hyperplanes in the mirror image across the diagonal .
hence , the number of solutions reads @xmath109 summing eqs .
and we obtain . for @xmath75
the strip becomes a diagonal line .
the counting with the help of vertical and horizontal line goes through in exactly the same way leading to eq . .
the diagonal solutions are , however , doubly counted since the strip is degenerate and must be subtracted .
this is precisely the first term of eq . .
therefore , for the number of solutions we get . starting with @xmath80 and @xmath105 , the sum @xmath111 is bounded only by together with @xmath112 .
so we insert @xmath113 and @xmath114 to and then the expression @xmath115 counts the number of horizontal steps from the polygon vertex point @xmath116 .
hence , the polygon s shape is now more complicated
there are two more vertices on the line given by @xmath92 .
it is advantageous to let the vertical hyperplanes ( recall that @xmath105 is being considered ) count until they hit @xmath92 and then subtract the inadmissible solutions
those above the ` cut - off ' line @xmath117 .
the cut - off line is always @xmath114 since @xmath39 can reach it but can not go higher ( @xmath112 ) satisfies constraint unless @xmath118 which , however , corresponds to @xmath43 solved as a separate case . ] .
hence , for the number of solutions we get @xmath119 where the first term is identical to and the upper bound in the second sum is given by counting the inadmissible solutions : we set @xmath113 in @xmath92 , find @xmath39 and calculate @xmath120 . so the number of vertices on the @xmath39 axis is @xmath121 leading to the sum s upper bound .
it is also the expression in the parenthesis where the @xmath104 variable is set up such that vertical hyperplanes with the decreasing number of solutions ( by one ) are assigned the correct multiplicity factors ( in the form of the triangle numbers ) as revealed in lemma [ lem : diosimsystem ] .
the factor of two again accounts for the solutions from mirror image situation on the other side of the strip for @xmath122 ( using horizontal hyperplanes ) .
counting in the strip is the same as in .
this is because the cut - off line never violates the points inside the strip .
the cut - off line @xmath114 intersects @xmath92 at @xmath123 .
by inserting this value to the upper diagonal strip boundary @xmath124 we can see that the cut - off line can not even get to the strip boundary .
for @xmath122 we arrive at the same conclusion and so @xmath125 from and together with we get .
the case @xmath75 is again a special instance of thus reducing it to .
let us recall that @xmath43 translates into @xmath127 so now it is advantageous to separately investigate @xmath87 and @xmath128 . similarly to the @xmath45 case , the first inequality implies @xmath83 with the same consequences for the polygon vertices .
contrary to @xmath45 we will use the reciprocal hyperplanes ( just one to be precise ) to count the solutions .
this is because now all the solutions lie on . for @xmath80
the strip defined by becomes a line containing @xmath96 solutions .
we derived the number governing their multiplicity ( see right before ) to be @xmath129 and so the strip contributes with @xmath130 solutions . since @xmath124 intersect at @xmath131 there is @xmath132 points ( lying on ) between the strip boundary and the @xmath39 axis .
the multiplicity is now based on @xmath38 and taking into account doubling from the same argument for @xmath122 the number of solutions lying on reads @xmath133 summing and we get . for @xmath75 the strip intersections with
is just a point and we simply add and resulting in .
the negative contribution removes the overlapping point shared by the @xmath122 and @xmath134 solutions .
the presence of a cut - off line @xmath117 has again no effect on the intersection of the strip and .
as before , the lowest cut - off is @xmath135 and it intersects at @xmath136 , that is , precisely at the intersection boundary given by @xmath124 .
hence the number of solutions is as in and we write @xmath137 .
for a generic @xmath114 we find that the boundary intersects at @xmath138 and so there is @xmath139 points .
the multiplicity is governed by @xmath38 and for @xmath114 we get from @xmath140 .
as we approach the strip , @xmath38 increases by two with each lattice segment .
hence , considering the identical calculation for @xmath65 , we get @xmath141 and so follows .
the case @xmath75 follows as in .
[ lem : countingoddell ] given the assumptions of lemma [ lem : diosimsystem ] and for @xmath8 odd the number of nonnegative solutions of eq . is a sum of the following expressions : [ eq : noofsolsodd ] @xmath142 where @xmath77 denote @xmath45 and @xmath43 , respectively , and the subscripts @xmath78 and @xmath79 distinguish between @xmath80 and @xmath75 .
there does not seem to exist an easy way of applying the even @xmath8 results to the odd case .
the proof , however , bears similarities to the proof of lemma [ lem : countingevenell ] including the split into several ( eight ) cases .
this makes counting easier and also serves for the sake of proof of theorem [ thm : main ] .
one of the cases we need to consider separately is when @xmath75 .
it turns out to be given by the @xmath80 case ( by setting @xmath75 ) like in lemma [ lem : countingevenell ] .
here it is advantageous to distinguish between @xmath143 and @xmath144 .
the first inequality combined with implies @xmath145 .
since neither of @xmath84 can be greater than @xmath8 it follows that two polygon vertices lie on the quadrant axes ( connected by the line @xmath85 ) . when @xmath43 we separately investigate @xmath146 and @xmath147 for the same reason .
an important difference compared to lemma [ lem : countingevenell ] is the location of the positive axes @xmath38 and @xmath39 in the odd square lattice introduced in definition [ def : lattice ] .
the axis @xmath38 will be identified with @xmath148 and @xmath39 with @xmath149 .
the reason is that unlike the even case , the solution - counting vertices in the square lattice lie on the odd coordinates and the smallest odd number is one .
because there is no nonnegative solution lying on a @xmath150 or @xmath151 we will use the shifted coordinate system in the next four subsections .
consider @xmath80 .
we will use the reciprocal hyperplanes to count the solutions in the strip area as they are characterized by @xmath90 shown in lemma [ lem : diosimsystem ] .
exactly as in the proof of lemma [ lem : countingevenell ] ( case @xmath45 and @xmath81 ) we find the diagonal strip intersection to be at two points whose @xmath38 coordinates are @xmath97 and @xmath98 .
again , their distance ( projected onto the @xmath38 or @xmath39 axis ) is @xmath99 and here the analysis starts to differ .
the coordinates @xmath97 and @xmath98 are even so no solution can lie on any vertical or horizontal line intersecting them .
the closest ` odd ' points inside the strip are one segment away ( from each ` even ' point ) and so there is @xmath100 vertices .
consequently , the neighboring reciprocal hyperplane intersects @xmath96 vertices .
since the first and last hyperplane ( given by @xmath91 and @xmath92 , respectively ) counts @xmath100 solutions and the total number of hyperplanes is @xmath101 ( odd ) we get the first two summands of with different upper bounds .
the vertical hyperplanes will be used for the region outside the strip where @xmath105 since the multiplicity factor is given by their @xmath38 coordinate .
the upper diagonal strip boundary intersects the @xmath39 axis at @xmath152 and @xmath153 intersects it at @xmath154 .
so there is @xmath97 vertices with nonnegative solutions .
every time @xmath38 increases by two we get two points less and from the previous paragraph the maximal value of @xmath38 is @xmath107 .
hence there is @xmath155 vertical axes and the last summand of is found ( multiplied by two to account for the mirror case @xmath65 ) .
is obtained by setting @xmath75 .
the situation is very similar to the relation between and so we only highlight a different step .
the first three terms of are the same as in and the last term removes the inadmissible solutions above the cut - off line(s ) ( for @xmath105 and its diagonal mirror image @xmath65 ) . considering @xmath105 , the line @xmath153 intersects the @xmath39 axis at @xmath157 and the distance from the cut - off line @xmath114 is @xmath158 .
so the number of inadmissible vertices on the axis is @xmath159 .
this is the upper bound in the last sum of and the expression in the parenthesis .
the parameter @xmath104 is again set up to properly count the inadmissible nonnegative solutions on the vertical hyperplanes together with their multiplicities .
. is obtained by setting @xmath160 in .
all solutions lie on the reciprocal hyperplane given by @xmath86 .
the strip solutions lie between the points given by the intersection of and @xmath43 whose projection onto the @xmath38 axis equals @xmath101 and @xmath161 .
so the intersection point are odd and therefore containing admissible nonnegative solutions .
their ( projected ) distance is @xmath99 and the strip vertices lie between them ( on @xmath43 ) .
henceforth , there is @xmath162 of them and the multiplicity is calculated from @xmath53}+{\d/2}=\ell_{11}$ ] according to lemma .
this is the first term in . for the second term ,
if @xmath105 , there is @xmath97 segments between the strip upper boundary and the @xmath39 axis and so @xmath163 vertices .
the multiplicity is calculated from @xmath38 and the @xmath104 parameter in the second term of does precisely that .
the factor of two accounts for the @xmath65 situation .
is obtained by setting @xmath160 in .
as before , the cut - off line @xmath114 removes some solutions from @xmath43 ( in the @xmath105 case ) but always outside the strip .
so the first summand of is identical to the first summand of . for the part of @xmath43 outside and on the boundary of the strip we notice that the cut - off line @xmath114 intersects @xmath43 at the point @xmath140 which is @xmath164 segments away from the upper diagonal strip boundary point ( distance measured by projecting onto the @xmath38 axis )
hence there is only @xmath139 admissible vertices on @xmath43 and we recovered the upper bound of the second sum in .
we sum over the multiplicity governed by @xmath38 in this region and that is determined by the @xmath104 variable in the second sum .
as before , for @xmath65 the situation is identical and it brings an overall factor of two . the last expression , eq . , is obtained by setting @xmath160 in .
[ lem : ell1ell22 ] for @xmath165 the number of solutions of diophantine system is equal to the number of solutions for @xmath80 in lemma [ lem : countingevenell ] . that is @xmath166 and similarly for odd @xmath8 , eqs .
, in lemma [ lem : countingoddell ] .
the subscript @xmath167 denotes the case of interest @xmath165 .
invariance w.r.t .
the permutation @xmath168 is another symmetry of .
the permutation swaps with and with and keeps @xmath41 intact .
hence , if @xmath165 we permute and apply lemma [ lem : diosimsystem ] in order to calculate @xmath169 and @xmath170 in lemma [ lem : countingevenell ] and @xmath171 and @xmath172 in lemma [ lem : countingoddell ] . lemmas [ lem : countingevenell ] and [ lem : countingoddell ] counted the solutions for a given @xmath101 and @xmath161 so our task is to sum over all such pairs . because the lemmas are split into several cases we have to adjust the summation procedure accordingly .
basically , all the work is about finding the way to reconcile the condition @xmath80 or @xmath75 with the different investigated cases .
there is a technical assumption we have to make .
we found for @xmath8 even and for @xmath8 odd .
however , the summing slightly differs between @xmath173 and @xmath174 in the former case and @xmath175 and @xmath176 in the latter case .
the results are identical and we present the derivation only for @xmath173 and @xmath175 in order not to overblow the proof . for @xmath45 the inequalities @xmath81 and @xmath80 @xmath177 where in the first sum we set @xmath118 that must be treated separately . in fact
, the first summand equals zero .
for @xmath110 we get @xmath178 the first term starts counting where the first sum of the second term in terminated .
the second term in starts summing where the second sum of the second term in terminated . by summing and
we obtain @xmath179 if @xmath75 the counting is simpler . in the first case we have @xmath180 (
note that @xmath181 is excluded since only @xmath182 is admissible and so it belongs to the @xmath43 case ) and in the second case we continue summing by @xmath183 the sum of the last two expressions reads @xmath184 for @xmath87 and @xmath80 we have @xmath185 while for @xmath128 the first term continues summing where the first sum in left off summing .
the second term continues the work of the second sum of : @xmath186 by summing and we arrive at @xmath187 when @xmath75 we obtain @xmath188 and @xmath189 their sum equals @xmath190 for @xmath45 the inequalities @xmath143 and @xmath80 @xmath191 where in the first sum we set @xmath192 to be treated separately . for @xmath156
we get @xmath193 eqs . and sum to @xmath194 for @xmath75
we obtain @xmath195 ( note that @xmath181 is excluded since then only @xmath182 is admissible and so it belongs to the @xmath43 case ) and in the second case we continue summing : @xmath196 the sum of and equals @xmath197 for @xmath146 and @xmath80 we find @xmath198 similarly to @xmath8 even , for @xmath147 the first term continues summing where the first sum in ended and the second term continues where the second sum of terminated : @xmath199 the sum of and equals @xmath200 when @xmath75 we obtain @xmath201 and @xmath202 with their sum being @xmath203 denoting @xmath44 by @xmath204 , lemma [ lem : diosimsystem ] tells us that @xmath205 and @xmath206 .
lemma [ lem : ell1ell22 ] brings the solutions for @xmath165 : @xmath207 and @xmath208 .
the same holds for odd @xmath8 and we get @xmath209 and @xmath210 .
. is obtained from @xmath211 and eq . from @xmath212
if we set @xmath213 in then it becomes a simpler linear system with an interesting number of nonnegative solutions .
[ prop : floyd ] the number of nonnegative solutions of the following system of linear diophantine equations [ eq : floyd ] @xmath214 is @xmath215 for @xmath8 even and zero for @xmath8 odd . by setting @xmath216
we get @xmath217 eq .
obtained after the rescaling of is the sum of rows , columns or diagonals of a normal magic square of the size @xmath218 ( sometimes called the magic constant ) .
it is also known as the sum of rows in floyd s triangle .
[ lem : floyd ] let @xmath34 for @xmath219 .
then there exists a nonnegative solution of if and only if @xmath220 the direct part follows from summing any two of the three equations and subtracting the third one .
we get three equations of the form @xmath221 since we are looking for @xmath222 it is necessary the following to be true : @xmath223 and @xmath224 .
the first inequality is the rhs of and combining the last two expressions we get the lhs . for the converse we may assume that is violated ( the first or second inequality ) .
then from we immediately see that @xmath5 is negative .
note that the second inequality in may not be saturated for some @xmath225 .
that is , not all @xmath47 satisfying are actually admissible .
this will become relevant in the next proof .
first we show that for @xmath8 odd there is no solution to . in that case @xmath47
are odd as well and by plugging them to we always get the rhs to be an odd number .
but then @xmath5 can not be an integer .
so from now on we focus on @xmath8 even .
since @xmath226 , then from it follows that the number of non - empty nonnegative solutions of are determined by @xmath47 satisfying @xmath227}.\ ] ] to count the number of solutions means to count the number of triples @xmath228 satisfying .
this is because for any such triple we get a triple of @xmath5 via .
contrary to the original system there are no multiplicities and this significantly simplifies the counting .
the rhs of splits into two cases when @xmath87 and @xmath128 .
the counting of admissible triples @xmath228 in is invariant w.r.t .
the relabelling @xmath168 and so the number of solutions for @xmath165 equals that of @xmath80 .
thus , assuming @xmath36 , the number of solutions for the first case of is @xmath229 and in the second case it is @xmath230 in the geometric picture what we did is to count the number of points lying on a line between two endpoints .
we divided by two because the values of @xmath47 are a multiple of two and added one not to omit a boundary point .
two cases must be distinguished .
we use the simple fact that @xmath231 in the following text . considering the previously mentioned symmetry and , there are four cases to investigate in this section . for @xmath75 and @xmath87
we get from @xmath232 similarly , for @xmath75 and @xmath128 we find from @xmath233 the factor @xmath234 comes from finding the ` smallest ' solution for @xmath75 and @xmath128 which is @xmath235 .
then , the distance between @xmath236 and @xmath237 is @xmath238 giving us @xmath234 after dividing by two and adding one ( a boundary point ) .
for @xmath87 and @xmath80 we get @xmath239 the second term counts the number of points lying between @xmath87 and @xmath240 .
the smallest @xmath161 consistent with the last case ( @xmath128 and @xmath80 ) is @xmath241 ( follows from minimally saturating the inequalities : @xmath242 and @xmath243 for @xmath244 and extracting @xmath161 ) . then , for every @xmath161 from @xmath245 to its maximal value ( equal to @xmath8 ) we find the corresponding @xmath101 .
this can be done in the following way . by solving for @xmath101 and inserting it to the second equation
we get @xmath246 ( @xmath244 gives us the previous minimal saturation ) .
since @xmath161 increases by two , the closest allowed value after @xmath247 is @xmath248 . by recalling @xmath249 it follows that there are now three possibilities : @xmath250 and for every increment of @xmath251 by four we add two more solutions .
this gives us the necessary counting and taking into account we may write @xmath252 the inner upper / lower bound is calculated from @xmath253 .
this expression follows after we find the minimal @xmath254 ( again from setting @xmath244 ) .
since the minimal @xmath161 equals @xmath245 and @xmath255 the sum s bounds follow .
we get from @xmath256 .
the factors of two account for the number of solutions for @xmath165 which is equal to the studied case @xmath80 .
we again tacitly use @xmath257 .
the derivation is very similar so let us stress common points and noteworthy differences .
essentially , the main difference comes from the fact that the boundaries @xmath75 and @xmath258 ( leading to the split to four cases ) intersect at @xmath259 which is odd .
but the values of @xmath225 are never odd in our problem and so it is mostly about adjusting the sums bounds ( to be shifted by one to start / end counting at an even point ) .
so for @xmath75 and @xmath87 we now get @xmath260 ( @xmath261 is the last even point consistent with the inequalities ) . for @xmath75 and @xmath128
the smallest solution is now @xmath262 and @xmath263 .
thus @xmath264 in the case @xmath87 and @xmath80 the counting argument goes through exactly like for except that in order to satisfy the inequalities we stop the counting of @xmath101 on the last even number ( which is @xmath265 ) @xmath266 finally , for @xmath128 and @xmath80 , we get as well but to get to the closest admissible even @xmath161 for @xmath174 we have to shift it by one : @xmath267 recalling @xmath268 , we get two minimally saturating solutions ( @xmath269 ) and as before , by increasing @xmath161 by two , two more solutions are always added .
hence @xmath270 where the outer lower bound comes from the rhs of for the minimal solutions .
for the inner bounds we get the same expressions like in but the derivation is modified by realizing that there are two minimal solutions @xmath271 for the minmal @xmath272 .
we again get from @xmath273 .
we conclude this work with several remarks . we found a closed expression counting the number of nonnegative solutions of linear diophantine system of equations for @xmath274 and for its special case of @xmath213 .
the main linear system is motivated by counting the perturbative contributions for an interaction lagrangian in interacting quantum field theory for bosons .
in particular , the number of nonnegative solutions is closely related to counting the feynman diagrams for two interacting fields in the scalar @xmath0 model to an arbitrary perturbative order @xcite . it would be quite interesting to generalize the presented result to a linear system given by generalizing in the following way : @xmath275 for @xmath276 .
this would provide a very general counting method of feynman diagrams for an arbitrary number of interacting fields , to an arbitrary perturbative order and for any scalar @xmath0 model of interacting bosons .
the secondary problem was motivated purely by curiosity as what happens if we simplify the diophantine system and has no bearing to high - energy physics .
unexpectedly , after rescaling @xmath216 and for @xmath8 even , the number of nonnegative solutions of such a system ( eq . ) turn out to be the magic constant of order @xmath8 the sum of all rows , columns and diagonals of a normal magic square of order @xmath218 . | we derive a closed expression for the number of nonnegative solutions of a certain system of linear diophantine equations .
the motivation comes from high energy physics where the nonnegative solutions play a crucial role in the perturbative calculation for a class of lagrangians describing the interaction of an atom with a boson field or a non - linear interaction of boson fields among themselves ( the so - called interacting @xmath0 models ) .
the linear system can be solved and the nonnegative solutions enumerated but a closed expression for the number of solutions is preferable to counting the solutions .
interestingly , the problem led to a construction of a simpler linear diophantine system whose nonnegative number of solutions turns out to be the magic constant .
[ theorem]corollary [ theorem]lemma [ theorem]proposition [ theorem]example [ theorem]conjecture |
the progress of spintronics and physics of heterostructures which can be observed during the last years @xcite is focused on various physical phenomena , and one of them which attracts a considerable attention is the spin - dependent transport through semiconductor / spin - polarized junctions @xcite . the physics of carrier polarization and its influence on transport in composite structures such as semiconductor / ferromagnet has been studied both theoretically @xcite and experimentally @xcite .
one of the models describing the spin - resolved carrier concentrations and currents at the junction is the two - component drift - diffusion model @xcite which predicted highly nonlinear and saturating current - voltage dependence at a single semiconductor / feromagnet junction due to the effect of spin blockade @xcite . in this model
the detailed structure of the charge and current distribution at the junction area @xcite as well as the schottky barriers @xcite , the charge redistribution effects @xcite , and the bound states @xcite are not taken into consideration .
still , the qualitative and distinguishable behaviour of current saturation due to the effect of spin blockade is reliably predicted under various system parameters such as the junction / semiconductor resistance ratio .
the spin blockade regime arises from the spatial distribution of the spin - minority carriers which can not enter the ferromagnet region and form a cloud near the junction which growth prevents the further increase of spin - majority carrier transport if the current exceeds a threshold value called the critical current .
further studies have shown the importance and promising applications of this effect also for non - stationary phenomena such as spin memory effects @xcite .
the models described above were applied mainly to gaas - based semiconductor channels , but is is known that the silicon - based structures are also of big interest for spintronics due to the dominating place of silicon in currently available electronic technologies .
more , the technologies of fabricating the silicon / ferromagnet structures such as si / si : mn formed on a basis of diluted magnetic semiconductors have been intensively developed during the last few years @xcite which makes their future applications in spintronics promising and creates certain questions about the phenomena described above .
is there a spin blockade regime in a silicon / ferromagnet junction at specific values of applied voltage , carrier mobility and concentration ?
if so , what is the critical current density and how it depends on the silicon and ferromagnet parameters such as the carrier polarization in ferromagnets and the conductivity of the semiconductor channel ? how deep can we modulate the current in the channel by manipulating the polarization of emitter or collector ferromagnets relative to each other ? in the present manuscript we study these problems in the framework of a simple but effective model of transport in the spin blockade regime @xcite which we generalize for the case of arbitrary carrier polarizations in the emitting and collecting ferromagnetic regions of the channel as well as for wide range of low- and high - ohmic n - doped and p - doped silicon samples .
it is found that the current can be deeply modulated by changing the spin alignment in the emitter and/or collector ferromagnet since the critical current density is very sensitive to it .
we find the analytical expression for the critical current density and calculate the current - voltage dependencies for various combinations of the channel / contact resistance ratios , as well as for n- and p - type of doping with both high and low concentrations .
the manuscript is organized as follows : in section 2 we derive a model generalizing the description of the spin blockade regime for the two - ferromagnet channel with arbitrary polarizations in the emitter and collector ferromagnets and discuss the properties of the critical current density , in section 3 we plot and discuss the current - voltage characteristics for various combinations of system parameters , and the conclusions are given in section 4 .
the schematic view of the ferromagnet - silicon - ferromagnet channel is shown in figure [ fsif ] .
the collector ferromagnet with the junction resistance @xmath0 is separated from the emitter ferromagnet by a bulk silicon channel with length @xmath1 which we consider as exceeding the spin diffusion length @xmath2 given by @xcite @xmath3 where @xmath4 and @xmath5 are the carrier diffusion coefficient and drift mobility , respectively , @xmath6 is the electric field inside the channel , and @xmath7 ns is the typical spin relaxation time @xcite .
one can see from ( [ ls ] ) that @xmath2 is maximal at zero electric field when @xmath8 and when the diffusion coefficient and spin relaxation time are big , or , since @xmath4 is coupled to the mobility via the einstein relation @xmath9 @xcite , it is clear that the spin diffusion length grows with the mobility . since the mobility in the bulk silicon is typically lower than the one for gaas @xcite , for the given parameters of the silicon the condition @xmath10
is already fulfilled if @xmath11 mkm which is a reasonable channel length of bulk semiconductor structures .
hence , from the point view where the spin - resolved concentration decay length is considered , the distance between emitter and collector ferromagnets here can be taken as infinite which simplifies the boundary conditions .
the key parameters of our model are the variable polarization degrees @xmath12 and @xmath13 in the collector and emitter ferromagnets describing the state of their non - ideality as well as the chosen direction of polarization for the majority of carriers .
for example ,
the pair @xmath14 , @xmath15 corresponds to the previously investigated case @xcite where the carriers are fully polarized in the collector and fully unpolarized in the emitter .
correspondingly , the choice @xmath16 , @xmath17 describes the situation when the collector and emitter are highly polarized in the same direction while the values @xmath16 , @xmath18 describe the opposite polarization at the emitter .
we adopt the labeling @xmath19 for the spin - resolved current with spins in ferromagnets aligned up ( down ) . if @xmath20 is the total current density , than the spin - resolved current at the collector @xmath21 which means that @xmath22 corresponds to an ideal ferromagnet where magnetic moments are all aligned up ( down ) , and @xmath23 describes the unpolarized current .
the same labeling is adopted for the carrier polarization generated by the emitter ferromagnet with polarization @xmath13 .
the main purpose of the present paper is to find out how deep is the influence of the various @xmath12 and @xmath13 differences reflecting the polarization switch on the total electrical current in the channel .
we include in our model the channel and contact zero - bias resistances @xmath24 and @xmath0 ( see figure [ fsif ] ) with variable ratio , and the direction of the carrier motion is labeled as @xmath25 ( for the electrons the actual current is @xmath26 ) .
the spin blockade may occur in the contact region at the collector ( left f in figure [ fsif ] ) if the polarizations @xmath12 and @xmath13 significantly differ from each other , which enhances the contact resistance from the zero - bias value @xmath0 as long as the current builds up . and
@xmath13 in collector and emitter ferromagnets .
the channel and contact zero - bias resistances are @xmath24 and @xmath0 , and the direction of the carrier motion @xmath25 is shown ( for the electrons the actual current is @xmath26 ) .
the spin blockade may occur in the contact region at the collector ( left f ) if the polarizations @xmath12 and @xmath13 significantly differ from each other , which enhances the contact resistance from the zero - bias value @xmath0 as long as the current builds up.,width=321 ] the model for the spin - resolved current densities @xmath27 and spin - resolved concentrations @xmath28 where @xmath29 is the total carrier concentration is well - known @xcite and consists of the continuity equation @xmath30 and the equation for the spin - resolved current @xmath31 these equations should be accompanied by the boundary conditions at the collector contact @xmath32 and at the other boundary of the channel @xmath33 which is in our problem can be considered as @xmath34 . in this case
the steady - state solution of ( [ ce ] ) exists @xcite which has the form of two exponents decaying to their concentrations defined by the boundary conditions at the emitter . in our model the emitter ferromagnet is described by the arbitrary polarization @xmath13 which generalizes the unpolarized case @xmath15 considered previously @xcite , so @xmath35 the steady - state solution of ( [ ce ] ) satisfying the boundary condition ( [ nem ] ) at the emitter and the normalizing condition @xmath28 has the form @xmath36 where @xmath37 is the inverse decay length defined in ( [ ls ] ) .
the parameter @xmath38 should be determined from the boundary condition at the collector when @xmath32 .
if the collector is a non - ideal ferromagnet with the polarization @xmath12 , then the spin - resolved current at the collector @xmath39 and the other boundary condition at the emitter @xmath40 is satisfied automatically .
the total current density @xmath20 is related to the electric field in the channel via usual relation @xmath41 which allows to construct a closed equation for the current density and applied voltage . from ( [ jcoll ] )
it follows that the conductivity of the collector junction is proportional to the concentration of the majority carriers , and at the absence of the current the junction resistance has a predetermined value @xmath0 while the silicon channel with length @xmath1 and cross - section @xmath42 is described by the resistance defined in a usual way by the carrier mobility , concentration and geometric dimensions as @xmath43 . after a simple algebra we obtain the ohmic law @xmath44 which reads in our case similar to the one obtained previously @xcite for a non - polarized emitter with @xmath15 , @xmath45 ^ 2}-1}}\right)=v , \label{cv}\ ] ] but in our generalized model the parameter @xmath46 defined as the critical current density
@xcite depends on both the collector and emitter ferromagnet polarizations @xmath12 and @xmath13 : @xmath47 where @xmath48 is the critical current density for the case of fully polarized collector ferromagnet with @xmath14 and fully unpolarized emitter with @xmath15 @xcite .
indeed , for @xmath14 and @xmath15 the value of @xmath46 from ( [ jc ] ) is equal to @xmath49 . measured in units of @xmath49 on the polarizations @xmath12 and @xmath13 in collector and emitter ferromagnets .
the case of fully polarized carriers in the collector with @xmath14 and fully unpolarized electrons in the emitter @xmath15 considered in @xcite is shown by a reference point a when @xmath50 .
the magnitude of @xmath46 diverges along the line @xmath51 which means that in the case of a perfect coincidence of the spin alignment in the emitter and collector the critical current can never be reached and there is no spin blockade in this limit . in the opposite limit of maximum difference between @xmath12 and @xmath13 the value of @xmath46 is considerably smaller than @xmath49 , so the spin blockade here occurs at lower current densities.,width=321 ] the main property of equation ( [ cv ] ) determining the current - voltage characteristic is the current saturation occurring at current densities comparable to @xmath46 .
thus , it is of interest to investigate the critical current density ( [ jc ] ) in more details with respect to the tunable polarization parameters @xmath12 and @xmath13 of the collector and emitter ferromagnets . in figure [ fjc ]
the three - dimensional plot @xmath52 is shown for the critical current density ( [ jc ] ) measured in units of @xmath49 .
the case of fully polarized carriers in the collector with @xmath14 and fully unpolarized electrons in the emitter @xmath15 considered in @xcite is shown by a reference point a when @xmath50 .
it is clear from ( [ jc ] ) that @xmath46 diverges along the line @xmath51 which means that in the case of a perfect coincidence of the spin alignment in the emitter and collector the critical current can never be reached and there is no spin blockade in this limit . in the opposite limit of maximum difference between @xmath12 and @xmath13 the value of @xmath46 is considerably smaller than @xmath49 , so the spin blockade here occurs at lower current densities .
the observed large variations of the critical current density are important for the desired high degree of control on the electrical current flowing through the ferromagnet - silicon - ferromagnet channel which can be achieved by variations of at least one ferromagnet polarization , say the emitter polarization @xmath13 .
below we shall see that these expectations are confirmed by the current - voltage dependencies in both n - doped and p - doped channels with both low- and high - ohmic resistance .
the current - voltage characteristics for n - doped and p - doped silicon channels with high and low carrier concentration are shown in figures [ fvah1 ] and [ fvah2 ] , respectively .
mkm and @xmath53 @xmath54 shown in figure [ fsif ] with ( a ) n - doped and ( b ) p - doped silicon with high carrier concentration @xmath55 @xmath56 . on each plot
the three families of curves correspond to three positions of the `` spin valve '' emitter while the carriers at the collector are always highly polarized with @xmath16 .
each family has two curves labeled 1 and 2 ( bold ) which corresponds to the low and high ratios of the junction and channel resistances @xmath57 and @xmath58 .
the solid curves n1 , n2 describe the case of non - polarized carriers at the emitter with @xmath59 which was considered in @xcite , the dashed curves p1 , p2 are for the emitter polarization @xmath17 which is close to the collector polarization @xmath16 , and the dash - dotted curves a1,a2 are for the emitter polarization @xmath18 which differs significantly from the one of the collector .
for both type of carriers and for both low and high ratios of the contact / junction resistance the current saturation is observed being the indication of the spin blockade regime .
the switch between different emitter polarizations leads to strong modulations of the current saturation amplitude , creating the possibility of the magnetization control on the current.,width=321 ] the data for the room temperature drift mobility of electrons and holes in the bulk silicon doped with various concentrations was taken from the standard reference plots @xcite , allowing to estimate both channel resistance for a bulk silicon sample with @xmath60 mkm and @xmath53 @xmath54 and the reference critical current density @xmath49 from ( [ jc0 ] ) . in figure [ fvah1 ] the results are shown for the silicon channel with high carrier concentration @xmath55 @xmath56 where the reference critical current density @xmath49 is within the range of @xmath61 @xmath62 , and the channel resistance @xmath63 ohm . on each plot
the three families of curves correspond to three positions of the `` spin valve '' emitter while the carriers at the collector are always highly polarized with @xmath16 .
each family has two curves labeled 1 and 2 ( bold ) which corresponds to the low and high ratios of the junction and channel resistances @xmath57 and @xmath58 .
the solid curves n1 , n2 describe the case of non - polarized carriers at the emitter with @xmath59 which was considered in @xcite , the dashed curves p1 , p2 are for the emitter polarization @xmath17 which is close to the collector polarization @xmath16 , and the dash - dotted curves a1,a2 are for the emitter polarization @xmath18 which differs significantly from the one of the collector . for both type of carriers and for both low and high ratios of the contact / junction resistance ratio the current saturation
is observed being the indication of the spin blockade regime .
the switch between different emitter polarizations leads to strong modulations of the current saturation amplitude , creating the possibility of the magnetization control on the current .
but for the ( a ) n - doped and ( b ) p - doped silicon channels with low carrier concentration @xmath64 @xmath56 .
the labeling of curves is the same as in figure [ fvah1 ] , and the maximum achievable currents are lower due to significantly smaller carrier concentration which decreases the conductivity .
the same conclusions about the spin blockade manifestation and polarization control of the current amplitudes mentioned in the caption for figure [ fvah1 ] can be applied here for the case of low carrier concentrations.,width=321 ] to compare the results for low - ohmic samples with the high - ohmic ones , we present in figure [ fvah2 ] the current - voltage characteristics for the silicon channel doped with low carrier concentration @xmath64 @xmath56 where the reference critical current density is much lower and is within the range of @xmath65 @xmath62 .
the channel resistance is correspondingly much higher , and we consider @xmath66 kohm for n - doped and p - doped samples , respectively .
the labeling of all curves is the same as in figure [ fvah1 ] , and the maximum achievable currents are much lower due to significantly smaller carrier concentration which decreases the conductivity .
one can draw the same conclusions about the spin blockade manifestation and polarization control of the current amplitudes mentioned above which obviously can be applied also for the case of low carrier concentrations .
we can conclude that the main goal of the present model which is the achievement of a deep current modulation by the magnetization switch at the emitter can be reached in both low- and highly - doped silicon samples , with both n - type and p - type doping and with both low and high ratios of the junction / channel resistances .
hence , the proposed model of the tunable current - voltage characteristics in a ferromagnet - silicon - ferromagnet channel seems to be applicable to a rather wide range of ferromagnet - semiconductor structures .
we have studied the steady - state current - voltage characteristics of ferromagnet - silicon - ferromagnet channels with long bulk silicon sample having the length exceeding the spin diffusion length .
the current behaviour was investigated in the presence of spin blockade regime at the collector junction , and the dependence of the critical current on both collector and emitter polarizations has been obtained analytically .
it was found that the current amplitude can be effectively tuned by varying the difference between the collector and emitter ferromagnet polarizations which allows to perform the magnetic manipulation on the electrical current in wide class of both n- and p - doped , low- and high - ohmic semiconductor channels coupled to ferromagnetic leads .
the author is grateful to e.s .
demidov for many helpful discussions .
the work was supported by the ministry of education and science rf under the program `` development of the higher school research potential '' ( grant no .
2.1.1/2833 ) . | the steady current - voltage characteristics of ferromagnet - silicon - ferromagnet channels with tunable emitter and collector polarizations are investigated in the presence of spin blockade generalizing the model developed by pershin yu v and di ventra m ( 2008 _ phys .
rev .
_ b * 77 * 073301 ) .
the dependence of the critical current on both collector and emitter polarizations is obtained analytically .
it is found that the current amplitude in the channel can be effectively tuned by varying the difference between the collector and emitter ferromagnet polarizations which allows to perform the magnetic manipulation of the electrical current in wide class of both n- and p - doped , low- and high - ohmic semiconductor channels coupled to ferromagnetic leads . |
the 4f - electronic systems with spin and orbital degrees of freedom in rare earth compounds frequently reveal electric quadrupole orderings in addition to magnetic dipole orderings at low temperatures .
the cubic compounds based on ce@xmath18 ion with a @xmath8 quartet ground state in particular have received much attention because the competitive phenomena associated with magnetic dipole , electric quadrupole and magnetic octupole degrees of freedom are expected .
the direct product of @xmath19 is reduced to a direct sum @xmath20 .
the magnetic dipole @xmath21 , @xmath22 , @xmath23 belonging to @xmath24 symmetry are order parameters for magnetic orderings .
the quadrupole orderings of @xmath10 , @xmath25 with @xmath26 or @xmath27 , @xmath28 , @xmath29 with @xmath13 are interesting phenomena in the @xmath8 system .
we refer to ceag exhibiting the ferroquadrupole ( fq ) ordering of @xmath10 at @xmath30 k.@xcite ceb@xmath2 is known as the antiferroquadrupole ( afq ) ordering of @xmath29-type with the propagation vector of = [ 111 ] at @xmath31 k.@xcite the octupole moments @xmath32 with @xmath33 symmetry , @xmath34 , @xmath35 , @xmath36 with @xmath24 and @xmath37 , @xmath38 , @xmath39 with @xmath13 may play a role in the @xmath8 system.@xcite cage of 4a site with o@xmath40 symmetry consisting of ge and pd2 atoms and cage of 8c site with t@xmath41 of pd1 and pd2 atoms in clathrate compound ce@xmath0pd@xmath1ge@xmath2 . the 4a site ce2 forms a simple cubic lattice , while 8c site ce1 makes a face centered cubic one .
the ferroquadrupole ordering below @xmath11 is relevant for the 8c sites and the antiferromagnetic ordering below @xmath42 occurs at 4a sites .
the @xmath13 rattling motion originates from the off - center ce1 atom in 4a - site cage . ]
a cerium - based ternary compound ce@xmath0pd@xmath1ge@xmath2 with the @xmath8 ground state has received much attention because the competition between quadrupole and magnetic orderings is expected at low temperatures.@xcite ce@xmath0pd@xmath1ge@xmath2 crystallizes in a cubic cr@xmath3c@xmath2-type structure with a space group @xmath43 consisting of four molecular units with 116 atoms in a unit cell.@xcite the twelve ce sites located in cages are divided into two nonequivalent sites in crystallography . as shown in fig .
[ eps01 ] the ce ion at 4a site in a cage consisting of twelve pd - atoms and six ge atoms possesses point group symmetry o@xmath40 , while the ce ion at 8c site in a cage of sixteen pd atoms has t@xmath41 .
the 4a sites form a face - centered cubic lattice , while the 8c sites make a simple cubic lattice .
inelastic neutron scattering on ce@xmath0pd@xmath1ge@xmath2 revealed overlapping two peaks for the crystalline electric field ( cef ) potentials , which correspond to magnetic dipole transitions from the @xmath8 ground quartet to the @xmath44 excited doublet at 60 k of the 4a site and from the @xmath8 ground quartet to the @xmath44 at 46 k of 8c site.@xcite the entropy obtained by low - temperature specific heat measurement on ce@xmath0pd@xmath1ge@xmath2 also indicates the ground state @xmath8 quartet at both 4a and 8c sites.@xcite the low - temperature specific heat of ce@xmath0pd@xmath1ge@xmath2 shows a rounded small peak at @xmath7 k and a sharp @xmath45-peak at @xmath12 k.@xcite magnetic susceptibility shows a clear cusp at @xmath42 , but exhibits no sign of anomaly at @xmath11.@xcite in addition to these experimental results , an elastic softening of @xmath4 in our preliminary paper suggests that the paramagnetic phase i transforms to the fq phase ii at @xmath11 and successively changes to the antiferromagnetic ( afm ) phase iii at @xmath42.@xcite the neutron scattering on ce@xmath0pd@xmath1ge@xmath2 reveals a paramagnetic state of ce ions at both 4a and 8c sites even in phase ii between @xmath11 and @xmath42 .
the afm ordering in phase iii with a propagation vector @xmath46 $ ] for cerium ions at 4a site is observed below @xmath42.@xcite even in phase iii below @xmath42 , the 8c site still remains to be the paramagnetic state .
the afm ordering with incommensurate structure at 8c site appears only below @xmath47 k. the clathrate compounds exhibiting the rattling motion or off - center motion in a cage have received attention because their remarkable reduction of thermal conductivity is favorable for application to thermoelectric device with a high figure of merit.@xcite the ultrasonic waves are scattered by the rattling motion in an over - sized cage of a semiconductor sr@xmath48ga@xmath49ge@xmath50 and a filled skutterudite compound pros@xmath51sb@xmath52.@xcite the off - center tunneling motion of oh ion doped in nacl gives rise to elastic softening at low temperatures.@xcite the rattling motion in the present compound ce@xmath0pd@xmath1ge@xmath2 with clathrate structure has not been reported so far . in the present paper
we show ultrasonic measurements on ce@xmath0pd@xmath1ge@xmath2 in order to examine lattice effects associated with the quadrupole ordering and rattling motion in the system . the thermal expansion measurement is also employed to detect the spontaneous distortion below @xmath11 . in sec .
ii , the experimental procedure and apparatus are described .
the results of the elastic constant , magnetic phase diagram , thermal expansion are presented in sec .
the ultrasonic dispersion due to rattling motion is also argued in sec .
iii . in sec .
iv , we present concluding remarks .
single crystals of ce@xmath0pd@xmath1ge@xmath2 used in the present measurements were grown by a czochralski puling method . we have made the ultrasonic velocity measurements using an apparatus consisting of a phase difference detector .
piezoelectric plates of linbo@xmath0 for the ultrasonic wave generation and detection are bonded on plane parallel surfaces of sample .
x_-cut plate of linbo@xmath0 is available for transverse ultrasonic waves and the 36@xmath53_y_-cut plate is for longitudinal waves .
the ultrasonic velocity _ v _ was measured by fundamental frequencies of 10 mhz and overtone excitations of 30 , 50 and 70 mhz . in the estimation of the elastic constant @xmath54
, we use the mass density @xmath55 g/@xmath56 for ce@xmath0pd@xmath1ge@xmath2 with a lattice parameter @xmath57 .@xcite a homemade @xmath58he - refrigerator equipped with a superconducting magnet was used for low - temperature measurements down to 450 mk in magnetic fields up to 12 t. a @xmath58he-@xmath59he dilution refrigerator with a top - loading probe was used for the ultrasonic measurements in low - temperature region down to 20 mk in fields up to 16 t. low input - power condition provides the low - temperature ultrasonic measurements free from a self - heating effect in the ultrasonic transducers . the sample length as a function of temperature or applied magnetic field
was measured precisely by a capacitance dilatometer in the @xmath58he - refrigerator .
the elastic constants of @xmath60 and @xmath61 of ce@xmath0pd@xmath1ge@xmath2 in fig . [ eps02 ] were measured by the longitudinal ultrasonic waves with frequencies 10 mhz propagating along the [ 100 ] and [ 110 ] directions , respectively .
the elastic constant @xmath4 of ce@xmath0pd@xmath1ge@xmath2 in fig .
[ eps03 ] was measured by the transverse ultrasonic wave of 10 mhz propagating along the [ 110 ] direction polarized to the @xmath62 $ ] one .
the elastic constant @xmath5 of ce@xmath0pd@xmath1ge@xmath2 in fig .
[ eps03 ] was determined by the transverse wave of 30 mhz propagating along [ 100 ] polarized to [ 010 ] .
the bulk modulus @xmath63 in fig .
[ eps02 ] was calculated by @xmath60 in fig .
[ eps02 ] and @xmath4 in fig .
[ eps03 ] .
it is remarkable that @xmath4 exhibits a huge softening of 50% with decreasing temperature down to @xmath7 k. in phase ii below @xmath11
the ultrasonic echo signal of the @xmath4 mode completely disappears due to a marked ultrasonic attenuation .
the softening of the longitudinal @xmath60 and @xmath64 modes in fig .
2 originates from the softening of @xmath4 , because @xmath60 and @xmath64 involve @xmath4 in part . the softening of @xmath4 above @xmath11 and the spontaneous tetragonal distortion below @xmath11 , that will be shown in sec .
iii d , provide evidence for the fq ordering in phase ii .
the @xmath5 in fig .
[ eps03 ] also exhibits a softening of 2.5% down to @xmath11 .
the low - temperature behavior of @xmath60 and @xmath5 shown in insets of figs .
[ eps02 ] and [ eps03 ] indicates the transition to the fq phase ii at @xmath11 and successive transition to the afm phase iii at @xmath12 k. on the other hand , @xmath65 shows monotonic increase with decreasing temperature .
neutron scattering on ce@xmath0pd@xmath1ge@xmath2 revealed the para - magnetic state for ce ions at both 4a and 8c sites in phase ii , which is consistent with the present scenario of the fq ordering at 8c site in phase ii below @xmath11.@xcite the afm ordering in phase iii at 4a site below @xmath42 has been detected by the neutron scattering .
it has been proposed that the inter - site quadrupole interaction among 8c sites brings about the fq ordering at 8c sites in phase ii and ce ions at 4a sites still remain to be the para - state even in phase ii .
the inter - site magnetic interaction among 4a sites gives rise to the afm ordering in phase iii below @xmath42 .
the magnetic ordering at 8c sites appears only below 0.4 k. we discuss about this transition in the following sec .
iii c. in order to analyze the elastic softening of @xmath4 and @xmath5 in ce@xmath0pd@xmath1ge@xmath2 of fig .
[ eps03 ] , we introduce the coupling of the quadrupole @xmath66 of ce ions to the elastic strain @xmath67 as @xcite @xmath68 where the summation @xmath69 takes over ce ions in unit volume and @xmath70 is a coupling constant .
the inter - site quadrupole interaction mediated by phonons and conduction electrons is written in a mean field approximation as @xmath71 where @xmath72 denotes a mean field of the quadrupole and @xmath73 means a coupling constant for the inter - site quadrupole interaction . by differentiating the total free energy consisting of 4f - electron and lattice parts with respect to the elastic strain @xmath67
, we obtain the temperature dependence of the elastic constant @xmath74 as @xcite temperature dependence of the elastic constants @xmath60 , @xmath64 and the bulk modulus @xmath65 of ce@xmath0pd@xmath1ge@xmath2 .
inset shows the anomalies of @xmath60 around the ferroquadrupole transition @xmath7 k and the antiferromagnetic transition @xmath12 k. ] temperature dependence of the elastic constants @xmath4 and @xmath5 corresponding to shear waves in ce@xmath0pd@xmath1ge@xmath2 .
inset shows the anomalies of @xmath5 around the ferroquadrupole transition @xmath7 k and the antiferromagnetic transition @xmath12 k. solid lines are the calculation by the quadrupole susceptibility for the @xmath8 ground state and @xmath44 state at 46 k of ce ions .
the broken lines were back ground @xmath75 as shown in the text .
a shoulder in @xmath5 around 30 k means the ultrasonic dispersion due to the @xmath13 rattling motion . ]
@xmath76 here @xmath77 denotes a background elastic constant without the quadrupole - strain interaction and _ n _ is the number of ce ions in unit volume . the quadrupole susceptibility of @xmath78 in eq .
( 3 ) is written as @xmath79 where @xmath80 is a second - order perturbation energy with respect to @xmath67 for cef state .
@xcite the first part in right hand of eq .
( 4 ) corresponds to the van vleck - term and the second part to the curie term .
the ce ions at both 4a and 8c sites in ce@xmath0pd@xmath1ge@xmath2 have the @xmath8 ground state , while the @xmath44 state has excited energies of 46 k at 8c site and 60 k at the 4a site .
as was already mentioned , the neutron scattering revealed that the fq ordering of 8c sites occurs at @xmath7 k and the afm ordering of 4a sites appears at @xmath12 k. these facts indicate that the inter - site quadrupole interaction of eq .
( 2 ) among the ce ions at 8c sites dominates the softening of @xmath4 as a precursor of the fq ordering at @xmath11 . in the following analysis
we simply assume the quadrupole - strain interaction of eq .
( 1 ) and quadrupole interaction of eq .
( 2 ) for the 8c site with the cef splitting of @xmath8 ( 0 k ) and @xmath44 ( 46 k ) .
the solid lines for @xmath4 and @xmath5 with eq .
( 3 ) in fig .
[ eps03 ] reproduce well the softening in paramagnetic phase i above @xmath11 .
it should be noted that the softening above @xmath11 proportional to the reciprocal temperature @xmath6 originates from the curie term of eq .
the coupling constants were determined to be @xmath81 k , @xmath82 k for @xmath4 and @xmath83 k , @xmath84 k for @xmath5 .
the back ground @xmath85 j / m@xmath58 and @xmath86 j / m@xmath58 indicated by broken lines in fig .
[ eps03 ] was used .
the positive value of @xmath87 are consistent with the fq ordering in ce@xmath0pd@xmath1ge@xmath2 . a shoulder like anomaly in @xmath5 around 10 k results from ultrasonic dispersion that is caused by a rattling motion of the rare - earth ion at 4a site in an oversized cage of fig .
[ eps01 ] .
we discuss about this remarkable behavior in sec .
iii e. in order to examine the magnetic phase diagrams of the fq and afm orderings in ce@xmath0pd@xmath1ge@xmath2 , we have made the low - temperature ultrasonic measurements of @xmath60 , @xmath5 and @xmath4 under magnetic fields . the softening of @xmath60 of fig .
[ eps04 ] reduces with increasing fields applied along the [ 001 ] direction parallel to the propagation direction of longitudinal wave .
the fq transition points @xmath11 indicated by downward arrows in fig .
[ eps04 ] shift to higher temperatures and become indistinct in high fields up to 6 t. in fig .
[ eps05 ] , the fq transition points @xmath11 also shift to higher temperatures accompanied by reduction of the softening in @xmath5 with increasing applied fields along the [ 001 ] direction . low - temperature behavior of @xmath60 of ce@xmath0pd@xmath1ge@xmath2 under magnetic fields along the [ 001 ] direction .
downward arrows indicate the ferroquadrupole ordering temperature @xmath11 and upward arrow means the antiferromagnetic ordering temperature @xmath42 . ]
low - temperature behavior of @xmath5 of ce@xmath0pd@xmath1ge@xmath2 under magnetic fields along [ 001 ] .
downward arrows indicate the ferroquadrupole ordering temperature @xmath11 and upward arrow means the antiferromagnetic ordering temperature @xmath42 . ]
low temperature behavior of @xmath4 of ce@xmath0pd@xmath1ge@xmath2 under magnetic fields along the [ 001 ] direction .
the successive phase transitions i - ii - iii are indicated by arrows . ]
low temperature behavior of @xmath4 of ce@xmath0pd@xmath1ge@xmath2 under magnetic fields along the [ 110 ] direction .
the successive phase transitions i - ii - ii@xmath88 are indicated by arrows . ] in figs .
[ eps06 ] and [ eps07 ] , we show the field dependence of @xmath4 applying fields along the [ 001 ] and [ 110 ] directions , respectively . in zero magnetic field
, the @xmath4 mode exhibits the considerable softening of 50% and the strong ultrasonic attenuation losing the echo signal in the vicinity of the fq transition @xmath7 k. the magnitude of the softening decreases abruptly with increasing fields along both [ 001 ] and [ 110 ] directions . in magnetic fields , clear minima of @xmath4 corresponding to the transitions @xmath11 from the paramagnetic
phase i to the fq phase ii shift to higher temperatures . only broad round anomalies around @xmath11 have been observed in high fields of 5 t. this behavior of @xmath11 is similar to the results of the fq ordering accompanied by the soft @xmath5 mode in hob@xmath2.@xcite the neel temperature @xmath42 in fig .
[ eps06 ] shifts slightly to lower temperatures with increasing magnetic fields . in fig .
[ eps07 ] the anomalies associated with the transition between phase ii and ii@xmath88 below @xmath89 have been found . for the investigation of low - temperature and high - field behavior in fq ii and afm iii phases , we have measured field dependence of the @xmath5 and @xmath61 employing the dilution refrigerator . in fig . [ eps08 ]
we show @xmath5 versus @xmath90 at 30 mk in fields up to 12 t applied along [ 001 ] .
inset of fig .
8 is expanded view below 2.5 t. an anomaly of the phase ii - iii boundary at 2.1 t indicated by a vertical line has been observed .
furthermore , several anomalies at 0.5 , 1.2 and 1.6 t associated with sub - phases of the phase iii have been found .
it should be emphasized that appreciable hysteresis phenomena between increasing and decreasing field sequences have been found only in phase iii . in fig .
[ eps09 ] we show the low - temperature field dependence of the @xmath64 in fields up to 16 t applied along [ 110 ] .
low - field behavior below 2.5 t is shown in inset of fig .
we have observed a new phase boundary around 8.2 t in phase ii , which is probably a sub - phase of the fq phase ii .
however , this phase boundary is absent in fields along [ 001 ] as shown in fig .
[ eps08 ] . at low field in phase iii
, we have found several anomalies in @xmath64 of fig .
[ eps09 ] showing a hysteresis behavior .
as can been seen in inset of fig .
[ eps09 ] , this hysteresis becomes pronounced with decreasing temperature .
these sub - phases with hysteresis behavior in phase iii in magnetic fields along both [ 001 ] and [ 110 ] are well consistent with the results of neutron scattering experiments that detect weak incommensurate magnetic bragg peaks with a propagation vector = [ 0 0 1-@xmath91 , ( @xmath92 0.06 ) at 8c site.@xcite the magnetic phase diagrams of ce@xmath0pd@xmath1ge@xmath2 in figs .
[ eps10 ] and [ eps11 ] are obtained in fields along the [ 001 ] and [ 110 ] directions , respectively .
we present the results of the ultrasonic measurements together with the results of thermal expansions in sec .
iii d. it is of importance that the fq phase ii is stabilized in fields for the [ 001 ] direction in fig .
[ eps10 ] and the [ 110 ] direction in fig .
[ eps11 ] .
the fq sub - phase ii@xmath88 was added to the phase diagram and the upper limit at 8.2 t of the phase ii@xmath88 newly determined in fields along [ 110 ] of fig .
[ eps11 ] .
however , the fq sub - phase is absent in fields along [ 001 ] of fig .
[ eps10 ] .
this result indicates strong anisotropy of the quadrupole interaction of @xmath10 in ce@xmath0pd@xmath1ge@xmath2 .
field dependence of the @xmath5 at 30 mk in fields along [ 001 ] up to 12 t. inset is expanded view below 2.5 t indicating the magnetic transitions . ]
field dependence of the @xmath61 at various temperatures in fields along [ 110 ] up to 16 t. inset is expanded view below 2.5 t indicating the magnetic transitions ] magnetic phase diagram of ce@xmath0pd@xmath1ge@xmath2 under fields along the [ 001 ] direction . the boundary from paramagnetic phase i to the ferroquadrupole phase ii shifts to higher temperatures with increasing fields , while the boundary from phase ii to the antiferromagnetic phase iii shifts to lower temperatures in fields .
] magnetic phase diagram of ce@xmath0pd@xmath1ge@xmath2 under fields along the [ 110 ] direction . the sub - phase ii@xmath88 exists in the afq phase ii . ]
the series of r@xmath0pd@xmath1x@xmath2 compounds usually show two successive afm orderings of 8c site at @xmath93 with a propagation vector @xmath94 $ ] and of 4a site at @xmath95 with @xmath96 $ ] , nd@xmath0pd@xmath1ge@xmath2 ( @xmath97 k , @xmath98 k ) @xcite , nd@xmath0pd@xmath1si@xmath2 ( @xmath99 k , @xmath100 k ) @xcite , tb@xmath0pd@xmath1si@xmath2 ( @xmath101 k , @xmath102 k ) @xcite , dy@xmath0pd@xmath1si@xmath2 ( @xmath103 k , @xmath104 k ) @xcite and so on .
one can reasonably expect that the transition temperature @xmath93 at 8c site is always higher than @xmath42 at 4a site since the distance @xmath105 6.2 between rare - earth ion of 8c site is much shorter than the one @xmath105 8.8 of 4a site . in the present ce@xmath0pd@xmath1ge@xmath2 , at first the fq ordering at 8c site with a structural change from cubic lattice to tetragonal one occurs at @xmath7 k. therefore , the afm ordering at 8c site is hard to take place because the favorable propagation vector @xmath94 $ ] of 8c site does not match to the tetragonal lattice in phase ii .
in other words , the afm ordering at 8c site is replaced by the fq ordering in ce@xmath0pd@xmath1ge@xmath2 . while , the afm ordering of 4a site with a propagation vector @xmath96 $ ] is easy to occur even in tetragonal structure below @xmath42 .
neutron experiments detected large enough value of saturation cerium moments @xmath106/ce that is expected from the ground state quartet @xmath8 perpendicular to the @xmath46 $ ] in ce@xmath0pd@xmath1ge@xmath2 far below @xmath42 at 50 mk.@xcite in order to examine the structural change due to the fq ordering at @xmath7 k , we have measured the thermal expansion along the [ 001 ] direction in ce@xmath0pd@xmath1ge@xmath2 .
the sample lengths along the [ 001 ] and [ 111 ] directions are written by the symmetry strains as @xmath107}=\varepsilon_{zz}=\varepsilon_{\rm b}/3+\varepsilon_u/\sqrt3 $ ] and @xmath108}=\varepsilon_{\rm b}/3 + 2(\varepsilon_{yz}+\varepsilon_{zx}+\varepsilon_{xy})/3 $ ] . here
, @xmath109 is a volume strain with @xmath110 symmetry , @xmath111 is a tetragonal strain with @xmath26 and @xmath112 is a shear strain with @xmath13 . the length along [ 001 ] in fig .
[ eps12 ] shows a monotonous decrease with decreasing temperature in paramagnetic phase i above @xmath11 and abruptly expands about @xmath9 below @xmath11 . the thermal expansion along [ 001 ] in phase ii and the huge softening of 50% in @xmath4 of fig .
[ eps03 ] indicate the @xmath10-type fq ordering accompanied by the structural transition from cubic lattice to tetragonal one with the spontaneous strain @xmath113 in phase ii .
this spontaneous strain is proportional to the order parameter as @xmath114 in mean - field approximation .
below @xmath12 k , the @xmath115 along [ 001 ] slightly shrinks .
inset of fig .
[ eps12 ] is expanded view of @xmath115 and the coefficient of the thermal expansion @xmath116 at low temperatures
. a sharp anomaly in the coefficient @xmath116 has been found at the fq transition @xmath89 .
thermal expansion @xmath115 in ce@xmath0pd@xmath1ge@xmath2 .
inset shows the thermal expansion coefficient @xmath116 and thermal expansion @xmath115 at low temperatures . ]
thermal expansion @xmath115 along [ 001 ] in ce@xmath0pd@xmath1ge@xmath2 under fields up to 12 t along the [ 001 ] direction . ]
measurements of @xmath115 versus @xmath117 in various magnetic fields parallel to [ 001 ] are shown in fig .
[ eps13 ] .
the magnitude of the expansion @xmath115 in fields exhibits noticeable increase up to @xmath118 compared with that in zero magnetic field .
the sharp increase of @xmath115 at the transition point to the fq phase ii has been observed in low magnetic fields below 1 t. on the other hand , the gradual increase in the thermal expansion @xmath115 above 5 t up to 12 t indicates an obscure character of the i - ii phase boundary in high fields .
this is consistent with the fact that the elastic constants in fields of figs .
[ eps04 ] , [ eps05 ] , [ eps06 ] and [ eps07 ] show obscure transitions in fields .
this behavior is similar to the liquid - gas transition near the critical end point under hydrostatic pressures .
the thermal expansion along the [ 001 ] direction in fields parallel to [ 001 ] and the considerable softening of @xmath4 of 50% in ce@xmath0pd@xmath1ge@xmath2 strongly suggests that the order parameter of the fq ordering in phase ii is @xmath10 with @xmath26 symmetry .
the relatively small softening of 2.5% in @xmath5 in fig .
[ eps03 ] means that the quadrupole of @xmath29-type with @xmath13 is irrelevant for the transition at @xmath11 .
the thermal expansion of ce@xmath0pd@xmath1ge@xmath2 along [ 111 ] is required to examine an interplay of the spontaneous strain @xmath112 for the phase ii .
we refer to our recent study of the fq transition in hob@xmath2 and the phase iv in ce@xmath119la@xmath120b@xmath2 ( _ x_=0.75 , 0.70 ) , @xcite where the trigonal strain @xmath121 plays a significant role and the tetragonal strain @xmath122 is irrelevant .
these facts are well consistent with the pronounced elastic softening in @xmath5 of 70% in hob@xmath2 and of 31% in ce@xmath119la@xmath120b@xmath2 ( _ x_=0.75 , 0.70 ) .
the @xmath5 mode associated with the elastic strain @xmath124 of @xmath13 symmetry of ce@xmath0pd@xmath1ge@xmath2 in figs . [ eps03 ] and [ eps05 ] exhibits a shoulder like anomaly around 10 k in addition to the characteristic softening due to the quadrupolar coupling above @xmath7 k. it should be noted that this anomaly is absent for the @xmath4 mode associated with @xmath26 elastic strain @xmath125 and the bulk modulus @xmath65 with @xmath110 volume strain @xmath126 . in order to examine the origin of this anomaly ,
we have measured the frequency dependence of @xmath5 from 10 mhz up to 250 mhz .
the elastic constant @xmath5 of fig .
[ eps14](a ) exhibits shoulders showing remarkable frequency dependence .
an increase in ultrasonic attenuation around shoulder has also been found , but not discussed here .
we describe this frequency dependence of the elastic constant @xmath127 in terms of debye - type dispersion as @xmath128 where @xmath129 and @xmath130 are the elastic constants of high frequency limit and low frequency one , respectively . here
@xmath131 is an angular frequency of the ultrasonic wave and @xmath132 means the relaxation time of the system . in fittings of fig .
[ eps14](b ) , we take into account the superposition of two susceptibilities by the quadrupole one of eq.(3 ) and the debye - type dispersion of eq.(5 ) as @xmath133 .
the inflection points around 10 k in @xmath5 indicated by arrows in fig .
[ eps14](a ) mean the temperatures where the @xmath132 coincides with the @xmath131 as @xmath134 .
the ultrasonic attenuation is expected to be maximum at the temperatures for @xmath134 .
the solid lines of fig .
[ eps14](b ) being the calculations with eq .
( 5 ) well reproduce the experimental results of fig .
[ eps14](a ) .
the relaxation time obeying the arrhenius - type temperature dependence @xmath135exp@xmath15 with the attempt time @xmath16 sec and the activation energy @xmath17 k has been determined .
the parameter of @xmath136 j / m@xmath58 is used .
frequency dependence of the elastic constant of the transverse @xmath5 mode in ce@xmath0pd@xmath1ge@xmath2 .
( a ) represents the experimental results with the frequencies from 10 mhz up to 250 mhz .
( b ) is the calculation in terms of the debye - type dispersion of eq .
( 5 ) in text . ]
the ultrasonic dispersion due to electron thermal hopping has already found in the inhomogeneous valence fluctuation compounds of sm@xmath0x@xmath51 ( x = se , te ) , yb@xmath51(as@xmath137sb@xmath138)@xmath0 @xcite and sr@xmath52ca@xmath139cu@xmath140o@xmath141 @xcite .
it is remarkable that the very slow relaxation time @xmath132 and extremely low activation energy @xmath142 for ce@xmath0pd@xmath1ge@xmath2 are exceptional as compared to those of charge fluctuation compounds sm@xmath0se@xmath51 @xcite and sm@xmath0te@xmath51 @xcite , @xmath143 sec and @xmath144 k , and sr@xmath52ca@xmath139cu@xmath140o@xmath141 , @xmath145 sec and @xmath146 k.@xcite this discrepancy of the order of @xmath147 and @xmath142 between the present ce@xmath0pd@xmath1ge@xmath2 and the charge fluctuation compounds indicates that thermally activated rattling motion of heavy mass particle , which is probably rare - earth ion in a cage , gives rise to the ultrasonic dispersion in ce@xmath0pd@xmath1ge@xmath2 .
glass materials and charge fluctuation compounds exhibit frequently the ultrasonic dispersion , which results from a thermally activated motion in a double- or multi - well potential . the compounds such as sm@xmath0x@xmath51 ( x = s , se , te ) with different valence of sm@xmath148 and sm@xmath18 ions and sr@xmath52ca@xmath139cu@xmath140o@xmath141 with cu@xmath148 and cu@xmath18 ions cause the ultrasonic dispersions due to thermally assisted charge fluctuation in the temperature region between 100 - 200 k. the two - level system ( tls ) due to an atomic tunneling @xcite or electron tunneling @xcite manifests itself in glass materials at low temperatures , where the thermally activated motion dies out .
the tls yields the decrease in the elastic constant proportional to ln@xmath117,@xcite the specific heat to @xmath117 and thermal conductivity to @xmath149.@xcite besides in the case of sm@xmath0te@xmath51 , remarkable logarithmic decrease in the elastic constant appears below about 15 k down to the spin glass transition at @xmath150 k , which suggests the existence of the 4f - electron tunneling motion between sm@xmath148 and sm@xmath18 ions situated charge glass state.@xcite [ cols="^,^,^",options="header " , ] the present clathrate compound ce@xmath0pd@xmath1ge@xmath2 is a crystal possessing an ideal periodic arrangement of cages in space .
the stable trivalent ce ions in cages of ce@xmath0pd@xmath1ge@xmath2 are free from the valence fluctuation phenomena . as shown in fig .
[ eps01 ] , the clathrate compound ce@xmath0pd@xmath1ge@xmath2 is made up of the cage at 4a site consisting of pd and ge with distances @xmath151=3.332 , @xmath152=3.067 , and the cage at 8c site of pd with @xmath153=2.868 , @xmath154=3.373 . the trivalent ce - ion with radii @xmath155 inside the 4a - site cage in particular is expected to show the rattling motion over the off - center positions being away from the center of the cage
actually the neutron scattering on pr@xmath0pd@xmath1ge@xmath2 and nd@xmath0pd@xmath1ge@xmath2 revealed the sharp transition peaks indicating the stable cef splitting at 8c site and no indication for cef state at 4a site.@xcite these results may imply the obscure cef state due to the off - center ce ion at 4a site contrary to the well - defined cef splitting at 8c site being stable ce ion position .
notable finding of fig .
[ eps14 ] is that the ultrasonic dispersion in the @xmath5 mode associated with @xmath112-type strain indicates the rattling motion with specific @xmath13 symmetry in ce@xmath0pd@xmath1ge@xmath2 .
it is expected that the ce ion in 4a cage with cubic symmetry o@xmath40 favors off - center positions along one of the three principle directions of [ 100 ] , [ 110 ] and [ 111 ] .
as one can see the cage at 4a site in fig .
[ eps01 ] , it is of particular interest that no atom exists along the three - fold [ 111 ] directions , while the ge atom occupies along the four - fold [ 100 ] and the pd atom along the two - fold [ 110 ] ones .
this crystallographic character may promise a flat potential along the three - fold [ 111 ] directions and profound potentials along the four - fold [ 100 ] and two - fold [ 110 ] directions .
presumably the ce ion at 4a site prefers the off - center eight positions along the [ 111 ] directions , which are defined as @xmath157 , @xmath158 , @xmath159 , @xmath160 , @xmath161 , @xmath162 , @xmath163 , @xmath164 .
the atomic densities @xmath165 at the eight off - center positions are also defined .
when 48 symmetry operators of o@xmath40 point group are acted on the atomic densities @xmath166 , one can derive the transfer representation matrices with @xmath167 elements .
consequently , one obtains the characters @xmath168}$ ] for the rattling motion by tracing the diagonal elements of the representation matrices . using the characters @xmath168}$ ] and the character table for the irreducible representations of o@xmath40,@xcite the rattling motion over the eight off - center positions along the [ 111 ] direction
is reduced to the direct sum of @xmath110(1d)@xmath169@xmath33(1d)@xmath169@xmath24(3d)@xmath169@xmath13(3d ) . employing projection operators on appropriate atomic density @xmath166
, we obtain the rattling modes for the irreducible representations as listed in table 1 together with the elastic strains @xmath67 .
one can see the presence of @xmath13 rattling mode @xmath170 coupled to the strain @xmath171 , @xmath172 , @xmath112 contrary to the absence of @xmath26 rattling mode coupled to the strain @xmath122 , @xmath125 .
this is consistent with the fact that the ultrasonic dispersion reveals in @xmath5 and is absent in @xmath173/2 .
the ce atom in the cage of ce@xmath0pd@xmath1ge@xmath2 obeys a harmonic oscillation of @xmath174exp@xmath175 with a mean square displacement @xmath176 .
the attempt time @xmath16 sec and the mass @xmath177 where @xmath178 is a proton mass , leads to the mean square displacement @xmath179 being approximately twice of off - center distances @xmath180 as @xmath181 .@xcite the full symmetry @xmath110 rattling mode @xmath182 means the uniform atomic distribution with fraction 1/8 at each eight off - center positions . while the @xmath13 rattling mode , for instance @xmath183 , represents anisotropic atomic distribution being quadrupole @xmath29 at the lowest order such as fraction 1/4 at @xmath184 , @xmath185 , @xmath186 , @xmath187 and zero at @xmath188 , @xmath189 , @xmath190 , @xmath191 as shown in fig .
the present group theoretical analysis for the rattling mode is essentially the same treatment previously argued for the charge fluctuation mode.@xcite in the present system ce@xmath0pd@xmath1ge@xmath2 , the thermally activated @xmath13 rattling mode may be a ground state and the @xmath110 , @xmath33 , @xmath24 be excited states .
our group has recently found similar ultrasonic dispersion around 30 k in a heavy fermion superconductor pros@xmath51sb@xmath52 with a filled skutterudite structure .
it should be noted that ultrasonic dispersion in the @xmath4 mode of pros@xmath51sb@xmath52 indicates the @xmath26 rattling motion of pr atom over six fractionally occupied positions along [ 100 ] .
the dispersion of @xmath4 in pros@xmath51sb@xmath52 is contrary to the one of @xmath5 in the present compound of ce@xmath0pd@xmath1ge@xmath2.@xcite the attempted time @xmath192 sec , activation energy @xmath193 k and mean square displacement @xmath194 in pros@xmath51sb@xmath52 are comparable to the present results of ce@xmath0pd@xmath1ge@xmath2 . the thermally activated @xmath13 rattling motion with fractional atomic state in ce@xmath0pd@xmath1ge@xmath2 dies out with decreasing temperature . at further low temperatures ,
the off - center tunneling state of ce ions in the 4a - site cages will appear , which means a quantum state being occupied four positions , for instance at @xmath184 , @xmath185 , @xmath186 , @xmath187 for the case of @xmath195 , at the same time .
the charge glass state in the inhomogeneous mixed valence compound sm@xmath0te@xmath51 revealed the low - temperature softening in elastic constants proportional to ln@xmath117 being resemble the structural glass.@xcite the present scenario of the atomic tunneling of ce ions in 4a - site cages in ce@xmath0pd@xmath1ge@xmath2 is also expected to show the elastic softening at low temperatures .
however , the low - temperature quadrupole and magnetic orderings in ce@xmath0pd@xmath1ge@xmath2 may mask the character of the tunneling state in the present case . in order to clarify the tunneling and rattling in the present clathrate compounds , the low - temperature thermodynamic and ultrasonic measurements on la@xmath0pd@xmath1ge@xmath2 without 4f - electron in particular
is required .
schematic view for the @xmath13 rattling mode @xmath195 due to the off - center ce - ion along the three - fold [ 111 ] direction in the 4a - site cage .
the @xmath195 represents that fractional atomic density 1/4 is located at @xmath184 , @xmath185 , @xmath186 , @xmath187 and null at @xmath188 , @xmath189 , @xmath190 , @xmath191 .
the freezing of the thermally activated motion of the @xmath13 rattling mode brings about the atomic tunneling state at low temperatures . ]
in the present paper we have measured the elastic constants and thermal expansion of ce@xmath0pd@xmath1ge@xmath2 . the characteristic elastic softening in @xmath4 and @xmath5
is well described in terms of the quadrupole susceptibility for the @xmath8 ground state .
the important finding is that the @xmath4 shows the huge softening of 50% towards @xmath7 k and the @xmath5 exhibits the softening of 2.5% only .
this result strongly indicates the fq ordering with the order parameter of the @xmath26 symmetry in phase ii below @xmath11 .
actually we have successfully observed the sharp increase of @xmath9 in length along the [ 001 ] direction below @xmath11 .
this is the evidence for the @xmath10-type fq ordering accompanied by the structural change from cubic lattice to tetragonal one at @xmath11 in ce@xmath0pd@xmath1ge@xmath2 . for the investigation of the magnetic phase diagram concerning the fq phase ii and afm phase iii in ce@xmath0pd@xmath1ge@xmath2 ,
the elastic constants and thermal expansion in fields have been measured .
we have found that the boundary from the paramagnetic phase i to the fq phase ii of @xmath10 shifts to higher temperatures with increasing magnetic fields .
the result that the i - ii phase transition becomes obscure in fields is similar to the liquid - gas transition approaching to the critical end point under pressure .
this result consistent with the fact that the fq order parameter @xmath10 in ce@xmath0pd@xmath1ge@xmath2 has the total symmetry under fields along the [ 001 ] direction.@xcite the boundary from the phase ii to the afm ordering shifts to lower temperatures as similar as the conventional afm ordering .
we have found the ultrasonic dispersion in the @xmath5 mode indicating the rattling motion with @xmath13 symmetry .
taking into account the absence of the atom along the [ 111 ] direction in 4a - site cage , we have successfully picked up the specific @xmath13 rattling mode @xmath196 , @xmath197 , @xmath195 with the fractional atomic density over the eight minimum positions of potential along the four - fold [ 111 ] , @xmath198 $ ] , @xmath199 $ ] and @xmath200 $ ] directions .
the dispersion of the @xmath5 mode obeying the debye - formula revealed the thermally activated - type relaxation time @xmath135exp@xmath15 for the @xmath13 rattling mode with an attempt time @xmath16 sec and an activation energy @xmath17 k. the estimated mean square displacement @xmath201 for the harmonic oscillation of ce atom leads to the distance of the potential minima along the [ 111 ] direction as @xmath202 . in order to confirm the anisotropic atomic distribution in the 4a - site cage , the neutron or x - ray scattering
is required .
the freezing of the thermally activated motion due to the @xmath13 rattling mode with lowering temperature brings about the atomic tunneling state . by analogy of the charge glass compound
sm@xmath0te@xmath51 , the ce - ion tunneling is expected at low temperatures .
the ultrasonic investigation on la@xmath0pd@xmath1ge@xmath2 free from the long - range ordering due to 4f - electrons is now in progress to shed light on the tunneling and rattling in cages .
999 d. schmitt , p. morin , and j. pierre , j. magn .
magn . matter .
@xmath203 , 249 ( 1978 ) .
r. takke , n. dolezal , w. assmus , and b. lthi , j. magn .
magn . matter .
@xmath204 , 247 ( 1981 )
. h. nakao , k. magishi , y. wakabayashi , y. murakami , k. koyama , k. hirota , y. endoh , and s. kunii , j. phys .
@xmath205 , 1857 ( 2001 ) .
s. nakamura , t. goto , s. kunii , k. iwashita , and a. tamaki , j. phys .
@xmath206 , 623 ( 1994 ) .
r. shiina , h. shiba , and p. thalmeier , j. phys .
@xmath207 , 1741 ( 1997 ) ; r. shiina , o. sakai , h. shiba , and p. thalmeier , j. phys .
@xmath208 , 941 ( 1998 ) .
j. kitagawa , n. takeda , and m. ishikawa , phys .
b @xmath209 , 5101 ( 1996 ) .
a. v. grivanov , yu . d. seropegin , and o. i. bodak , j. alloys .
@xmath210 , l9 ( 1994 ) .
l. keller , a. dnni , m. zolliker , and t. komatsubara , physica b @xmath211-@xmath212 , 336 ( 1999 ) .
j. kitagawa , n. takeda , m. ishikawa , t. yoshida , a. ishiguro , n. kimura , and t. komatsubara , phys .
b @xmath213 , 7450 ( 1998 ) . o. suzuki , t. horino , y. nemoto , t. goto , a. dnni , t. komatsubara , and m. ishikawa , physica b @xmath211-@xmath212 , 334 ( 1999 ) ; t. goto , t. horino , y. nemoto , t. yamaguchi , a. dnni , o. suzuki , and t. komatsubara , physica b @xmath214-@xmath215 , 492 ( 2002 ) .
a. dnni , t. herrmannsdrfer , p. fischer , l. keller , f. fauth , k a mcewen , t. goto , and t. komatsubara , j. phys . :
matter @xmath216 , 9441 ( 2000 ) .
l. mihaly , nature @xmath217 , 839 ( 1998 ) ; v. keppens , d. mandrus , b.c .
sales , b.c .
chakoumakos , p. dai , r. coldea , m.b .
maple , d.a .
gajewski , e.j .
freeman , and s. bennington , nature @xmath217 , 876 ( 1998 ) . v. keppens , b.c .
sales , d. mandrus , b.c .
chakoumakos , and c. laermans , phil . mag .
lett . , @xmath218 , 807 ( 2000 ) . t. goto , y. nemoto , k. sakai , t. yamaguchi , m. akatsu , t. yanagisawa , h. hazama , k. onuki , h. sugawara , and h. sato , to be submitted phys . rev .
e. kanda , t. goto , h. yamada , s. suto , s. tanaka , t. fujita , and t. fujimura , j. phys .
@xmath219 , 175 ( 1985 ) . ,
h. yamada , s. tanaka , y. kayanuma , t. kojima , j. phys .
@xmath219 , 1180 ( 1985 ) .
p. thalmeier and b. lthi , handbook on the physics and chemistry of rare earths , ed .
k. a. gschneider jr . and l. eyring ( north - holland , 1991 ) vol .
t. goto , y. nemoto , y. nakano , s. nakamura , t. kajitani , and s. kunii , physica b @xmath220-@xmath221 , 586 ( 2000 ) ; t. yamaguchi , m. akatsu , y. nakano , t. washizawa , y. nemoto , t. goto , a. dnni , s. nakamura , and s. kunii , physica b , @xmath222-@xmath223 , 622 ( 2003 ) .
a. dnni , l. keller , p. fischer , y. aoki , h. sato , f. fauth , m. zolliker , t. komatsubara , and y. endoh , j. phys . : condens .
matter @xmath224 , 7219 ( 1998 ) ; n. kimura , n. tateiwa , m. nakayama , h. aoki , t. komatsubara , t. sakon , m. motokawa , y. koike , and n. metoki , physica b @xmath211-@xmath212 , 338 ( 1999 ) .
t. herrmannsdrfer , a. dnni , p. fischer , l. keller , e. clementyev , a. furrer , s. mango , b. van den brandt , and h. kitazawa , j. alloys . comp .
@xmath225-@xmath226 , 509 ( 2001 ) .
t. herrmannsdrfer , a. dnni , p. fischer , l. keller , g. bttger , m. gutmann , h. kitazawa , and j. tang , j. phys . : condens .
matter @xmath227 , 2929 ( 1999 ) .
t. herrmannsdrfer , a. dnni , p. fischer , l. keller , and h. kitazawa , physica b @xmath220-@xmath221 , 167 ( 2000 ) .
o. suzuki , t. goto , s. nakamura , t. matsumura , and s. kunii , j. phys .
@xmath208 , 4243 ( 1998 ) ; m. akatsu , t. goto , y. nemoto , o. suzuki , s. nakamura , and s. kunii , j. phys .
@xmath228 , 205 ( 2003 ) .
y. nemoto , h. aoki , t. goto , a. ochiai , and t. suzuki , physica b @xmath211-@xmath212 , 275 ( 1999 ) .
t. goto and b. lthi , advances in phys .
@xmath229 , 67 ( 2003 ) .
a. tamaki , t , goto , s. kunii , t. suzuki , t. fujimura , and t. kasuya , j. phys .
c : solid state phys .
@xmath230 , 5849 ( 1985 ) .
y. nemoto , t. goto , a. ochiai , and t. suzuki , phys . rev .
b @xmath231 , 12050 ( 2000 ) .
p. w. anderson , b. i. halperin , and c. m. varma , philos .
@xmath232 , 1 ( 1972 ) .
see for example ; s. hunklinger and w. arnold , physical acoustics ( academic , new york , 1976 ) , vol .
xii , p. 155 .
see for example ; amorphous solids : low - temperature properties , edited by w. a. phillips ( springer , berlin , 1981 ) . t. inui , y. tanabe , and y. onodera , group theory and its applications in physics ( springer , berlin , 1990 ) .
cox and a. zawadowski , advances in phys . , @xmath233 , 599 ( 1998 ) .
t. goto , y. nemoto , a. ochiai , and t. suzuki , phys .
b @xmath234 , 269 ( 1999 ) . | lattice effects in a cerium based clathrate compound ce@xmath0pd@xmath1ge@xmath2 with a cubic cr@xmath3c@xmath2-type structure have been investigated by ultrasonic and thermal expansion measurements .
elastic softenings of @xmath4 and @xmath5 proportional to the reciprocal temperature @xmath6 above @xmath7 k are well described in terms of the quadrupole susceptibility for the ground state @xmath8 quartet .
a huge softening of 50 % in @xmath4 and a spontaneous expansion @xmath9 along the [ 001 ] direction in particular indicate the ferroquadrupole ordering of @xmath10 below @xmath11 .
the elastic anomalies associated with the antiferromagnetic ordering at @xmath12 k and the incommensurate antiferromagnetic ordering are also found .
notable frequency dependence of @xmath5 around 10 k is accounted for by the debye - type dispersion indicating a @xmath13 rattling motion of an off - center ce ion along the [ 111 ] direction with eight fractionally occupied positions around the 4a site in a cage .
the thermally activated @xmath13 rattling motion obeying a relaxation time @xmath14exp@xmath15 with an attempt time @xmath16 sec and an activation energy @xmath17 k dies out with decreasing temperature , and then the off - center tunneling state of ce ion in the 4a - site cage will appear at low temperatures . |
a promising way to explain the late - time accelerated expansion of the universe is to assume that at large scales general relativity ( gr ) breaks down , and a more general action describes the gravitational field .
thus , in the latter context , infra - red modifications to gr have been extensively explored , where the consistency of various candidate models have been analysed ( see @xcite for a review ) .
note that the einstein field equation of gr was first derived from an action principle by hilbert , by adopting a linear function of the scalar curvature , @xmath0 , in the gravitational lagrangian density .
the physical motivations for these modifications of gravity were related to the possibility of a more realistic representation of the gravitational fields near curvature singularities and to create some first order approximation for the quantum theory of gravitational fields , and more recently in an attempt to explain the late - time cosmic acceleration . in this context , a more general modification of the hilbert - einstein gravitational lagrangian density involving an arbitrary function of the scalar invariant , @xmath1 , has been extensively explored in the literature , and recently a maximal extension of the hilbert - einstein action has been proposed @xcite .
the action of the maximal extension of the hilbert - einstein action is given by @xcite @xmath3 where @xmath4 is an arbitrary function of the ricci scalar @xmath0 , and of the lagrangian density corresponding to matter , @xmath5 .
the energy - momentum tensor of matter is defined as @xmath6 .
varying the action with respect to the metric @xmath7 , the gravitational field equation of @xmath8 gravity is provided by @xmath9 g_{\mu \nu } = \frac{1}{2 } f_{l_{m}}\left ( r , l_{m}\right ) t_{\mu \nu } \,.\end{aligned}\ ] ] for the hilbert - einstein lagrangian , @xmath10 , we recover the einstein field equations of gr , i.e. , @xmath11 . for @xmath12 , where @xmath13 , @xmath14 and @xmath15 are arbitrary functions of the ricci scalar and of the matter lagrangian density , respectively , we obtain the field equations of modified gravity with an arbitrary curvature - matter coupling @xcite .
an interesting application was explored in the context of @xmath16 gravity@xcite .
the @xmath2 models possess extremely interesting properties .
first , the covariant divergence of the energy - momentum tensor is non - zero , and is given by @xmath17 \frac{\partial l_{m}}{% \partial g^{\mu \nu } } \ , .
\label{noncons}\end{aligned}\ ] ] the requirement of the conservation of the energy - momentum tensor of matter , @xmath18 , provides the condition given by @xmath19 \partial l_{m}/ \partial g^{\mu \nu } = 0 $ ] .
secondly , the motion of test particles is non - geodesic , and takes place in the presence of an extra force . as a specific example , consider the case in which matter , assumed to be a perfect thermodynamic fluid , obeys a barotropic equation of state , with the thermodynamic pressure @xmath20 being a function of the rest mass density of the matter @xmath21 only , i.e. , @xmath22 , and consequently , the matter lagrangian density , becomes an arbitrary function of the energy density @xmath21 only , i.e. , @xmath23 ( for more details , we refer the reader to @xcite ) .
thus , the equation of motion of a test fluid is given by @xmath24 , where the extra - force @xmath25 is defined by @xmath26 \left ( u^{\mu } u^{\nu } -g^{\mu \nu } \right ) \,.\ ] ] note that @xmath25 is perpendicular to the four - velocity , @xmath27 , i.e. , @xmath28 .
the non - geodesic motion , due to the non - minimal couplings present in the model , implies the violation of the equivalence principle , which is highly constrained by solar system experimental tests .
however , it has recently been argued , from data of the abell cluster a586 , that the interaction between dark matter and dark energy implies the violation of the equivalence principle @xcite .
thus , it is possible to test these models with non - minimal couplings in the context of the violation of the equivalence principle .
it is also important to emphasize that the violation of the equivalence principle is also found as a low - energy feature of some compactified versions of higher - dimensional theories . in the newtonian limit of weak gravitational fields @xcite , the equation of motion of a test fluid in @xmath4 gravity
is given by @xmath29 where @xmath30 is the total acceleration of the system ; @xmath31 is the newtonian gravitational acceleration ; the term @xmath32 $ ] is identified with the hydrodynamic acceleration term in the perfect fluid euler equation .
now , by assuming that in the newtonian limit the function @xmath33 can be represented as @xmath34 , where @xmath35 , so that @xmath36 given by @xmath37\,,\ ] ] is a supplementary acceleration induced due to the modification of the action of the gravitational field .
in conclusion , the maximal extensions of gr , namely the @xmath2 gravity models open the possibility of going beyond the algebraic structure of the hilbert - einstein action . on the other hand ,
the field equations of @xmath2 gravity are equivalent to the field equations of the @xmath1 model in empty space - time , but differ from them , as well as from gr , in the presence of matter .
thus , the predictions of @xmath2 gravitational models could lead to some major differences , as compared to the predictions of standard gr , or other generalized gravity models , in several problems of current interest , such as cosmology , gravitational collapse or the generation of gravitational waves .
the study of these phenomena may also provide some specific signatures and effects , which could distinguish and discriminate between the various gravitational models .
in addition to this , in order to explore in more detail the connections between the @xmath2 theory and the cosmological evolution , it is necessary to build some explicit physical models .
fsnl acknowledges financial support of the fundao para a cincia e tecnologia through the grants cern / fp/123615/2011 and cern / fp/123618/2011 .
f. s. n. lobo , arxiv:0807.1640 [ gr - qc ] .
t. harko and f. s. n. lobo , eur .
j. c * 70 * , 373 ( 2010 ) .
t. harko , phys .
b * 669 * , 376 ( 2008 ) .
o. bertolami , c. g. boehmer , t. harko and f. s. n. lobo , phys .
d * 75 * , 104016 ( 2007 ) .
t. harko , t. s. koivisto and f. s. n. lobo , mod .
a * 26 * ( 2011 ) 1467 .
t. harko , f. s. n. lobo , s. i .
nojiri and s. d. odintsov , phys . rev .
d * 84 * , 024020 ( 2011 ) .
o. bertolami , f. s. n. lobo and j. paramos , phys .
d * 78 * , 064036 ( 2008 ) .
o. bertolami , j. paramos , t. harko and f. s. n. lobo , arxiv:0811.2876 [ gr - qc ] .
o. bertolami , f. gil pedro and m. le delliou , phys .
b * 654 * , 165 ( 2007 ) . | we consider a maximal extension of the hilbert - einstein action and analyze several interesting features of the theory .
more specifically , the motion is non - geodesic and takes place in the presence of an extra force .
these models could lead to some major differences , as compared to the predictions of general relativity or other modified theories of gravity , in several problems of current interest , such as cosmology , gravitational collapse or the generation of gravitational waves .
thus , the study of these phenomena may also provide some specific signatures and effects , which could distinguish and discriminate between the various gravitational models . |
the studies of nonlinear excitations of two - dimensional ( 2d ) and quasi 2d correlated spin systems are an important issue of modern physics of magnetism , and can be useful for development of general soliton concepts .
@xcite the topological textures like localized solitons ( skyrmions @xcite ) or magnetic vortices make an important contribution to the thermodynamics of magnetically ordered systems @xcite or even determine the character of its ordering as in the case of berezinskii - kosterlitz - thouless transition .
@xcite last years the interest for two - dimensional solitons has grown since they are frequently realized as ground state in the finite - size mesoscopic magnetic samples , so - called magnetic dots . @xcite the most `` famous '' topological solitons are magnetic vortices having @xmath9 topological charge .
these vortices are usually related to thermodynamic aspects of soliton physics .
@xcite also , they appear in mesoscopic nanostructures . @xcite easy - plane magnets with continuously degenerated ground state have vortices with the energy being logarithmically divergent as a function of system size .
the other and much less studied example of topological solitons are magnetic skyrmions which are present in isotropic or easy - axis magnets .
contrary to the above vortices , the latter textures are characterized by nontrivial @xmath0 topological charge and finite energy .
it is known that they determine the response functions of 2d magnets at finite temperatures @xcite and take part in long - range order breakdown in isotropic magnets .
@xcite the skyrmions form ground state of magnetic nanoparticles with easy - axis anisotropy .
@xcite their analysis is more complicated as compared to magnetic vortices and comprises many nontrivial features .
an important example of latter features is the problem of a skyrmion stability since due to hobart - derrick theorem the static solitons with finite energy are unstable for wide class of models including standard continuous magnetic models .
@xcite for magnetic vortices , the consideration of lowest possible topological charge @xmath10 is sufficiently as the vortex energy grows with @xmath3 , @xmath11 .
because of that it is advantageous for a vortex with @xmath12 to decay for @xmath13 vortices with @xmath10 and the vortices with @xmath7 can be stable in exceptional cases only .
@xcite the situation for skyrmions is not that simple .
the simplest continuous model for isotropic 2d ferromagnet ( fm ) @xmath14 ^ 2d^2x\ ] ] admits the well - known belvin - polakov ( bp ) solution , @xcite which reads @xmath15 where @xmath16 is normalized magnetization @xmath17 @xmath18 is a spin value , @xmath19 is a lattice constant of a 2d fm , @xmath20 is its exchange constant , @xmath21 and @xmath22 are polar coordinates in the @xmath23plane , @xmath24 is an arbitrary constant .
solution has the energy @xmath25 so that the state of bp skyrmions with @xmath12 merges or dissociates into several other similar skyrmions with different @xmath26 s ( the only rule that in such process the topological charge should conserve ) without their energy altering .
such exact degeneration is related to very high hidden symmetry , stemming from exact integrability of corresponding static model ( see , e.g. , @xcite ) .
this degeneration should certainly be lifted if we go beyond the model .
the most important characteristic here is the parameter @xmath27 , which is appropriate to call the _ topological energy density _ , @xmath28 where @xmath29 is the energy of a soliton with topological charge @xmath13 .
if @xmath27 is a growing function of @xmath13 , the most favorable state with a given @xmath13 comprises @xmath3 solitons with unit topological charge otherwise such state is unstable .
latter question is especially important for the investigation of general regularities of the highly excited magnet states evolution , ( see , e.g. ref . and references therein ) or for the analysis of essentially inhomogeneous magnet states under strong pumping .
@xcite latter states can be generated by the ultrafast pulses of magnetic field , see refs .
for details .
the preceding discussion demonstrates that the problem of obtaining and investigation of the stable skyrmions with higher topological charges is extremely important .
@xcite the present work is devoted to the analysis of skyrmions with higher @xmath0 topological charges in 2d heisenberg ferromagnet with uniaxial anisotropy .
we show that there exists a certain range of system parameters ( exchange and anisotropy constants ) , where stable precessional solitons with topological charge @xmath30 exist .
it turns out that in wide range of anisotropy constants , the topological energy density @xmath31 of the textures with @xmath30 is lower then that of the textures with @xmath32 . on the other hand ,
the solitons with @xmath5 and @xmath33 have monotonously growing phase boundary functions @xmath34 , while the case @xmath7 has peculiar nonmonotonous behavior , determining the transition regime from low to high topological charges .
this means that the preferable values of soliton topological charge are neither @xmath5 nor high charges , but rather @xmath7 or @xmath8 .
we begin with the discrete model of a classical 2d fm with uniaxial anisotropy , described by the following hamiltonian @xmath35.\end{aligned}\ ] ] here @xmath36 is a classical spin vector with fixed length @xmath18 on the site @xmath37 of a 2d square lattice .
the summations run over all lattice sites @xmath37 and nearest
neighbors @xmath38 , @xmath39 is the exchange integral and the constant @xmath40 describes the anisotropy of spin interaction . in subsequent discussion , we refer to this type of anisotropy as exchange anisotropy ( ea ) . additionally , we took into account single - ion anisotropy ( sia ) with constant @xmath41 .
we consider @xmath42axis to be easy magnetization direction so that @xmath43 or @xmath44 .
the analysis of real magnetic systems with discreet spins can be performed only numerically . in principle , it can be done by the same method as was described in ref . , but it is not easy to extract necessary information from the set of numerical data . on the other hand ,
if we neglect the specific effects of discreteness ( which appear at strong anisotropy only , @xmath45 ) , like the presence of pure collinear structures , see ref . , the consideration can be simplified using the generalized continuous approximation and classical landau - lifshitz equation . in this approximation
we can introduce the smooth function @xmath46 instead of discreet variable @xmath47 . in this case , the classical magnetic energy functional @xmath48 $ ] can be constructed expanding the discrete hamiltonian in power series of magnetization gradients , yielding @xmath49 where @xmath50 contains zeroth and second order contributions to magnetic energy , see , and @xmath51 contains the fourth powers of gradients .
the explicit ( and quite cumbersome ) expressions for @xmath50 and @xmath51 had been presented in ref . .
we note here , that single - ion anisotropy enters only @xmath50 , but not @xmath51 and higher terms , while exchange anisotropy enters every term of the expansion . for isotropic case @xmath52 and @xmath53
, @xmath50 coincides with the energy of the isotropic continuous model bearing bp soliton solutions of the form with degenerate ( with respect to topological charge ) topological energy density , see also .
a simple accounting of magnetic anisotropy in @xmath50 generates a model , which is typical example of the models governed by hobbard - derrick theorem - it does not admit static stable soliton solutions . in the model with @xmath54 only , the size of any texture like domain wall , soliton etc ( see , e.g. ref . for details ) is given by the characteristic length @xmath55 @xmath56 in the case of weak anisotropy , @xmath57 , the length scale @xmath58 so that the magnetization varies slowly in a space .
now we consider generalized model , including higher powers of gradients . in the expansion ,
we limit ourselves to the terms of fourth order only as they are playing a decisive role in solitons stabilization , see refs . for details .
the simplest possible generalized model with account for @xmath51 only can in principle admit the above static stable solitons .
@xcite simple scaling arguments can demonstrate that .
namely , if the soliton core has size @xmath59 , the specific calculations with energy functionals like ( see below ) yield the dependence of the energy @xmath60 on @xmath59 in the form @xmath61 where the first term comes from @xmath51 and the rest come from @xmath50 .
if @xmath62 had a minimum ( this occurs if @xmath63 ) , this would mean the existence of a stable static skyrmion .
unfortunately , for the real model of magnet @xmath64 so that the first term is negative and the dependence @xmath65 does not have a minimum .
we note also that the expansion and the above scale arguments are valid for any symmetry of initial 2d discrete lattice .
the only difference is in the coefficients before gradient powers .
although these coefficients influence the solitons properties , the main feature of these expansions , consisting in the fact that in eq .
@xmath66 remains the same .
in other words , we did not find any symmetry of 2d fm with nearest - neighbor ferromagnetic interaction , where @xmath67 has a minimum so that stable static soliton can exist .
@xcite in the absence of _ static _ two dimensional solitons , it is possible to construct stable soliton states with _ stationary dynamics _ , due to the presence of additional integrals of motion for magnetization fields .
the purely uniaxial model possesses the exact symmetry with respect to the spin rotation around @xmath68axis so that the energy functional @xmath69 $ ] does not depend explicitly on the variable @xmath70 .
this leads to the appearance of an additional integral of motion , @xmath42-projection of total spin .
this integral of motion can be conveniently parameterized via integer @xmath71 defining a number of bound magnons in a soliton @xmath71 , see ref . for details .
in continuous approximation it can be written as @xmath72 conservation of @xmath71 leads to the presence of so called _ precessional _ solitons characterizing by time - independent projection of magnetization onto the easy @xmath68axis and with the precession of the magnetization vector @xmath73 at constant frequency @xmath74 around the @xmath42 axis , @xmath75 which holds instead of ( [ qe2 ] ) in this case .
the analogs of such precessional solitons are known to occur in different field - theoretical models , it is enough to note the non - topological coleman s @xmath76-balls , @xcite which do not have topological properties as well as @xmath0 topological @xmath76 - lumps , @xcite see ref . for details . stable dynamical solutions with nonzero @xmath74 correspond to conditional ( for fixed @xmath71 value ) minimum of the energy functional @xmath77 .
namely , we may look for an extremum of the expression @xmath78 where @xmath79 is an internal soliton precession frequency , which in this case can be regarded as lagrange multiplier .
note , that functional is nothing but the lagrangian of 2d fm magnetization field calculated with respect to specific time dependence .
this condition leads to the relation @xcite @xmath80 , which determines the microscopic origin of the precessional frequency @xmath74 .
namely , an addition of one extra spin deviation ( bound magnon ) to a soliton changes its energy by @xmath81 .
thus , the dependencies @xmath82 and @xmath83 are very important for the problem of a soliton stability .
further analysis of above continuous model consists in the solution of differential equations for soliton structure , what is equivalent to the minimization of the functional .
as these equations can barely be solved analytically , here we analyze the solitons properties by direct variational method . as we have shown earlier by comparison of variational approach and direct numerical minimization of initial discreet energy on a lattice ,
@xcite the variational approach gives fairly good results for weak anisotropies @xmath84 , where the generalized continuous description is valid .
the continuous models like and are usually parameterized by angular variables so that the energy @xmath29 becomes a functional of these variables , @xmath85 $ ] .
having energy functional @xmath29 , we can write the corresponding landau - lifshitz equations and lagrangian . to apply direct variational method for minimization of the energy @xmath86 we use the trial function @xmath87 where @xmath88 is the mcdonald function with index @xmath3 .
@xcite note , that trial function ( [ macd ] ) is based on the interpolative solution , constructed in ref . .
the trial function gives correct asymptotics both for @xmath89 ( corresponding to bp soliton ) and for @xmath90 ( exponential decay with some characteristic scale @xmath91 ) , see refs . for details .
latter exponential asymptotics is absent for solitons in isotropic 2d fm , e.g. for bp soliton .
it can be shown that the exponential asymptotics occurs for anisotropic models , where the length of decay is proportional to @xmath55 .
it had been demonstrated that due to power - law asymptotics of @xmath92 for isotropic magnet , the integrals defining soliton energy and number of bound magnons ( see below eqs . ) are divergent for @xmath5 .
@xcite to avoid this divergency in anisotropic models like , the interpolative ( between bp asymptotics at @xmath89 and exponential one at @xmath93 ) solution had been put forward in ref . .
our analysis of continuous model shows that at @xmath5 the above divergence can be cut off by the exponential asymptotics only .
for @xmath7 the power - law asymptotics is sufficient for convergence of corresponding integrals , while their derivatives with respect to parameter @xmath94 are divergent . in this case , the divergences in derivatives are also eliminated by exponential asymptotics . at @xmath95
all integrals and their derivatives ( with respect to @xmath94 ) are convergent .
this means that the behavior of solitons with @xmath5 and 2 on one side and those with @xmath95 on the other side is different , being determined by the interplay between effects of anisotropy and higher spatial derivatives . in our minimization method ,
the parameter @xmath96 is variational , while the parameter @xmath59 is kept constant as it is related to @xmath71 , @xmath97 , see , e.g. @xcite . in other words , we minimize the energy @xmath98 with the trial function ( [ macd ] ) over @xmath96 for constant @xmath59 .
this approach has the advantage that it also permits to investigate the stability of obtained soliton texture .
namely , a soliton is stable if it corresponds to the conditional minimum of the energy at fixed @xmath71 , and it is unstable otherwise . to proceed further ,
we introduce following dimensionless variables @xmath99 and express trial function in terms of them .
we have @xmath100 then using the equation , we can calculate the number of bound magnons in the soliton @xmath101 in the variables the soliton energy assumes the form @xmath102 where @xmath103 , \nonumber \\ & & \gamma _ 4(z)=\int_0^{\infty } xdx\biggl\{\left ( \delta _ x\theta \right)^2\left ( 1+\kappa \sin ^2\theta \right ) + \theta^{\prime 4 } \left ( 1+\kappa \cos ^2\theta \right ) + \frac{\nu^2\sin ^2\theta } { x^2}% \left ( \frac{\nu^2}{x^{2}}+2\theta ^{\prime 2}\right ) + \nonumber \\ & & + \delta _ { x}\theta \sin 2\theta \left ( \kappa \theta ^{\prime 2}-\frac{% \nu ^2}{x^2}\right ) \biggr\},\ \ \theta ' = \frac{d\theta}{dx } , \ \delta _ x\theta = \frac{d ^2 \theta}{dx^2 } + \frac 1x\frac{d \theta}{dx}\;. \label{asm7}\end{aligned}\ ] ] thus , we express the energy and the number of magnons via two parameters , @xmath104 and @xmath42 .
it turns out , that initial dimensional variables @xmath96 and @xmath59 enter the problem only in the form of their product @xmath42 .
the dependence of @xmath71 and @xmath29 on @xmath42 enters the problem via a few complicated functions @xmath105 and @xmath106 , which can be written only implicitly in the form of integrals .
however , in terms of these functions , the dependence on @xmath107 ( [ dimen ] ) turns out to be quite simple .
this permits reformulation of the initial variational problem in terms of variables @xmath42 and @xmath71 only .
namely , we express @xmath108 and substitute this expression into the dimensionless energy ( [ dimen ] ) .
this gives us the expression for the energy of a soliton with given @xmath71 , as a function of variational parameter @xmath42 .
then we can find a minimum of @xmath29 with respect to @xmath42 , keeping @xmath71 constant .
the result of such numerical minimization in the form of the dependence @xmath109 is shown in fig.[fig : v1 ] for a magnet with purely exchange anisotropy @xmath110 .
the curves for higher @xmath40 are qualitatively the same . to justify the applicability of our direct variational approach ,
earlier we have shown,@xcite that for @xmath5 the dependencies @xmath111 found by variational and numerical minimizations of the energy , are identical at small enough anisotropy , @xmath112 .
our analysis shows ( see also fig.[fig : v1 ] ) that topological energy density grows slowly as function of @xmath71 .
we note here that this property holds for all anisotropy constants and topological charges .
even more interesting is the fact that all curves @xmath109 have threshold points @xmath113 so that solitons exist only at @xmath114 and @xmath115 .
these threshold values determine the minimal soliton energy , which is most important characteristic of soliton contribution into magnet thermodynamics and can be observed experimentally .
@xcite for instance ( see fig.[fig : v1 ] ) , at @xmath10 and @xmath116 @xmath117 , which is substantially higher then that expected from eq .
@xmath118 .
[ fig : v1 ] demonstrates one more unexpected soliton property , namely that at @xmath116 the energy density @xmath111 _ decreases _ with increase of @xmath13 and fixed @xmath71 value . as we will show below
, such behavior occurs for many values of @xmath119 , although there can be exceptions .
the most important feature , however , is the existence of above threshold energy and bound magnons number values .
the behavior of these values at variation of anisotropy constants is quite nontrivial .
for example , at @xmath116 ( the value chosen for fig .
[ fig : v1 ] ) @xmath120 but @xmath121 . at the same time , the corresponding threshold energies behave monotonically @xmath122 .
our extensive analysis of numerical curves @xmath109 for different anisotropies have shown that their overall behavior is dictated primarily by the threshold values : if @xmath123 , then the entire curve @xmath109 lies above corresponding curve @xmath124 in wide interval of @xmath71 s .
this shows the importance of the above threshold values for the properties of solitons .
the information about these values can be conveniently represented in the form of so - called solitons phase diagram , i.e. the dependence of @xmath125 and @xmath126 on @xmath127 .
to obtain the phase diagram , we should pay attention to the details of above minimization procedure .
namely , to obtain truly stable soliton texture we should demand that the conditional extremum of the energy @xmath60 is a minimum .
so , we keep track not only to the first derivative @xmath128 to be zero but also to the second derivative to be positive ( corresponding to a minimum ) at the point @xmath129 , where @xmath130 . when a soliton approaches the limit of its stability , the modulus of second derivative diminishes , becoming zero at the stability limit @xmath131 , corresponding to above values @xmath132 and @xmath133 .
thus the above soliton phase diagram is indeed determined by the instability point @xmath131 ( as a function of ratio @xmath134 ) where both first and second derivatives of the energy are zero .
the shape of the above phase diagram is determined by the character of anisotropy and @xmath3 value and is reported on fig .
[ fig : v2 ] for exchange anisotropy .
first of all , one can see that the dependence @xmath135 is quite complicated and its character change twice , at @xmath136 and at @xmath137 .
it is seen from fig.[fig : v2]a that while there is a quite large region of @xmath4 values , where @xmath138 and @xmath139 are smaller then @xmath140 , the region where @xmath141 is approximately three times smaller .
we have shown that @xmath142 is always larger then @xmath139 although still smaller then @xmath143 . the energy @xmath135 is growing with @xmath3 at @xmath144 only .
[ fig : v2]b demonstrates the divergence of @xmath145 at small @xmath146 as @xmath5 and 2 .
this is related to the fact that in bp soliton the integral describing @xmath71 diverges logarithmically as @xmath93 for @xmath10 , see also above .
overall , fig .
[ fig : v2 ] demonstrates the complicated behavior of solitons phase diagram , which is needed to be understood . to understand better the above complex behavior
, it is instructive to obtain the phase diagram analytically . without loss of generality
, we shall do so for the case of uniaxial anisotropy only .
such analytical treatment is possible since the anisotropy constant @xmath41 in this case enters the problem only via coefficient in the first term of eq . .
this permits to obtain the analytic dependencies @xmath147 and @xmath148 in the inverse form @xmath149 .
note , that the structure of eq .
suggests that the above analytical procedure is applicable for both exchange and uniaxial anisotropy .
our analysis shows that this does not change the situation qualitatively .
moreover , the quantitative results are close to each other , compare figs .
[ fig : v2 ] and [ fig : w4 ] . as it was shown above ,
the soliton phase diagram is determined by the instability point @xmath131 . to obtain an equation for latter point , we rewrite the eq . in the form @xmath150
( function @xmath151 is defined by eq . ) and equate to zero the first and second derivatives of @xmath27 with respect to @xmath42 .
this gives the equation for the dependence @xmath152 in the form @xmath153 where primes mean corresponding derivatives .
then , the equation for @xmath131 can be obtained by substitution of above @xmath154 into the one of the equations determining zero of first or second energy derivatives with respect to @xmath42 .
it turns out that this equation always has real solutions if we formally admit the existence of negative @xmath41 .
the dependence permits to obtain the equation for the phase diagram in the implicit form @xmath155 , which reads @xmath156,\nonumber \\ & & q(z)=\frac{\gamma_0(z)}{2\psi(z)}.\end{aligned}\ ] ] the dependence @xmath157 , obtained by inversion of eq .
, is reported on fig .
[ fig : w3a ] .
it is seen that while the entire curves @xmath157 at @xmath5 and 2 lie at @xmath43 , the curves for @xmath95 lie in this range only partially . since for uniaxial anisotropy @xmath41 can be only positive , only those parts of the curves with @xmath95 where @xmath43 correspond to physically realizable case .
this clarifies the reason why the dependence @xmath158 for skyrmions with @xmath95 begins from finite @xmath71 values , see fig .
[ fig : v2]b .
this different behavior is related to the different kind of convergence of corresponding integrals at @xmath5 and 2 as well as for @xmath95 , see discussion above .
the dependence @xmath159 can be easily recalculated to the solitons phase diagram @xmath160 and @xmath161 .
this diagram is shown on figs .
[ fig : w4]a , b .
it is seen the qualitative coincidence with the numerical curves from figs.[fig : v2]a , b .
the details of behavior can now be better seen then from above numerical curves .
our analysis shows that aforementioned different character of convergence of integrals influences asymptotics of the phase diagram curves at small anisotropies .
this influence can be seen from figs .
[ fig : v2]a and [ fig : w4 ] , where the dependence @xmath162 at @xmath5 and 2 is nonanalytical at @xmath163 , while for @xmath95 it is analytical .
the reason for such behavior can be seen from fig .
[ fig : w3a ] , showing that small @xmath164 correspond to small @xmath42 for @xmath5 and 2 , while for @xmath33 all values of @xmath164 including limiting case @xmath165 correspond to finite @xmath42 .
this means that at @xmath6 the asymptotic analysis of the curves @xmath166 can be done simply by taylor expansion at small @xmath164 , which yields simple monotonic behavior of @xmath167 and @xmath168 . at @xmath10 and 2
the situation is not so simple and requires more complicated analysis .
such analysis can be performed on the base of eqs . , and is quite cumbersome .
the main idea is that functions @xmath169 , @xmath170 and @xmath171 in at small @xmath42 can be represented via exponential integral functions , @xcite which , in turn , may be expanded in asymptotic series .
this procedure for @xmath10 gives following parametric dependence @xmath172 the asymptotics is shown by dashed line on fig .
[ fig : w4]a . in the main logarithmic approximation
this gives simple relations @xmath173 which coincide with the results obtained in ref . by direct numerical modeling of the discreet model , giving @xmath174 and @xmath175 .
these expressions give us an idea about behavior of energy and bound magnons number at small @xmath41 .
namely we see that the energy has the ( approximate ) square root singularity , while magnons number @xmath71 obeys inverse square root law .
the asymptotics is shown on fig .
[ fig : w4]a by dashed line . for @xmath7
the same approximation gives @xmath176 this also yields the divergent @xmath177 and square root peculiarity in the energy @xmath178 .
however , the coefficient before @xmath179 in the energy is much smaller then that at @xmath5 , which makes it almost invisible in the scale of fig .
[ fig : w4 ] .
the details of this behavior are reported on the inset to fig .
[ fig : w4]a .
this shows the similarities and differences between cases @xmath5 and @xmath7 , which can also be seen from fig .
[ fig : w4 ] .
namely , if the topological energy densities for skyrmions with @xmath5 and 2 have both square - root nonanalyticity , for @xmath7 this nonanalyticity reveals only as a small cusp in a close vicinity of @xmath52 .
other peculiarity of @xmath7 case is nonmonotonous behavior of @xmath180 ( decreasing at @xmath181 with subsequent increase ) .
our analysis shows , that the above peculiarities of the skyrmion with @xmath7 are due to interplay between exponential asymptotics of @xmath92 , necessary to cut - off the divergence at @xmath5 and 2 as well as power - law asymptotics , which is sufficient for the convergence at @xmath95 .
in this paper we present a comprehensive theoretical study of the localized topological solitons ( skyrmions ) , stabilized by precessional spin dynamics , for the classical 2d ferromagnet with easy - axial anisotropy on a square lattice .
our efforts were directed primarily towards the study of the role of higher @xmath182 topological charges @xmath30 on the above solitons properties .
our main conclusion is that the interplay between high topological charges , effects of lattice discreteness ( in the form of higher powers of magnetization gradients ) and uniaxial magnetic anisotropy makes many unexpected peculiarities into solitons properties as compared to those in the simplest isotropic continuous model with @xmath183 , containing bp solitons .
similar to previous studies , it turns out that the presence of even weak anisotropy makes solitons dynamic , i.e. , those with nonzero precession frequency for any number @xmath71 of bound magnons .
the minimal consideration of discreteness ( via higher degrees of gradients of magnetization ) yields the existence of some threshold value of both soliton energy ( topological energy density , which is more appropriate characteristic for solitons with @xmath30 ) and the number of bound magnons .
similar to the problem of cone state vortices , @xcite the instability is related to the joint action of discreteness and anisotropy . as a result ,
the critical values of bound magnons @xmath184 and soliton energy @xmath185 is present and nonanalytic dependencies of above threshold values on the anisotropy constant appear at @xmath5 and @xmath7 .
it was shown earlier , @xcite that the variational minimization of @xmath50 ( i.e. energy , incorporating only squares of magnetization gradients ) gives no threshold for soliton existence , i.e. the soliton exists everywhere up to @xmath186 .
this means , that mapping of the initial discrete model even for small anisotropy @xmath57 on the simplest continuum model is wrong and to get the correct description of solitons in 2d fm we have to consider at least fourth powers of magnetization gradients .
this seemingly paradoxical result is actually due to the fact that the terms with @xmath187 are scale - invariant ( so that the corresponding energy has a saddle point ) while the ( stable ) soliton size is determined by the fourth derivatives as well as by a magnetic anisotropy .
our analysis show , that higher ( then fourth ) powers of magnetization gradients do not change the situation qualitatively , rather , in the range of above - studied @xmath188 these terms make only a small quantitative contribution to the solitons phase diagram .
this means , that solitons with @xmath5 can be well studied within the model . in summary , we have shown the existence of stable topological @xmath189 solitons ( skyrmions ) in the 2d ferromagnet with uniaxial anisotropy on a square lattice . since for any 2d lattice symmetry the structure of corresponding energy functional is the same , we may speculate that such textures exist in any 2d magnet with uniaxial anisotropy . we also show that due to different character of the energy and magnons number integrals convergence at @xmath190 and @xmath191 determines the features of so - called solitons phase diagram .
the main nontrivial and unexpected result of the paper is that while the solitons with @xmath5 and @xmath33 have monotonously growing phase boundary functions @xmath34 , the case @xmath7 has peculiar nonmonotonous behavior , determining the transition regime from low to high topological charges .
this means that the designated value of soliton topological charge , which is expected for highly excited state of fm , is neither @xmath5 nor high charges , but rather @xmath7 or @xmath8 . c. e. zaspel , t. e. grigereit , and j. e. drumheller , phys .
lett . * 74 * , 4539 ( 1995 ) ; k. subbaraman , c. e. zaspel , and j. e. drumheller , phys .
lett . * 80 * , 2201 ( 1998 ) ; c. e. zaspel and j. e. drumheller , int . j. mod
b * 10 * , 3649 ( 1996 ) . | we show that stable localized topological soliton textures ( skyrmions ) with @xmath0 topological charge @xmath1 exist in a classical 2d heisenberg model of a ferromagnet with uniaxial anisotropy . for this model
the soliton exist only if the number of bound magnons exceeds some threshold value @xmath2 depending on @xmath3 and the effective anisotropy constant @xmath4 .
we define soliton phase diagram as the dependence of threshold energies and bound magnons number on anisotropy constant .
the phase boundary lines are monotonous for both @xmath5 and @xmath6 , while the solitons with @xmath7 reveal peculiar nonmonotonous behavior , determining the transition regime from low to high topological charges .
in particular , the soliton energy per topological charge ( topological energy density ) achieves a minimum neither for @xmath5 nor high charges , but rather for intermediate values @xmath7 or @xmath8 . |
there is a plethora of astrophysical evidence today , from supernovae measurements , the spectrum of fluctuations in the cosmic microwave background @xcite , baryon oscillations and other cosmological data , indicating that the expansion of the universe is currently accelerating .
the energy budget of the universe seems to be dominated at the present epoch by a mysterious dark energy component , but the precise nature of this energy is still unknown .
many theoretical models provide possible explanations for the dark energy , ranging from a cosmological constant @xcite to super - horizon perturbations @xcite and time - varying quintessence scenarios @xcite , in which the dark energy is due to a smoothly varying ( scalar ) field which dominates cosmology in the present era , such as a time - dependent dilaton field @xcite .
the current astrophysical data are capable of placing severe constraints on the nature of the dark energy , whose equation of state may be determined by means of an appropriate global fit . most of the analyses so far are based on effective four - dimensional robertson - walker universes , which satisfy on - shell dynamical equations of motion of the einstein - friedman form . even in modern approaches to brane cosmology , which are described by equations that deviate during early eras of the universe from the standard friedman equation ( which is linear in the energy density )
, the underlying dynamics is assumed to be of classical equilibrium ( on - shell ) nature , in the sense that it satisfies a set of equations of motion derived from the appropriate minimization of an effective space - time lagrangian .
however , cosmology may not be an entirely classical equilibrium situation @xcite .
the initial big bang or other catastrophic cosmic event , which led to the initial rapid expansion of the universe , may have caused a significant departure from classical equilibrium dynamics in the early universe , whose signatures may still be present at later epochs including the present era .
three of us ( je , nem , dvn ) have indeed proposed one specific model for the cosmological dark energy which is of this type , being associated with a rolling dilaton field that is a remnant of this non - equilibrium phase , described by a generic non - critical string theory @xcite .
we call this scenario ` q - cosmology '
. the central purpose of this paper is to confront the present cosmological data on high - redshift supernovae with non - equilibrium cosmologies @xcite , in which the dark energy relaxes at different rates , and compare the results with the predictions of the conventional @xmath0cdm model @xcite .
we also comment on a model with super - horizon perturbations superposed on an underlying einstein - friedman - robertson - walker universe @xcite .
as we explain in more detail below , care must be taken in interpreting the q - cosmology scenario proposed in refs . @xcite .
since such a non - equilibrium , non - classical theory _ is not described by the equations of motion derived by extremizing an effective space - time lagrangian _ , one must use a more general formalism to make predictions that can be confronted with the current data .
the approach we favour is formulated in the context of string / brane theory @xcite , the best candidate theory of quantum gravity to date .
our approach is based on non - critical ( liouville ) strings @xcite , which offer a mathematically consistent way of incorporating time - dependent backgrounds in string theory .
the basic idea behind such non - critical liouville strings is the following .
usually , in string perturbation theory , the target space dynamics is obtained from a stringy @xmath1-model @xcite that describes the propagation of strings in classical target - space background fields , including the space - time metric itself .
consistency of the theory requires conformal invariance on the world sheet , in which case the target - space physics is independent of the scale characterising the underlying two - dimensional dynamics .
these conformal invariance conditions lead to a set of target - space equations for the various background fields , which correspond to the einstein / matter equations derived from an appropriate low - energy effective action that is invariant under general coordinate transformations .
unfortunately , one can not incorporate in this way time - dependent cosmological backgrounds in string theory , since , to low orders in a perturbative expansion in the regge slope @xmath2 , the conformal invariance condition for the metric field would require a ricci - flat target - space manifold , whereas a cosmological background necessarily has a non - vanishing ricci tensor . to remedy this defect , and thus be able to describe a time - dependent cosmological background in string theory , the authors of ref .
@xcite suggested that a non - trivial rle should be played by a time - dependent dilaton background .
this approach leads to strings living in numbers of dimensions different from the customary critical number , and was in fact the first physical application of non - critical strings @xcite .
the approach of ref .
@xcite was subsequently extended @xcite to incorporate off - shell quantum effects and non - conformal string backgrounds describing other non - equilibrium cosmological situations , including catastrophic cosmic events such as the collision of two brane worlds , _
etc_. in our discussion of such models in this paper , we first review briefly their main predictions .
we then demonstrate the importance of off - shell corrections to the naive equations of motion of non - critical liouville string cosmology , which cause us to modify the naive equations of motion of q - cosmology .
we then demonstrate that the available supernova data are compatible with such non - critical - string - based cosmologies @xcite .
the structure of the article is as follows . in section [ sec : pheno ] we summarise the basic equations of the first set of models we study , namely @xmath0cdm and the super - horizon model .
then , in section [ sec : qcosmo ] we discuss the liouville q - cosmology model of refs . @xcite .
in particular , we discuss technical aspects of q - cosmology , placing the emphasis on the off - shell corrections and deriving the fitting formulae to be used in our analysis .
this part may be omitted by the reader who is more interested in the confrontation of this and other models with the available high - redshift supernova data that we present in section [ sec : data ] .
we find that the @xmath0cdm model fits the data very well , as does the super - horizon model .
we also show that the off - shell corrections to the q - cosmology enable this model to be reconciled with the supernova data on equal footing with @xmath0cdm .
our results and the possible outlook are summarised in section [ sec : summary ] .
in this section we review the basic formulae needed in the phenomenological analysis of observational constraints on the cosmological parameters derived from supernova data and ( in the case of the @xmath0cdm ) also baryon acoustic oscillations , in the context of on - shell models ( i.e. , excluding the q - cosmology model to be discussed later ) .
the relevant dynamics of any such cosmological robertson - walker model may be lumped into the expression for the hubble parameter @xmath3 in terms of the various components of the energy density of the universe , represented in units of the ( present - day ) critical density for a spatially - flat universe , @xmath4 , where @xmath5 for matter , including dark matter , @xmath6 for constant dark energy , i.e. , a cosmological constant @xmath0 , _
etc_. later in this work we discuss fits to the on - shell ( @xmath0cdm and super - horizon ) and off - shell ( q - cosmology ) models discussed in the current and the next sections , respectively . in this case
, we consider a spatially flat robertson - walker universe containing dust - like matter that can be described as an ideal fluid , and a cosmological constant @xmath8 corresponding to a _ constant _ equation of state of the form : @xmath9 where @xmath10 is the pressure ( energy density ) and @xmath11 .
the standard einstein - friedmann equations for such a universe yield : @xmath12 where @xmath13 is the present - day value of the hubble constant . in this model , the universe is assumed to be filled with only non - relativistic matter , and there is no dark energy of any sort .
the metric takes the form : @xmath14 where @xmath15 denotes the gravitational potential .
the authors of ref .
@xcite considered perturbations @xmath15 that correspond to conformal contraction or stretching of the three - dimensional euclidean space , @xmath16 .
splitting the potential @xmath17 into contributions from long - wavelength ( super - horizon ) modes , @xmath18 , and short - wavelength ( sub - horizon ) modes , @xmath19 , and ignoring the latter within our hubble volume , one obtains @xcite : @xmath20 with the normalization @xmath21 .
the analysis of ref .
@xcite used the appropriate einstein s equations to demonstrate that , up to trivial rescaling of spatial coordinates and absorbing the time - independent part of @xmath18 , the time dependence of @xmath22 may be cast in the form : @xmath23 one therefore finds that the expansion rate of the universe , as observed by a local observer who is restricted to observe within our hubble volume , is given by : @xmath24 where @xmath25 . combining and
, one may extract the following relation between @xmath26 and @xmath27 : @xmath28 this model predicts acceleration of the universe at the current era : @xmath29 it can easily be shown by direct comparison with conventional models that the super - horizon perturbations mimic a dark energy model with equation of state @xmath30 in order to evaluate the constraints imposed by astrophysical data on this super - horizon model , we use the following integral of the hubble parameter @xmath31 over @xmath27 : @xmath32 where @xmath33 and @xmath34 are the standard and imaginary error function , respectively .
we assume values @xmath35 , in order for perturbation theory to remain valid @xcite .
we reiterate that we do not address any of the deeper theoretical issues of this model .
there are many dark energy models in which the energy density relaxes in time .
for example , relaxation at a rate proportional to @xmath36 , where @xmath37 is the cosmic time , may arise in certain non - critical string theory models @xcite .
in such a model , the rle of the quintessence field is played by the dilaton field @xcite .
however , at least in the string theory context , a complete treatment of such a rolling dilaton field requires a non - trivial discussion of off - shell physics , which we present in the next section . for the moment
, we neglect this complication ( as might be valid for some non - string models of relaxing dark energy ) , and assume a naive on - shell physical framework in which the dark energy density relaxes according to a generic power law @xmath38 . within this general framework
, @xmath39 corresponds to a cosmological constant @xmath0 and @xmath40 corresponds to a naive interpretation of the q - cosmology model of ref .
@xcite , in which only the conventional type of matter exists and the off - shell terms are not taken into account .
such a relaxing dark energy model would predict a hubble parameter of the form : @xmath41 where @xmath42 represents the present - day dilaton dark energy in the on - shell formulation .
in such a model with @xmath40 , the rolling dilaton dark energy would resemble a spatial curvature contribution , and such naive on - shell models are easily excluded on the same observational grounds that support experimentally the spatial flatness of the universe .
however , @xmath39 is certainly allowed , and it is important to know the permissible range of @xmath43 .
there is a formal correspondence between and a generalized @xmath0cdm model in which the vacuum energy @xmath0 obeys an equation of state with @xmath44 , namely @xmath45 .
however , the underlying physics may be entirely different . for models with relaxing dark energy ,
the standard data analysis shows that both the ` gold ' and the snls dataset favour a value of the parameter @xmath43 appearing that is @xmath46 .
indeed , the combined data prefer @xmath47 ( @xmath48 ) , i.e. , a ( nearly ) constant vacuum energy , as in the @xmath0cdm model .
in this section we explore in more detail q - cosmologies with relaxing vacuum energy in the context of liouville strings , which provide a consistent treatment of off - shell , out - of - equilibrium effects . for the benefit of the reader , we review here the basic formalism and summarize the results , giving details of their derivation in the next two subsections , that may be skipped by a reader not interested in technical aspects .
there are many specific cosmological models that may be formulated within the general framework of liouville strings : see @xcite for an extended discussion and references . here
we limit ourselves to the main features of this cosmological framework , which do not depend on the details of the underlying microscopic string / brane models or compactification .
we simply assume the theory to have been compactified ( somehow ) to a four - dimensional robertson - walker space - time , with a scale factor @xmath49 in the so - called einstein cosmic frame @xcite . in the absence of matter , the liouville - dressing approach of ref .
@xcite and the ( dynamical ) identification of the liouville zero mode with target time lead to generalized conformal invariance conditions for the fields of the gravitational multiplet of the string propagating in a four - dimensional background : @xmath50 where the overdot denotes differentiation with respect to the world - sheet zero mode of the liouville field @xmath51 , the set of @xmath52 are string background fields ( metric , dilaton , _ etc . _ ) , the @xmath53 are the respective world - sheet renormalization - group @xmath54-functions , and @xmath55 is the square root of the central - charge deficit that describes the departure from criticality .
the negative sign in front of @xmath53 on the right - hand side of corresponds to supercriticality for the string , in which case the central charge exceed its critical ( conformally - invariant ) value .
it is in this case that the liouville mode has a time - like signature and can be identified with the target time @xcite . in critical strings
, the vanishing of the @xmath53 would correspond to the ordinary einstein / matter equations of the low - energy field theory derived from strings .
the vanishing conditions correspond to an equilibrium situation . in non - critical string , however , these equations are replaced by , which express the restoration of conformal invariance by the liouville mode . the physical reasons for an initial departure from conformal invariance might be a catastrophic cosmic event , such as the collision of two brane worlds or a big bang @xcite .
such an event would cause a departure from equilibrium , which may not be inappropriate for the early era of the universe . following the identification of target time with the liouville mode ,
the generalized conformal invariance equations yield new dynamical equations that describe the non - equilibrium dynamics of the liouville strings @xcite .
if one further assumes that , after compactification to four space - time dimensions , one obtains a spatially - flat cosmological robertson - walker metric background , e.g. , in the context of a brane model , the equations define the dynamics of the resulting ` liouville q - cosmology ' , as we term such a non - critical string cosmology .
the equations _ are not _ the ordinary equations of motion corresponding to a four - dimensional gravitational effective action , but describe the dynamics of an _ off - shell relaxation process_. as explained in ref .
@xcite , the inclusion of matter does not change qualitatively this non - equilibrium , off - shell nature of such a model .
apart from the formal addition of extra terms in the gravitational @xmath54-functions , resulting from the coupling of matter to the gravitational field , which are needed for general coordinate invariance in the target - space , in most models of physical interest may be included in such a way that the main source of non - criticality is the gravitational and moduli sector , while the matter @xmath53 functions themselves are ( almost ) vanishing .
in such a case , the matter fields feel the presence of the off - shell liouville terms on the left - hand - side of via their gravitational couplings .
the presence of these off - shell liouville terms , which are exclusive to non - critical strings , results in a modification of the critical energy density condition for the liouville cosmology , which ensures spatial flatness .
this critical density is therefore _
different _ from that in conventional ( on - shell ) friedman - robertson - walker ( frw ) cosmologies .
the essential formalism is that of ref .
@xcite , in which all physically relevant quantities should be expressed in the einstein frame , in terms of the einstein cosmic time . the four - dimensional matter action ( including radiation fields ) couples to the dilaton field non -
trivially , in a way that is specific to the various matter species , as a result of purely stringy properties of the effective action @xcite .
a generic @xmath1-model - frame effective four - dimensional action with dilaton potential @xmath56 , which could even include higher string - loop corrections , has the form : @xmath57 - \\ & \quad\frac{1}{16\pi}\int d^4x \sqrt{g}\frac{1}{\alpha ( \phi)}f_{\mu\nu}^2 - i_{\rm m}(\phi , g , { \rm matter})~ , \end{split}\label{matteraction}\ ] ] in the notation of ref .
@xcite , with the various factors @xmath58 encoding information about higher string loop corrections .
also , @xmath59 denotes the radiation field strength and @xmath60 represents matter contributions , which couple to the dilaton @xmath51 in a manner dictated by string theory , with specific scaling laws @xcite under shifts of the dilaton field @xmath61 . in cases @xcite
where only the string tree level plays a rle for late times , the various form factors simplify , e.g. , @xmath62 , _ etc .
_ however , for reasons of generality , here we keep the form . as we shall see , there is sensitivity in the present data to terms in the dilaton potential with higher powers of the string coupling .
when higher loop corrections are important , these factors have a complicated form , for instance one has @xmath63 , with various constants @xmath64 , with the powers of the square of the string coupling @xmath65 counting closed string loops , as appropriate for the gravitational multiplet . for simplicity , in the present work
we ignore the antisymmetric - tensor four - dimensional field , which , as discussed in ref .
@xcite , corresponds to an axion field .
higher loop corrections may result in a modified dilaton potential for the noncritical string theory of the form @xmath66 where the central charge deficit @xmath67 encodes the microscopic non - equilibrium physics @xcite . in the models we are considering in this work ,
@xmath68 and the string theory is supercritical @xcite . assuming a normal fluid form for matter or radiation , with stress tensor @xmath69 in the einstein frame @xcite
, we obtain the following gravitational equations of motion in our case ( we work here in units @xmath70 , where @xmath71 is the four - dimensional planck constant ) @xcite : @xmath72 where @xmath73 is the einstein - frame robertson - walker cosmic time @xcite and the potential of the dilaton assumes the form , including loop corrections @xmath74 .
we note that the equations differ from those holding in on - shell cosmologies by the liouville out - of - equilibrium contributions @xmath75 , which are exclusive to our treatment @xcite .
the equations need to be supplemented by the appropriate curci - paffuti equation @xcite expressing the renormalisability of the @xmath1-model fields @xcite , which is also exclusive to our non - critical string approach .
this equation relates the dilaton world - sheet @xmath54-function to the rest of the @xmath54-functions in the problem ( for the graviton , _ _ e__tc . ) , and essentially expresses the ( time - dependent ) central charge deficit in terms of the physical fields in the problem , @xmath76 where the overdot henceforward denotes a derivative with respect to the cosmic time @xmath73 , the ` @xmath77 ' indicate terms due to higher order in string loops , and the values of @xmath78 and @xmath79 are given by @xmath80 with @xmath81 the deceleration parameter : @xmath82 for completeness , we mention that the equations lead , after standard manipulations @xcite , to the ( non-)conservation equation of matter in the presence of the non - equilibrium contributions : @xmath83 these equations are used below to obtain a general expression for the hubble parameter as a function of the redshift , which we use later in our fit to data .
the above analysis contains non - linear equations that are hard to solve analytically .
a detailed numerical analysis is performed in ref .
@xcite , where we refer the interested reader for details . instead , for our phenomenological purposes here , we shall attempt an approximate analytic solution at late epochs of the universe , appropriate for our phenomenological fits . throughout this work we work in units where the present scale factor is normalised to one ,
@xmath84 . to solve in the various epochs of the universe , it is convenient first to split the energy density of matter into radiation @xmath85 , baryonic @xmath86 and dark - matter @xmath87 components , and to use simple equations of state for the dilaton fluid , as supported by our theoretical model @xcite : @xmath88 we have assumed a non - trivial equation of state @xmath89 for the dark matter component , which as we discuss below , stems from the fact that its scaling with the scale factor @xmath26 is _ exotic _ , different from the ordinary dust scaling @xmath90 .
one can split the matter evolution equation into various components , using . in the solution of ref .
@xcite without matter , at late eras the dilaton @xmath51 and the scale factor @xmath91 vary as follows with the cosmic time @xmath73 : @xmath92 where @xmath93 , @xmath94 are appropriate positive constants .
the form of the scale factor implies the following form for the deceleration parameter at late epochs of the universe : @xmath95 the values of the constants are such that the present era in the history of the universe is characterised by @xmath96 , as follows from the fact that the deceleration parameter of the universe is currently observed @xcite to be @xmath97 .
we notice that the asymptotic behaviour of the model of ref .
@xcite without matter , at very late times , indicates that @xmath98 and thus @xmath99 implying that @xmath100 , varying very little at late epochs . in the present case ,
where matter effects are important , the formulae need some modification . in general
, the dilaton and the scale factor differ from the particular form @xcite , in which case one obtains a much more complicated behaviour for @xmath101 than the one given in , which can only be studied numerically .
a complete numerical analysis is performed in ref .
@xcite , where we refer the reader for details .
we recall , for completeness , that the analysis of ref .
@xcite indicates evidence for a past deceleration of the universe at redshifts larger than @xmath102 . for our generic phenomenological purposes in this work we can make several physically sensible simplifying assumptions , which allow us to carry some of the results of ref .
@xcite through to the present case , allowing for some analytic treatment .
specifically , the dilaton and scale factors are assumed to take the form @xmath103 as dictated by a general analysis of the dilaton equation in the model .
notice that , in view of the general discussion in ref .
@xcite , we no longer maintain the explicit form of @xmath104 used in the matter - less model of ref .
@xcite , however we do maintain the scaling relation , which , according to the numerical analysis of ref .
@xcite , seems to be a good approximation in the eras of interest to us here .
in general @xcite , there may be a more complicated behaviour than or .
for instance , loop and other corrections may yield power - series corrections in the ( perturbative ) string coupling @xmath105 such corrections could lead to appreciable deviations in the behaviours of various fields and other cosmological parameters even at the present era .
in fact , the numerical analysis of ref .
@xcite indicates evidence for a past deceleration of the universe at redshifts smaller than @xmath106 .
nevertheless , even in these more general models , the hubble parameter maintains @xcite a scaling similar to at the late eras of interest to us here , which implies that some of the qualitative features of our analytic approach used here are valid , to a good approximation , in these more complicated cases .
on using and performing some elementary manipulations , we easily obtain : @xmath107 a consistent assumption , as we shall demonstrate now , is that the matter energy density , @xmath108 , includes dust ( baryonic matter ) and exotic scaling ( dark ) components , which can behave like radiation - like contributions : @xmath109 indeed , taking into account , we observe that the above assumption is consistent with the conservation equation provided that the central charge deficit at late eras scales as @xmath110 with @xmath111 a constant , namely the central charge of the asymptotic conformal field theory to which the string model flows at infinite time @xcite .
notice that , if one allowed an exotic @xmath112 scaling of matter of the form @xmath113 , which characterises the dilaton dark energy @xmath114 , as we discuss below , then @xmath67 should contain also terms of the form @xmath115 .
this would lead , in general , to logarithmic scaling @xmath116 of the various energy density components , which would complicate the situation , as far as the validity of the approximation is concerned .
although such terms may exist in some models , we do not consider them in this work .
here we assume the scaling for the matter energy densities in the eras of interest to us .
for more complete , numerical treatments we refer the interested reader to ref .
@xcite . to proceed with a consistency check and estimates on the scaling behaviour of @xmath117 ,
it is necessary to make some assumptions about the form of the potential @xmath74 .
as already mentioned , we assume that it is generated by higher string loops . in brane - inspired models such as those considered in refs .
@xcite , one has open strings living on the brane world .
the corrections then will be proportional to an extra power of the string coupling @xmath118 . assuming loop corrections of the form @xmath119 and taking into account , we observe that the dark matter density yields : @xmath120 as we see from , , , , the quantity @xmath121 is negative .
we also note that the late - era scaling , which is required for consistency ( on account of ) , can easily be guaranteed , provided the terms @xmath122 which otherwise would yield a @xmath112 scaling are suppressed for the redshift region @xmath123 of interest to us .
this would require a @xmath124 of order @xmath125 , according to our assumption that @xmath81 does not change much in the region in which we are interested .
, indicating past deceleration of the universe .
however , even in such a case , the absence of @xmath112 terms in matter today might be a consistent assumption .
in fact , the numerical solution of ref .
@xcite indicates that the matter does develop an exotic scaling @xmath112 but at much later eras than the present one . ]
notice that the dust - like scaling does not receive contributions from the string loop corrections , and in fact a comparison between and yields a mathematically consistent result with @xmath126 free parameter to be constrained by fitting data and/or other microscopic model considerations .
in contrast , the exotic scaling @xmath127 receives contributions from the loop corrections , proportional to @xmath54 , whose sign can not be fixed by our generic considerations , and depends on the details of the microscopic string / brane theory model , and contributions due to the off - shell liouville terms @xmath128 which are positive according to our assumption that @xmath129 : @xmath130 the asymptotic behaviour of the dilaton dark energy may be evaluated from the definition of @xmath114 : @xmath131 where the dots have the same meaning as in .
it is interesting to observe that the dilaton dark energy component contains a part scaling like @xmath112 , which is positive and of order @xmath132 , but also parts scaling like @xmath90 and @xmath127 , which appear also in the matter energy density .
however , in contrast to the matter case , the ` dust'-like contributions to the dilaton dark energy do not have a fixed sign or magnitude , as they depend on the string loop correction parameter @xmath133 , @xmath134 when combined with the corresponding parts in @xmath108 , in order to obtain the precise scaling of the hubble parameter with the redshift @xmath27 , @xmath3 , from the first of equations , we observe that @xmath135 where we have been careful to indicate the relative signs of the various terms , as predicted by the model with the approximations made so far . scaling , denoted by dots .
thus , the exotic ( dark ) matter @xmath136 could be positive .
this case seems to be favoured by the data , as discussed in the text , and it is in agreement with the numerical analysis of ref . @xcite . ]
notice the independence of the exotic scaling on the loop parameter @xmath54 , which does not enter the expression for @xmath3 .
thus , it is a feature of this liouville model that _ the sign of the dust - like contribution appearing in @xmath3 is not fixed _ , since such terms receive contributions from the dilaton dark energy .
indeed , in modern string / membrane models , one may obtain negative dust - like contributions to the dilaton potential from compactification of kaluza - klein graviton modes in brane models , _ etc .
in contrast , the exotic scaling terms appear positive . when fitting the data , in order to distinguish ordinary matter from dilaton dark energy effects that scale similarly with the redshift , one would have to perform also measurements of the equation of state of each component , which we do not attempt in this work . in the specific model @xcite
considered here as our pilot study , the values of the various @xmath137 appearing in the expression for @xmath3 are in principle predicted by the underlying microscopic dynamics of the model , as we have seen above , but we leave this exercise for future work . however ,
since some of the @xmath137 involve highly model - dependent coefficients of string loop corrections and other , currently unknown , details of microscopic models , and , moreover , in the above derivation several assumptions have been made , e.g. , about the form of the potential for the dilaton , the form of the dilaton and the scale factor at late epochs , _ etc .
_ , which may not always be valid @xcite , we consider at present the various parameters appearing in the expression for @xmath3 as arbitrary parameters to be fixed by the data . to be even more general , we keep the exponent of the exotic scaling as a free fitting parameter as well . from the above considerations , in particular , , it is also clear that in general @xmath138 may be negative , as its sign depends on the sign of the corrections to the dilaton potential . the final result of our parametrisation for @xmath3 in the q - cosmology framework , which will be compared with the astrophysical data in the next section , in order to obtain information on the basic cosmological parameters of the model , is therefore : @xmath139 with the densities @xmath140 corresponding to present - day values ( @xmath141 ) and @xmath142 the reader should recall that in the model described in this section @xmath143
. however , in view of the various approximations employed in our analytic treatment of @xmath55-cosmology above , which , as already mentioned , may not be valid in the present era , when matter contributions are important , the exponent @xmath144 is treated from now on as a fitting parameter .
the above formulae are valid for late eras , such as the ones pertinent to the supernova and other data ( @xmath145 ) that we use in this work .
we stress once again that the various @xmath137 contain contributions from _ both _ dark energy and matter energy densities .
as explained previously , @xmath146 does not merely represent ordinary matter effects , but also receives contributions from the dilaton dark energy .
in fact , the sign of @xmath146 depends on details of the underlying theory , and it could even be _
negative_. for instance , kaluza - klein graviton modes in certain brane - inspired models @xcite yield negative dust contributions . in a similar vein
, the exotic contributions scaling as @xmath147 are affected by the off - shell liouville terms of q - cosmology .
it is because of the similar scaling behaviours of dark matter and dilaton dark energy that we reverted to the notation @xmath137 , @xmath148 in . to disentangle the ordinary matter and dilaton contributions
one may have to resort to further studies on the equation of state of the various components , which we do not study in this article .
more generally , one could have included a cosmological constant @xmath149 contribution in , which may be induced in certain brane - world inspired models .
we do not do so in this work , as our primary interest is to fit q - cosmology models @xcite , which are characterised by dark energy densities that relax to zero .
we now discuss the observables available to test the above models using mainly data on high - redshift supernovae and ( to a limited extent ) baryon acoustic oscillations .
we use the supernovae data reported in refs .
@xcite , which are given in terms of the distance modulus @xmath150 we use for @xmath151 the standard formula relating the luminosity distance to the redshift @xmath27 @xmath152 where @xmath153 is given by , and in the three models discussed in the previous sections , namely @xmath0cdm , super - horizon and q - cosmology respectively .
we note that this observable depends on the expansion history of the universe from @xmath27 to the present epoch , and recall that most of the available supernovae have @xmath154 , though there is a handful with larger values of @xmath27 . for standard on - shell cosmologies ,
complementary information is provided by the data on baryon acoustic oscillations ( bao ) @xcite , which show up in the galaxy - galaxy correlation function at @xmath155 . however , for reasons to be discussed later on in the article , the application of such an analysis to the off - shell q - cosmology model is an open issue , since the underlying theoretical framework needs to be re - evaluated .
the formulae for @xmath3 in the various models discussed above , namely , and , serve as the basis for our analysis below .
we now present our fits of the above models to the presently available astrophysical data on high - redshift supernovae @xcite , and ( in the case of @xmath0cdm ) also baryon acoustic oscillations ( bao ) @xcite .
we start our discussion with the @xmath0cdm and super - hubble - horizon models , and then proceed to the case of off - shell liouville q - cosmology models @xcite .
measurements @xcite are available of the distance moduli for 157 supernovae in a so - called ` gold ' sample @xcite , observed by ground - based facilities and the hubble space telescope : see fig .
[ fig : gold_datafit ] .
we compare these with the measurements of 71 other high - redshift supernovae published more recently @xcite by the supernova legacy survey ( snls ) , which are accompanied by a reference sample of 44 nearby sn ia , yielding a total of 115 data points : see fig .
[ fig : snls_datafit ] . for both samples ,
the data are expressed in terms of the observed distance modulus @xmath156 derived from the sn ia light curves and the corresponding redshift @xmath27 .
the best - fit cosmological parameters are then extracted by minimizing : @xmath157 where @xmath158 and @xmath159 are the measured value for the distance modulus and the corresponding error for a specific supernova @xmath160 , respectively , @xmath161 is the intrinsic dispersion of the absolute magnitudes of the supernovae , and @xmath162 is the theoretical prediction for a given model as calculated by , , where @xmath151 is given in units of megaparsecs .
this fit and all subsequent analyses are performed with the root @xcite implementation of the minuit function minimization and error analysis code @xcite .
( black dashed - dotted line ) ; ( iii ) @xmath0cdm model for @xmath163 ( red long - dashed line ) ; ( iv ) the super - horizon model for @xmath164 ( blue short - dashed line ) ; and ( v ) off - shell q - cosmology model for @xmath165 and @xmath166 ( green thick solid line ) . ] ( black dashed - dotted line ) ; ( iii ) @xmath0cdm model for @xmath167 ( red long - dashed line ) ; ( iv ) the super - horizon model for @xmath168 ( blue short - dashed line ) ; and ( v ) off - shell q - cosmology model for @xmath169 and @xmath170 ( green thick solid line ) . ] both the data and the predictions of the various models are expressed in the following as residuals , @xmath171 , from their predicted values in an empty universe ( @xmath172 ) : @xmath173 in addition to presenting the ` gold ' and the snls data sets respectively , figures [ fig : gold_datafit ] and [ fig : snls_datafit ] also display the predictions of various cosmological models to be detailed later on in the article .
in addition to the 157 sn ia belonging to the ` gold ' sample analysed in ref .
@xcite , a so - called ` silver ' dataset of 29 sn 1a is also available , with lesser spectrometric and photometric quality compared to the ` gold ' one . although the results presented here were obtained with the ` gold ' dataset , the analysis was also repeated including the ` silver ' supernovae data with a total of 186 supernovae with results comparable to those presented here , proving the robustness of the analysis .
the analysis was also repeated for the combined sample of the ` gold ' ( 157 supernovae ) plus the snls sample ( 71 supernovae ) , which yields a sample of 228 supernovae .
the @xmath174 values obtained for the best fits to the various models are discussed in the remainder of this section . for the @xmath0cdm model @xcite , assuming a flat universe , we find the @xmath174 and parameter values listed in table [ tb : lcdm ] , which are in agreement with those presented in ref . @xcite .
the corresponding confidence limits in the @xmath175 plane are shown in fig . [
fig : lcdm ] . [ cols="<,<,<,<,<",options="header " , ] we observe from fig .
[ fig : deceler ] that , with the best - fit values of the q - cosmology parameters in the model of refs .
@xcite , the transition from a decelerating expansion of the universe to an accelerating expansion would have taken place when @xmath176 .
this is almost compatible at the 1-@xmath1 level with the result of the analysis of ref .
@xcite , which indicated a critical @xmath177 .
however , this was based on a very simple ( linear ) assumed form for the deceleration @xmath178 , which is not the case of our model , as also seen in fig .
[ fig : deceler ] .
it should be possible in the future to explore observationally nonlinear forms of @xmath179 , with the aim of improving the fit and reducing the uncertainty in the value of the critical @xmath180 for the transition from deceleration to acceleration .
we have shown in this paper that the two main data sets for high - redshift supernovae are compatible , at least as far as their constraints on the cosmological models studied in this paper are concerned .
the standard @xmath0cdm model fits the supernova data _ very well _ , and the super - horizon dark matter model also fits the supernova data _ quite well_. both of these models are on - shell , i.e. , they satisfy the pertinent einstein s equations . however , off - shell cosmology models can still be compatible with the data .
as we discussed above , off - shell effects are important in our q - cosmology model .
introducing an extra parameter to allow for these off - shell effects , we find that _ the q - cosmology model may fit the supernovae data as well as the standard @xmath7cdm
model_. * in the @xmath0cdm model that combines a cosmological constant with cold dark matter , assuming a flat universe we find @xmath181 ( @xmath182 ) for @xmath183 , corresponding to a value of @xmath184 for the dark energy density .
this value is consistent with the one predicted in ref .
@xcite . * in the super - horizon dark matter model ( shcdm ) of ref .
@xcite , we find @xmath185 ( @xmath186 ) for the best - fit case with @xmath187 * for the q - cosmology model of refs .
@xcite , assuming @xmath188 : @xmath189 ( @xmath190 ) : for @xmath191 , @xmath192 , yielding @xmath193 .
just as this analysis was being completed , the three - year wmap data ( wmap3 ) were released @xcite . assuming a flat universe , the wmap3 data by themselves yield within the @xmath0cdm model the result @xmath194 .
this central value is slightly lower than our result , but the error is larger , and the two results are compatible within their errors . the wmap collaboration has combined their data with other cmb data , and the most precise result , obtained by combining wmap3 with the 2df galaxy redshift survey , is @xmath195 , which differs from our combined result by about @xmath196 .
on the other hand , combining wmap3 with large - scale structure data gives larger central values of @xmath197 .
the smallest errors are those obtained by combining wmap3 with the sloan digital sky survey lrg data set : @xmath198 and with the cfhtls lensing data : @xmath199 .
these results straddle our combined fit to the supernova data .
the data on baryon acoustic oscillations ( bao ) provide complementary information @xcite . within the @xmath0cdm model and assuming a flat universe , the bao yield an independent estimate @xcite @xmath200 , which is completely compatible with the value we obtained from the supernova data . combining the two results , we find @xmath201 .
we leave for a future occasion the exploration of the constraints imposed by the bao data on the q - cosmology models .
when confronting the bao data with the q - cosmology model , it will be desirable also to formulate more precisely the predictions of this model , which is also an important subject for future work .
there is one important issue which should be stressed at this point .
the exotic matter scaling : @xmath202 with @xmath203 of the q - cosmology model at late eras , if valid at earlier times , would have dominated the epochs characterised by redshifts @xmath204 . as a result ,
the positions of the peaks of the baryon acoustic oscillations would have been modified , leading to stringent constraints on the models . however , in such a scenario the phenomenological observables currently used in analyses of the bao @xcite would be inapplicable to q - cosmology models .
we hope to come back to this issue in a future publication .
n.e.m . wishes to thank g. diamandis , v. georgalas and a. b. lahanas for discussions , and the physics department of athens university for its hospitality during the last stages of this work .
the work of n.e.m . is partially supported by funds made available by the european social fund ( 75% ) and national ( greek ) resources ( 25% ) - epeaek b - pythagoras .
v.a.m.acknowledges support by the european union through the rtn contract : hprn - ct-2002 - 00292 , _ the third generation as a probe for new physics_. the work of d.v.n . is supported by d.o.e .
grant de - fg03 - 95-er-40917 .
a. g. riess _ et al .
_ [ supernova search team collaboration ] , astron . j. * 116 * , 1009 ( 1998 ) [ arxiv : astro - ph/9805201 ] ; b. p. schmidt _ et al .
_ [ supernova search team collaboration ] , astrophys .
j. * 507 * , 46 ( 1998 ) [ arxiv : astro - ph/9805200 ] ; s. perlmutter _ et al .
_ [ supernova cosmology project collaboration ] , astrophys .
j. * 517 * , 565 ( 1999 ) [ arxiv : astro - ph/9812133 ] ; j. p. blakeslee _ et al . _
[ supernova search team collaboration ] , astrophys . j. * 589 * , 693 ( 2003 ) [ arxiv : astro - ph/0302402 ] ; a. g. riess _ et al .
_ [ supernova search team collaboration ] , astrophys .
j. * 560 * , 49 ( 2001 ) [ arxiv : astro - ph/0104455 ] .
a. upadhye , m. ishak and p. j. steinhardt , phys .
d * 72 * ( 2005 ) 063501 [ arxiv : astro - ph/0411803 ] .
a. g. riess _ et al .
_ [ supernova search team collaboration ] , astrophys .
j. * 607 * , 665 ( 2004 ) [ arxiv : astro - ph/0402512 ] . n. a. bahcall , j. p. ostriker , s. perlmutter and p. j. steinhardt , science * 284 * , 1481 ( 1999 ) [ arxiv : astro - ph/9906463 ] .
e. w. kolb , s. matarrese , a. notari and a. riotto , arxiv : hep - th/0503117 ; e. w. kolb , s. matarrese and a. riotto , arxiv : astro - ph/0506534 .
i. antoniadis , c. bachas , j. r. ellis and d. v. nanopoulos , phys . lett .
b * 211 * , 393 ( 1988 ) ; nucl .
b * 328 * , 117 ( 1989 ) ; phys .
b * 257 * , 278 ( 1991 ) .
j. r. ellis , n. e. mavromatos and d. v. nanopoulos , phys . lett .
b * 619 * , 17 ( 2005 ) [ arxiv : hep - th/0412240 ] ; j. r. ellis , n. e. mavromatos , d. v. nanopoulos and m. westmuckett , int .
j. mod .
phys . a * 21 * ( 2006 ) 1379 [ arxiv : gr - qc/0508105 ]
. g. a. diamandis , b. c. georgalas , n. e. mavromatos and e. papantonopoulos , int .
j. mod .
a * 17 * , 4567 ( 2002 ) [ arxiv : hep - th/0203241 ] ; g. a. diamandis , b. c. georgalas , n. e. mavromatos , e. papantonopoulos and i. pappa , int .
j. mod .
a * 17 * , 2241 ( 2002 ) [ arxiv : hep - th/0107124 ] .
g. a. diamandis , b. c. georgalas , a. b. lahanas , n. e. mavromatos and d. v. nanopoulos , arxiv : hep - th/0605181 , phys .
b , in press ( doi:10.1016/j.physletb.2006.09.035 ) .
green , j.h .
schwarz and e. witten , _ superstring theory _ ,
i & ii ( cambridge university press , 1987 ) . f. david , mod .
lett . a * 3 * , 1651 ( 1988 ) ; j. distler and h. kawai , nucl .
b * 321 * , 509 ( 1989 ) ; j. distler , z. hlousek and h. kawai , int .
j. mod .
a * 5 * , 391 ( 1990 ) ; see also : n. e. mavromatos and j. l. miramontes , mod .
a * 4 * , 1847 ( 1989 ) ; e. dhoker and p. s. kurzepa , mod .
lett . a * 5 * , 1411 ( 1990 ) .
j. r. ellis , n. e. mavromatos and d. v. nanopoulos , phys .
b * 293 * , 37 ( 1992 ) [ arxiv : hep - th/9207103 ] ; mod .
a * 10 * , 1685 ( 1995 ) [ arxiv : hep - th/9503162 ] . _
invited review for the special issue of j. chaos solitons fractals _ , vol .
10 , ( eds . c. castro amd m.s .
el naschie , elsevier science , pergamon 1999 ) 345 [ arxiv : hep - th/9805120 ] ; m. gasperini , f. piazza and g. veneziano , phys .
d * 65 * , 023508 ( 2002 ) [ arxiv : gr - qc/0108016 ] . g. curci and g. paffuti , nucl .
b * 286 * , 399 ( 1987 ) .
m. minamitsuji , m. sasaki and d. langlois , phys .
d * 71 * , 084019 ( 2005 ) [ arxiv : gr - qc/0501086 ] .
p. astier
_ , astron .
astrophys .
* 447 * ( 2006 ) 31 [ arxiv : astro - ph/0510447 ] . for a critical comprehensive recent review
see : e. v. linder , arxiv : astro - ph/0507308 and references therein .
d. j. eisenstein _ et al .
_ , astrophys .
j. * 633 * ( 2005 ) 560 [ arxiv : astro - ph/0501171 ] ; s. cole _ et al .
_ [ the 2dfgrs collaboration ] , mon . not .
soc . * 362 * , 505 ( 2005 ) [ arxiv : astro - ph/0501174 ] .
r. brun and f. rademakers , nucl .
. meth .
a * 389 * ( 1997 ) 81 .
f. james and m. roos , comput .
commun .
* 10 * ( 1975 ) 343 . | we discuss fits of cosmological dark energy models to the available data on high - redshift supernovae .
we consider a conventional model with cold dark matter and a cosmological constant ( @xmath0cdm ) , a model invoking super - horizon perturbations ( shcdm ) and models based on liouville strings in which dark energy is provided by a rolling dilaton field ( q - cosmology ) .
we show that a complete treatment of q - cosmology requires a careful discussion of non - equilibrium situations ( off - shell effects ) .
the two main high - redshift supernova data sets give compatible constraints on @xmath0cdm and the other models .
we recover the well - known result that @xmath0cdm fits very well the combined supernova data sets , as does the super - horizon model .
we discuss the model - dependent off - shell corrections to the q - cosmology model that are relevant to the supernova data , and show that this model fits the data equally well .
this analysis could be extended to other aspects of cosmological phenomenology , in particular to the cmb and baryon acoustic oscillations , which have so far been treated using on - shell models .
, , , observational cosmology , dark energy , supernovae98.80.es , 95.36.+x , 98.65.dx |
after the successful applications of lscher s method @xcite to the elastic scattering of _ elementary _ scalar and fermionic particles ( see , e.g. , @xcite ) , we have chosen the massive schwinger model ( qed@xmath0 ) with 2 flavours to determine scattering phases in meson - meson scattering .
similar to four - dimensional qcd , mesons occur because of confinement .
this two - dimensional model also exhibits charge screening and it possesses a nontrivial vacuum structure .
moreover , there exist analytical predictions which enable us to test the numerical results .
lscher s relation connects elastic scattering phases in infinite volume and two - particle energies in finite volumes . in 2 dimensions it has the simple form @xmath1 using the energy - momentum relation , @xmath2 can be calculated from the mass and the two - particle energies , which are accessible to monte carlo simulations .
the massive schwinger model with 2 flavours has the following continuum euclidean action : @xmath3 bosonization of the model leads in the strong coupling limit to the sine - gordon model , plus corrections @xcite . since the particle spectrum of the sine - gordon model
is known analytically , one can derive that the particle spectrum of the massive two - flavour schwinger model consists of a pseudoscalar isotriplet ( `` pion '' ) with g - parity g=+1 and a mass @xmath4 according to @xmath5 and a scalar isosinglet with g=+1 and mass @xmath6 @xcite . in the sine - gordon model
also the elastic scattering phases have been calculated @xcite and will form the basis for our numerical tests . from the corrections to the sine - gordon model only the `` @xmath7''-particle ( a pseudoscalar isosinglet , g=1 ) with mass @xmath8 is known @xcite . & & + & & + & & + @xmath9 & @xmath10 & @xmath11 & @xmath12 & @xmath13 & @xmath14 & @xmath12 & @xmath13 & @xmath14 & @xmath12 & @xmath13 & @xmath14 & @xmath12 & @xmath13 & @xmath14 & @xmath12 & @xmath13 & @xmath14 & @xmath12 & @xmath13 & @xmath14 + && & 0.05 & + & & 0.05 in ( 1,2 ) ( 0,0)(1,0)2(0,1)2 ( 0,0)(0,1)3(1,0)1 & & & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & & & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & + & & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & & & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & + & + &&+ & 0.05 & + & + & 0.05 in ( 1,2 ) ( 0,0)(1,0)2(0,1)2 ( 0,0)(0,1)3(1,0)1 & &+ & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & &+ & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & + & + & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & &+ & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & + & + + &+ & & 0.05 & & & 0.05 in ( 1,2 ) ( 0,0)(1,0)2(0,1)2 ( 0,0)(0,1)3(1,0)1 & + & & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & & & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & + & & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & & & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & + & + &+ & + & 0.05 & &+ & 0.05 in ( 1,2 ) ( 0,0)(1,0)2(0,1)2 ( 0,0)(0,1)3(1,0)1 & + & + & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & &+ & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & + & + & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & &+ & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & + & + + + & & & 0.05 & & & 0.05 & + & & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & & & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & + & & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & + & & 0.05 in ( 2,2 ) ( 0,0)(1,0)3(0,1)2 ( 0,0)(0,1)3(1,0)2 & & + + & &+ & 0.05 & &+ & 0.05 & + & + & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & &+ & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & + & + & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & + & + & 0.05 in ( 2,2 ) ( 0,0)(1,0)3(0,1)2 ( 0,0)(0,1)3(1,0)2 & &+ + + & + & & 0.05 & & & 0.05 & + & & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & & & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & + & & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & & & 0.05 in ( 2,2 ) ( 0,0)(1,0)3(0,1)2 ( 0,0)(0,1)3(1,0)2 & + & + + & + & + & 0.05 & &+ & 0.05 & + & + & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1
( 0,0)(0,1)2(1,0)4 & &+ & 0.05 in ( 4,1 ) ( 0,0)(1,0)5(0,1)1 ( 0,0)(0,1)2(1,0)4 & + & + & 0.05 in ( 3,2 ) ( 0,0)(1,0)2(0,1)2 ( 2,1)(1,0)2(0,1)1 ( 0,0)(0,1)1(1,0)1 ( 0,1)(0,1)2(1,0)3 & &+ & 0.05 in ( 2,2 ) ( 0,0)(1,0)3(0,1)2 ( 0,0)(0,1)3(1,0)2 & + & + + the lattice formulation of the schwinger model with staggered fermions and compact wilson action for the gauge field is given by : @xmath15 energy eigenstates are classified according to irreducible representations of the group of symmetry transformations leaving the time slices fixed .
the representation @xmath16 of the continuum time slice group ( cts ) is characterized by the representation @xmath12 of the su(2 ) and the quantum numbers @xmath13 and @xmath14 associated with parity and g - parity .
the representation @xmath17 of the lattice time slice group ( lts ) is characterized by the quantum numbers @xmath9 , @xmath10 and @xmath11 , which correspond to shift in space , inversion and charge conjugation on the lattice .
restriction of an irreducible cts representation to the subgroup lts will in general lead to reducible representations . for the one-(two-)particle operators , which transform according to the representations @xmath12 of rank 2 ( 4 ) , every lattice symmetry sector couples to two ( four ) continuum sectors , as shown in table [ tab : lts - cts ] .
[ fig : m_pi(m)-b2b4 ]
we first verified the analytical prediction for the relation of the pion mass to the bare mass of the fermions ( see fig .
[ fig : m_pi(m)-b2b4 ] ) .
we have good agreement between the analytical formula and our simulations : for @xmath18 = 4 we extract an exponent of 0.689(10 ) as compared to 2/3 of eq .
( [ eq : m_pi(m ) ] ) . the deviation between the measured points and the analytical curve decreases with increasing @xmath18 , which can be explained by the fact that the continuum limit corresponds to @xmath19 .
this is supported by additional measurements for higher values of @xmath18 , up to @xmath18 = 10 , for which the points tend towards the analytically predicted curve . [
fig : m_eff(m)-b7 ] in fig .
[ fig : m_eff(m)-b7 ] we show the particle content of the model found by us after the investigation of almost all lattice symmetry sectors .
unfortunately we are not yet able to identify numerically the analytically predicted singlet states mentioned above .
with increasing @xmath18 , a problem related to the topological properties of the model arises : the tunnelling rate between sectors of different topological charges decreases strongly , so that we are facing an ergodicity problem . to estimate its impact
we analyzed the dependence of the pion mass on the topological charge @xmath20 of the configurations by comparing measurements on configurations associated with different topological charges ( see fig .
[ fig : m_pi(q)+p(q)-b1m01 ] ) . within the error bars ,
no significant deviation is visible .
[ fig : m_pi(q)+p(q)-b1m01 ] because of the unexpected complexity of the one - particle spectrum for the massive schwinger model we have not yet been able to determine scattering phases satisfactorily .
helpful discussions with m. gckeler are gratefully acknowledged .
furthermore we wish to thank the hlrz jlich and the rechenzentrum of the rwth aachen for providing the necessary computer time .
9 m. lscher , nucl .
b354 ( 1991 ) 531 , nucl .
b364 ( 1991 ) 237 .
m. gckeler , h. a. kastrup , j. westphalen and f. zimmermann , nucl .
b425 ( 1994 ) 413 , and references therein .
m. gckeler , h. a. kastrup , j. viola and j. westphalen , nucl .
b(proc . suppl . ) 47 ( 1996 ) 831 .
s. coleman , ann .
( 1976 ) 239 .
a. b. zamolodchikov and a. b. zamolodchikov , ann .
( 1979 ) 253 . | we discuss the possibility of extracting phase shifts from finite volume energies for meson - meson scattering , where the mesons are fermion - antifermion bound states of the massive schwinger model with su(2 ) flavour symmetry .
the existence of analytical strong coupling predictions for the mass spectrum and for the scattering phases makes it possible to test the reliability of numerical results . |
during the brief history of extrasolar planet investigations , our understanding of the relative populations of different types of planets has been limited by the observational biases of the techniques employed . with the advent of sophisticated transit searches and hypersensitive radial velocity measurements ,
significant progress has been made discovering various types of planets that orbit stars with periods up to a few years .
less progress has been made in discovering planets in longer orbits , and particularly around nearby m dwarfs , which account for at least 74% of the stellar population within 10 pc @xcite .
m dwarfs offer fertile ground for companion searches , as @xcite have inferred that a high fraction of m dwarfs host terrestrial planets at short orbital periods .
less is known about the populations of jupiter - mass planets and brown dwarfs around m dwarfs , particularly at orbital periods longer than a few years . to understand how m dwarf planetary systems form and evolve , we must probe the full regime of companion masses and orbital periods .
transit techniques are geometrically biased towards companions with small orbits , while radial velocity techniques are biased towards massive companions with short periods that exert large gravitational accelerations on their host stars .
direct imaging techniques are limited to young , giant planets at large separations .
astrometric techniques , which measure the positions of stars on the plane of the sky , are most sensitive to jovian - type planets in jovian - type orbits . while radial velocity observing programs are now becoming sensitive to such companions @xcite , the astrometric results presented here have longer observational baselines , of up to 13 years .
furthermore , astrometry can detect companions with a large range of inclinations and orientations , and allow for the determination of orbit inclinations and accurate companion masses . to date the majority of nearby extrasolar planets around m dwarfs
have been discovered by radial velocity searches , which tend to select the brightest m dwarfs . as discussed in more detail in @xmath2[sec : analysis ] , in ground - based imaging programs the brightest targets generally have the noisiest astrometric residuals due to the short exposures required and the lack of comparably bright reference stars . with the exception of gj 1214 ,
five m dwarfs in our sample were found to have planets using radial velocity techniques , and are among the brightest targets in our astrometric program .
an extreme case is the k dwarf bd @xmath010 3166 , for which we are not sensitive to sub - stellar companions , but for which we provide the first accurate parallax . for comparison ,
we have included six additional m dwarfs that are less bright , less massive , and closer , and therefore more favorable to companion detection via astrometry . to calibrate our analysis ,
we have also included three confirmed stellar binaries with clear photocentric perturbations for which we have characterized the orbits .
these binaries were chosen from the roughly two dozen binaries in our observing program with clear astrometric perturbations because we have observed multiple orbital periods , and can most accurately characterize the orbits .
astrometric solutions for proper motion and parallax are given for each of the 16 systems targeted , plus orbital solutions for three binaries .
a detailed analysis of the astrometric residuals is given to search for companions to the 12 m dwarf systems without close stellar companions .
periodograms of the astrometric residuals have been generated , along with detection limits based on simulations of 10 million hypothetical companions to each star .
these are the first results of a larger recons survey for companions orbiting more than 250 red dwarfs within 25 pc for which we have at least five years of coverage . as observations continue
, this sample will grow , further constraining the population of brown dwarf and super - jupiter companions in long period orbits around m dwarfs . finally , to provide context for these results we provide a comprehensive list of the 17 m dwarfs within 25 pc having exoplanets as of 1 july 2014 , including the six targeted in this work .
the 0.9 m telescope at ctio is equipped with a 2048 @xmath3 2048 tektronix ccd camera with 0.401 pixel@xmath4 plate scale @xcite .
only the center quarter of the chip is used for astrometric and photometric observations , yielding a 6.8 square field of view .
astrometric exposures are taken through one of four filters , @xmath5 ( old ) , @xmath5 ( new ) , @xmath6 , or @xmath7 ( old ) , @xmath5 ( new ) , @xmath6 , and @xmath7 filters are 5438 , 5475 , 6425 , and 8075 , respectively . ] ( hereafter without subscripts , and the @xmath8 filters combined ) .
depending on the brightnesses of the science targets , reference stars , and sky conditions , exposure times vary from 20 to 1200 s for targets with 9 @xmath9 .
for optimal centroiding , exposure times are set so that either science or reference stars have maximum peak adu of @xmath1050,000 ( digital saturation occurs at 65,537 adu ) .
observations are almost always made within @xmath11 30 minutes of a science target s transit to minimize the corrections required for differential color refraction , as described in @xcite .
three to five frames are typically taken each night , depending primarily on the exposure time required . to enable routine calibration of the science images , bias and dome flat frames are taken nightly .
instrument setups for most stars have been kept constant during the 13 years of observations .
however , we have used two @xmath8 filters , dubbed the `` old '' tek#2 @xmath8 filter ( @xmath12 = 5438 , @xmath13 = 1026 ) and `` new '' tek#1 v filter ( @xmath12 = 5475 , @xmath13 = 1000 ) , because the `` old '' filter cracked in 2005 february .
the `` new '' @xmath8 filter was used between 2005 and 2009 .
the `` old '' @xmath8 filter was reinstated in 2009 july after confirming that the crack in the corner did not significantly affect astrometric residuals . as discussed in @xcite
, a reliable parallax can be obtained using data from both filters as long as at least 1 - 2 years of data ( depending on observing frequency ) have been taken in each filter .
reductions containing both `` old '' and `` new '' @xmath8 frames can exhibit offsets of a several milliarcseconds ( mas ) in residuals on both axes .
this has been mitigated by choosing close - in reference stars , and only using frames taken near the meridian . in total ,
7 of the 16 systems discussed in this paper were observed astrometrically in the @xmath8 filter .
further details about the filters and their effects on the astrometry can be found in @xcite and @xcite .
the paths traced on the sky by science stars result from the combinations of proper motions and parallactic shifts .
details of the data reduction process used to separate these motions are given in @xcite and @xcite .
briefly , we ( 1 ) use sextractor @xcite to measure centroids , ( 2 ) define a six - constant plate model to find plate constants ( given in equation ( 4 ) of @xcite ) , ( 3 ) assume that ensembles of reference stars have zero mean parallax and proper motion , ( 4 ) solve least - square equations for multi - epoch images ( given in equation ( 5 ) of @xcite ) , and ( 5 ) convert from relative parallax to absolute parallax by estimating the distances of the reference stars photometrically .
our typical centering precision is 2.1 - 3.5 mas , or 0.5 - 0.9% of a pixel , depending on the filter , with @xmath14 being the best and @xmath8 being the worst . to correct the relative parallax to an absolute parallax ,
photometric distances are estimated by comparing @xmath15 colors to @xmath16 for single , main - sequence stars in the recons 10 pc sample @xcite .
a distance is estimated for each reference star , and the correction to absolute parallax is then computed using the weighted mean distance of the entire reference field . the uncertainty on the correction is determined using equation ( 6 ) in @xcite . in the case of a binary with a given combination of magnitude and mass differences , where @xmath17 is the photocentric semi - major axis of the orbit of the primary , @xmath18 is the fractional mass ( @xmath19 ) , @xmath20 is the relative luminosity ( @xmath21 ) , and @xmath22 is the semi - major axis of the relative orbit of the two components .
the perturbation we detect here is @xmath17 . ]
, we detect its photocenter orbit around its barycenter , in addition to the motions due to parallax and proper motion .
hence , the residuals of our typical binary star s parallax reduction are significantly offset from zero . in order to get a better parallax result and calculate the photocenter s orbital elements
, we first carry out a standard reduction for proper motion and parallax .
we then fit a photocentric orbit to the residuals , i.e , we treat these residuals as a binary orbit , using the techniques described in @xcite .
based on the orbital elements we calculate , this photocentric orbit in components of r.a . and decl .
is then removed from the centroids of the science star at each epoch .
finally , we re - calculate proper motion and parallax using these corrected centroids .
the final `` cleaned '' proper motions and parallaxes are the values given for the binaries in table [ tab1 ] .
after one iteration , the parallax errors are reduced to those typical of similar program stars , and the residuals are significantly reduced and consistent with the mean errors found for our overall program .
@xmath15 photometry was obtained at the ctio 0.9 m using the same instrumental setup used for the astrometry frames . as for astrometry observations , bias and dome flat frames
are taken nightly for basic image calibration .
all science stars were observed at airmass @xmath23 .
exposure times were chosen to reach a signal - to - noise ratio ( s / n ) @xmath24 100 for science stars in each of the @xmath15 filters .
combinations of fields that provided 10 or more standard stars from @xcite and/or e - regions from @xcite were observed several times each night to derive transformation equations and extinction curves .
further details of photometric data reductions , the definition of transformation equations , errors , etc .
, can be found in @xcite and @xcite .
spectroscopic observations used to the provide spectral types in table 2 were made at the ctio 1.5 m using the r - c spectrograph and loral 1200 @xmath3 800 ccd camera between 2003 and 2006 .
grating # 32 was used in first order with a tilt of 15.@xmath251 , and observations were made using a 2 slit .
the order - blocking filter og570 was utilized to provide spectra covering the range of 6000 - 9000 with a resolution of 8.6 . for calibration , bias frames , dome flats , and sky flats
were taken at the beginning of each night .
further details regarding the 1.5 m spectroscopy program and associated data reduction , including assignment of spectral types , can be found in @xcite .
astrometric residuals represent the deviation in a target s measured position from the solution for proper motion and parallax , given for all 16 target stars in table [ tab1 ] .
the residuals for each star are plotted in figure [ fig1 ] .
each filled circle represents the mean of typically three to five frames taken in a single night , with a corresponding estimate of the nightly mean error .
open circles represent nights with only one frame , which are included in the parallax measurement but are not included in the following analysis of the residuals .
the three binaries ( gj 748ab , lhs 1582ab , and lhs 3738ab ) exhibit large , periodic perturbations in both r.a . and decl .
, indicating the presence of a companion in each case .
because the systems are unresolved in our images , we calculate a photocentric semi - major axis , as previously discussed .
the remaining targets generally have flat residuals , although some are more scattered ( gj 849 ) than others ( gj 1128 ) .
targets with the flattest residuals , such as denis j1048 - 3956 , generally have at least 8 reference stars that closely and evenly surround the target , and are bright enough to have a least 1,000 peak counts when the target has 50,000 .
fainter targets tend to have the flattest residuals because they have a larger number of suitable reference stars and require longer exposures , which smooth out short - term seeing effects that produce psfs with poorly - defined centroids .
the residuals were analyzed for the presence of companions using lomb - scargle periodograms , shown in figure [ fig2 ] , generated by the method given in @xcite .
this method weights each data point by its estimated nightly mean error , and produces a normalized periodogram for the data sequence with power ranging from 0 to 1 .
we only attempt to detect companions with periods greater than two years but less than the length of the observations , and have only generated periodograms for those ranges .
the data were oversampled in increments of one day to create smooth periodograms .
as @xcite have shown , oversampling the periodogram does not significantly increase the risk of false periodogram peaks . as the data are irregularly spaced , it was not possible to use an analytical formula for the false alarm probabilities ( fap ) . instead , empirical fap distribution functions were calculated as outlined in @xcite by generating periodograms for 10,000 sets of random noise .
each of the 10,000 data sets has the same number of observations as the actual data , with the same observation times and errors as the actual data .
the randomized values of each point within a given data set are assigned from a normal distribution with a standard deviation representing the distribution of offsets from zero of the actual data points .
each of these 10,000 periodograms was then sampled in a grid of periods with five day increments between two years and the length of observations to find the highest power occurring in each periodogram .
the fap for an observed periodogram peak of a given power is equal to the fraction of the 10,000 data sets with peaks greater than or equal to that power .
the normalized powers corresponding to fap values of 0.1 , 0.01 , and 0.001 are shown as dashed lines in figure [ fig2 ] .
this corresponds to 10% , 1% , and 0.1% probabilities that the peak is due solely to random noise and the cadence of observations . as a calibration , periodograms and fap functions
were also generated for the three confirmed stellar binaries for which astrometric perturbations are clearly evident .
these binaries have significant periodogram peaks ( fap @xmath26 0.001 ) in both the r.a . and decl .
axes at periods corresponding to their orbital periods , while the remaining 13 targets have no significant peaks .
this indicates that our periodograms are sensitive to astrometric perturbations , and that we have not detected any significant periodicity in the residuals of the remaining 12 m dwarfs and one k dwarf without companions .
given that there are no companions evident in the residuals of the non - binary targets , we aim to establish lower limits for the companion masses and periods to which we are sensitive , and would have detected were they present .
for each target we ran a simulation of 10 million hypothetical companions orbiting each star with masses chosen randomly from a uniform distribution between 0.5 and 80 @xmath1 , and periods chosen randomly from a uniform distribution between 2 years up to the length of observations .
although we have observed perturbations as short as 1.2 years , our practical lower limit is 2 years .
all geometric parameters ( inclination , eccentricity , time of periastron , longitude of ascending node , and longitude of periastron ) were assigned randomly from a uniform distribution of all possible values , including eccentricity , which was allowed to be as high as 0.99 .
the program then calculates the astrometric perturbation that each simulated companion would induce on the primary , using a primary mass calculated from the mass - luminosity relations in @xcite ( equations 5a and 5b ) and the revised relation for the lowest mass stars in @xcite . for den j1048 - 3956 and lp 944 - 020 , which are too faint for the relations , we assume masses of 0.08 @xmath27 .
we assume that the companion does not contribute significantly to the overall flux in each system , so that the photocenter of the system is concentric with the primary star .
this assumption proves problematic when considering brown dwarf companions to the two latest m dwarfs , den j1048 - 3956 and lp 944 - 020 .
however , the trigonometric parallaxes of these targets agree well with their photometric distance estimates ( columns 12 and 13 of table [ tab2 ] ) , and we conclude that neither have nearly equal luminosity companions .
the goodness - of - fit between the observed data @xmath28 with errors @xmath29 , and a flat line with @xmath30 was determined using the reduced chi squared statistic @xmath31 where @xmath32 is the number of degrees of freedom , given by @xmath33 for @xmath34 data points and @xmath35 fitted parameters . because we did not attempt to fit a model to the data , but only to analyze how well a simulation fits the data , we set @xmath35 equal to zero .
as data points with small error bars are more heavily weighted , we discarded epochs that have unrealistic errors smaller than 1 mas representing less than 5% of the epochs because such points overconstrain the orbits that can be fit .
simulated orbits for which @xmath36 was greater than 4 in at least one of either r. a. or decl .
we consider to be orbits that we would have detected .
this threshold of 4 is based empirically on the values plotted in figure [ fig3 ] , where the lower panel is an inset of the upper .
the plotted @xmath36 values were calculated for ( 1 ) the stars included in this paper , ( 2 ) calibration stars with flat residuals we use to monitor potential fluctuations in equipment and our data reduction pipeline , and ( 3 ) additional known binaries with perturbations .
we compare these targets residuals to the case of flat residuals , i.e. , an exact astrometric solution with no perturbation .
as expected , the binaries with perturbations have large @xmath36 values , indicating that a flat line is a poor fit to the data .
the two solid points inside the @xmath37 box are long term perturbations with large gaps in the astrometric observations .
as more data are collected , those two points will move to larger @xmath36 values outside the box , but we include the points here for completeness . for those stars with no perturbations , all have @xmath36 less than 4 in both axes , and are centered around 1 .
this indicates that we are accurately calculating our measurement errors , and provides an empirical @xmath36 value of 4 , above which we are sensitive to perturbations .
table [ tab1 ] gives the parallax and proper motion results for the 16 systems , with details about the astrometric observations ( filters used , number of seasons observed , number of frames used in reductions , time coverage , span of time , and the number of reference stars ) and results ( relative parallaxes , parallax corrections , absolute parallaxes , proper motions , position angles of the proper motions , and the derived tangential velocities based on relative proper motions and parallaxes ) .
all but two of the sixteen systems have parallax errors of @xmath102 mas or less .
bd @xmath010 3166 and gj 876 have larger errors due to combinations of faint reference stars and short exposures .
corrections to absolute parallax are generally less than @xmath102 mas , so systematics in the corrections should not significantly affect the results . three targets ( gj 317 , gj 667c , and bd @xmath010 3166 , ) have corrections of @xmath103@xmath38-@xmath384 mas due to reddening of the reference stars , which skews their photometric distance estimates . in these cases ,
we adopt a generic correction of 1.50 @xmath11 0.50 mas .
the per observation precision for each target is listed in column 16 , representing the mean of the observation errors in r.a . and decl .
the percentage of companions eliminated listed in column 17 is discussed in @xmath2[sec : discuss ] .
lccccccccrrrrrrccc + gj 317 & 08 40 59.21 & @xmath023 27 22.6 & r & 5c & 75 & 2009.04 - 2013.38 & 4.35 & 7 & 64.04 @xmath11 1.45 & 1.50 @xmath11 0.50 & 65.54 @xmath11 1.53 & 930.7 @xmath11 1.1 & 330.5 @xmath11 0.13 & 65.5 & 4.99 & & + bd @xmath010 3166 & 10 58 28.79 & @xmath010 46 13.4 & i & 7s & 71 & 2004.43 - 2011.50 & 7.07 & 6 & 13.84 @xmath11 3.04 & 1.50 @xmath11 0.50 & 15.34 @xmath11 3.08 & 185.9 @xmath11 1.5 & 269.1 @xmath11 0.67 & 52.4 & 10.85 & & + gj 581 & 15 19 26.83 & @xmath007 43 20.1 & v & 14s & 267 & 2000.58 - 2013.38 & 12.80 & 11 & 157.67 @xmath11 1.57 & 1.12 @xmath11 0.17 & 158.79 @xmath11 1.58 & 1224.3 @xmath11 0.4 & 266.0 @xmath11 0.03 & 36.5 & 7.93 & 96% & a + gj 1214 & 17 15 18.92 & @xmath3904 57 50.1 & i & 4c & 80 & 2010.39 - 2013.38 & 3.00 & 9 & 68.20 @xmath11 1.26 & 1.88 @xmath11 0.18 & 70.08 @xmath11 1.27 & 945.5 @xmath11 1.4 & 142.0 @xmath11 0.17 & 63.9 & 5.02 & & + gj 667c & 17 18 58.82 & @xmath034 59 48.6 & v & 11s & 140 & 2003.52 - 2013.38 & 9.86 & 5 & 139.38 @xmath11 1.98 & 1.50 @xmath11 0.50 & 140.88 @xmath11 2.04 & 1154.1 @xmath11 0.6 & 101.0 @xmath11 0.05 & 38.8 & 7.66 & 93% & + gj 849 & 22 09 40.34 & @xmath004 38 26.8 & v & 11s & 135 & 2003.52 - 2013.39 & 9.86 & 5 & 113.78 @xmath11 1.97 & 2.27 @xmath11 0.30 & 116.05 @xmath11 1.99 & 1118.0 @xmath11 0.5 & 90.8 @xmath11 0.04 & 45.7 & 9.53 & 81% & + gj 876 & 22 53 16.75 & @xmath014 15 49.2 & v & 11s & 85 & 2003.52 - 2013.39 & 9.87 & 6 & 210.97 @xmath11 3.99 & 2.14 @xmath11 0.57 & 213.11 @xmath11 4.03 & 1149.4 @xmath11 1.1 & 125.7 @xmath11 0.11 & 25.6 & 8.43 & 99% & + + gj 1061 & 03 35 59.72 & @xmath044 30 45.5 & r & 13s & 194 & 1999.62 - 2012.95 & 13.32 & 7 & 269.92 @xmath11 1.29 & 0.94 @xmath11 0.08 & 270.86 @xmath11 1.29 & 827.7 @xmath11 0.3 & 117.7 @xmath11 0.04 & 14.5 & 7.59 & 79% & a + lp 944 - 020 & 03 39 35.25 & @xmath035 25 43.8 & i & 8s & 59 & 2003.95 - 2012.94 & 8.99 & 10 & 154.53 @xmath11 1.03 & 1.36 @xmath11 0.10 & 155.89 @xmath11 1.03 & 408.3 @xmath11 0.3 & 48.5 @xmath11 0.07 & 12.4 & 2.13 & 94% & a + gj 1128 & 09 42 46.36 & @xmath068 53 06.1 & v & 13s & 167 & 2000.23 - 2013.12 & 12.89 & 8 & 153.54 @xmath11 0.75 & 0.73 @xmath11 0.11 & 154.27 @xmath11 0.76 & 1123.0 @xmath11 0.2 & 356.1 @xmath11 0.02 & 34.5 & 3.17 & 80% & a + denis j1048 - 3956 & 10 48 14.56 & @xmath039 56 07.0 & i & 13s & 200 & 2001.15 - 2013.27 & 12.13 & 11 & 247.23 @xmath11 0.60 & 0.85 @xmath11 0.10 & 248.08 @xmath11 0.61 & 1531.6 @xmath11 0.2 & 229.5 @xmath11 0.01 & 29.3 & 2.92 & 97% & a + scr 1138 - 7721 & 11 38 16.76 & @xmath077 21 48.5 & i & 11s & 134 & 2003.25 - 2013.27 & 10.03 & 12 & 119.60 @xmath11 1.01 & 0.81 @xmath11 0.07 & 120.41 @xmath11 1.01 & 2143.3 @xmath11 0.4 & 287.8 @xmath11 0.02 & 84.4 & 4.20 & 69% & a + proxima cen & 14 29 43.02 & @xmath062 40 46.7 & v & 14s & 205 & 2000.57 - 2013.25 & 12.68 & 5 & 766.41 @xmath11 0.91 & 1.72 @xmath11 0.50 & 768.13 @xmath11 1.04 & 3850.8 @xmath11 0.6 & 282.4 @xmath11 0.02 & 23.8 & 4.83 & 99% & a + + lhs 1582ab & 03 43 22.08 & @xmath009 33 50.9 & r & 11s & 102 & 2000.87 - 2012.94 & 12.06 & 7 & 48.84 @xmath11 1.18 & 2.00 @xmath11 0.26 & 50.84 @xmath11 1.21 & 509.4 @xmath11 0.3 & 52.7 @xmath11 0.06 & 47.5 & 3.88 & & a + gj 748ab & 19 12 14.60 & @xmath3902 53 11.0 & v & 10s & 154 & 2004.45 - 2013.39 & 8.95 & 11 & 97.77 @xmath11 1.15 & 2.22 @xmath11 0.41 & 99.99 @xmath11 1.22 & 1857.8 @xmath11 0.5 & 107.4 @xmath11 0.02 & 88.1 & 5.80 & & + lhs 3738ab & 21 58 49.13 & @xmath032 26 25.5 & r & 12s & 151 & 1999.64 - 2012.81 & 13.17 & 12 & 50.82 @xmath11 1.01 & 1.40 @xmath11 0.21 & 52.22 @xmath11 1.03 & 535.2 @xmath11 0.3 & 229.1 @xmath11 0.06 & 48.6 & 2.50 & & a + [ tab1 ] lrrrcrrrcccrrcc + gj 317 & 12.01 & 10.84 & 9.37 & 3 & 7.934 & 7.321 & 7.028 & m3.5 v & 1 & 0.35 & 15.26 @xmath11 0.36 & 9.70 @xmath11 1.53 & 12 & + bd @xmath010 3166 & 10.03 & 9.58 & 9.19 & 3 & 8.611 & 8.300 & 8.124 & k3.0 v & 2 & 0.85 & 65.19 @xmath11 13.64 & & & a + gj 581 & 10.56 & 9.44 & 8.03 & 3 & 6.706 & 6.095 & 5.837 & m3.0 v & 2 & 0.30 & 6.30 @xmath11 0.06 & 6.60 @xmath11 1.03 & 12 & + gj 1214 & 14.71 & 13.27 & 11.50 & 3 & 9.750 & 9.094 & 8.782 & m4.5 v & 1 & 0.14 & 14.27 @xmath11 0.24 & 12.42 @xmath11 2.00 & 12 & + gj 667c & 10.34 & 9.29 & 8.09 & 3 & 6.848 & 6.322 & 6.036 & m1.5 v & 2 & 0.36 & 7.10 @xmath11 0.10 & 9.41 @xmath11 1.49 & 12 & + gj 849 & 10.38 & 9.27 & 7.87 & 3 & 6.510 & 5.899 & 5.594 & m3.0 v & 2 & 0.42 & 8.62 @xmath11 0.15 & 5.73 @xmath11 0.92 & 12 & + gj 876 & 10.18 & 8.97 & 7.40 & 3 & 5.934 & 5.349 & 5.010 & m3.5 v & 2 & 0.27 & 4.69 @xmath11 0.09 & 3.46 @xmath11 0.54 & 12 & + + gj 1061 & 13.09 & 11.45 & 9.47 & 6 & 7.523 & 7.015 & 6.610 & m5.0 v & 3 & 0.11 & 3.69 @xmath11 0.02 & 3.55 @xmath11 0.60 & 12 & + lp 944 - 020 & 18.69 & 16.39 & 13.98 & 3 & 10.725 & 10.017 & 9.548 & m9.0 v & 4 & 0.08 & 6.42 @xmath11 0.04 & 7.04 @xmath11 1.32 & 11 & + gj 1128 & 12.74 & 11.36 & 9.62 & 3 & 7.953 & 7.385 & 7.037 & m4.0 v & 2 & 0.15 & 6.48 @xmath11 0.03 & 6.33 @xmath11 1.00 & 12 & + denis j1048 - 3956 & 17.37 & 14.98 & 12.47 & 4 & 9.538 & 8.905 & 8.447 & m8.0 v & 2 & 0.08 & 4.03 @xmath11 0.01 & 4.48 @xmath11 0.73 & 10 & + scr 1138 - 7721 & 14.78 & 13.20 & 11.24 & 4 & 9.399 & 8.890 & 8.521 & m5.0 v & 3 & 0.11 & 8.31 @xmath11 0.07 & 9.45 @xmath11 1.71 & 12 & + proxima cen & 11.13 & 9.45 & 7.41 & 3 & 5.357 & 4.835 & 4.384 & m5.0 v & 2 & 0.11 & 1.30 @xmath11 0.01 & 1.15 @xmath11 0.18 & 12 & b + + lhs 1582ab & 14.69j & 13.33j & 11.60j & 4 & 9.799j & 9.177j & 8.854j & m4.5 vj & 1 & & 19.67 @xmath11 0.47 & 13.27 @xmath11 2.25 & 12 & c + gj 748ab & 11.10j & 9.95j & 8.47j & 3 & 7.087j & 6.572j & 6.294j & m3.5 vj & 2 & & 10.00 @xmath11 0.12 & 7.69 @xmath11 1.26 & 12 & c + lhs 3738ab & 15.78j & 14.29j & 12.46j & 3 & 10.654j & 10.091j & 9.761j & m4.5 vj & 5 & & 19.15 @xmath11 0.38 & 18.50 @xmath11 2.96 & 12 & c
+ [ tab2 ] nine of the 16 targets in this paper have parallaxes previously published by recons , and are noted in column 18 .
the results presented here supersede those published previously by recons because additional data and improved reduction techniques have been used , as discussed in detail in @xcite .
the identical parallax of lp 944 - 020 is also presented in @xcite as part of a study of the stellar hydrogen burning limit . for bd @xmath010 3166 ,
we did not run simulations because it is too massive and far away for us to detect any type of substellar companion .
however , we do provide the first accurate parallax , and conclude that bd @xmath010 3166 is not physically related to the star with a similar proper motion , lp 731 - 076 , that is @xmath40 away @xcite .
photometric and spectroscopic results are provided in table [ tab2 ] .
@xmath15 photometry was taken using the ctio 0.9 m ( number of nights of photometry in column 5 ) , with errors in @xmath41 mag @xcite .
@xmath42 photometry was retrieved from the two micron all sky survey ( 2mass , @xcite ) catalog .
spectral types are given in column 9 with references in column 10 .
mass estimates were calculated as discussed in @xmath2[sec : analysis ] .
the photometric distances are calculated using the @xmath43 distance relations ( number of relations in column 14 ) detailed in @xcite . for systems with photometric and trigonometric distances that agree within the errors , we conclude that they lack nearly equal luminosity companions .
the trigonometric distances of gj 748ab and lhs 1582ab are greater than their photometric distances due to companion contributions to the systems total flux .
the two distances of lhs 3738ab agree well , indicating that the companion is significantly fainter than the primary .
gj 317 and gj 849 have discrepant ( at 3.5@xmath44 and 3.1@xmath44 , respectively ) photometric and trigonometric distances , which does not necessarily mean that these stars have stellar companions , as main sequence stars within the same spectral type can vary somewhat in luminosity .
@xcite note a radial velocity drift in their observations of gj 849 .
@xcite also note this drift , and constrain the minimum companion mass to @xmath45 .
our astrometry would show a photocenter shift for unequal mass components , as discussed in @xmath2[sec : observe ] .
only components of roughly equal luminosity and mass would provide the additional flux with no perturbation . such a companion would have been observed to separations as close as 1 in our images , which corresponds to @xmath109 au . at a semimajor axis of 9 au ,
the orbital period is 29.5 years for an equal mass companion .
this results in velocities for each component 9.1 km / s for edge - orbits .
thus , for most orbital inclinations , such a stellar companion is ruled out by the radial velocity data .
therefore , it is unlikely that a stellar companion similar to the primary is contributing to the overluminosity we observe .
figure [ fig4 ] shows the range of periods and masses for which 90% of simulated companions would have been detected , based on the simulations for objects with masses from 0.5 to 80 @xmath1 . as discussed in @xmath2[sec : analysis ] , the masses , periods , and orbital parameters of the simulated companions were assigned randomly from a uniform distribution . for this discussion we set the dividing line between planets and brown dwarfs at 13 @xmath1 , and the dividing line between brown dwarfs and stars at 80 @xmath1 .
the bottom panel of figure [ fig4 ] is an inset of the top , showing the best case targets in more detail .
the noisy nature of the lines is due to the sizes of the bins used in the simulations .
the bin sizes were chosen to achieve a reasonably high resolution , while still having enough simulated companions in each bin .
the minimum detectable companion mass is smallest for companions with long periods , which produce the largest amplitude perturbations in the astrometric data .
the lower panel indicates that for the best case targets ( stars at close distances and of low mass ) .
we are most sensitive to jovian - type planets in jovian - type orbits . for a companion at a given mass and orbital period
, the amplitude of the resulting astrometric perturbation depends on the orbital parameters of the system , and the mass and distance of the primary .
the detection limits we report are based on simulations of companions with a wide range of masses , periods , and orbital parameters .
therefore we give a few representative examples of how the results in figure [ fig4 ] translate into astrometric perturbations in mas . in the case of a face - on , circular orbit ,
a 20 @xmath1 companion in a 4 year orbit around gj 581 would cause a 16 mas perturbation , while a 15 @xmath1 companion in an 8 year orbit would cause a 20 mas perturbation , and a 10 @xmath1 companion in a 12 yr orbit would cause a 17 mas perturbation . for circular ,
face - on orbits around proxima centauri , companions of 1.5 , 1 , and 0.5 @xmath1 in orbits of 4 , 8 , and 12 years would cause perturbations of 12 , 13 , and 8 mas , respectively .
these values are significantly greater than the per observation precisions listed table [ tab1 ] 7.93 mas for gj 581 and 4.83 mas for proxima .
thus , the 90% detection thresholds given in figure [ fig4 ] are reasonable for the four planet hosts observed longer than eight years , column 17 of table [ tab1 ] gives the percentages of simulated _ brown dwarf _ companions , ranging from 81 99% , eliminated with orbits between two and eight years , and masses between 13 and 80 @xmath1 .
approximately 92% of all simulated brown dwarfs have been eliminated as companions to those stars known to host exoplanets .
for the six more astrometrically favorable targets , we calculate the percentages of simulated _ planetary _ companions eliminated with orbits between two and eight years and masses between 1 and 13 @xmath1 , with results ranging from 69 99% .
we have eliminated @xmath1086% of all simulated planets with masses of 1 13 @xmath1 around these six astrometrically favorable stars , and effectively all brown dwarf companions in orbital periods of 2 8 years .
photocentric orbital solutions for the three binaries are shown in figure [ fig5 ] with the corresponding orbital parameters given in table [ tab3 ] . from our astrometric data for gj 748 ab
, we find an orbital period of 2.504 @xmath11 0.025 years , which is consistent with the two detailed studies of the system by @xcite , who found p = 2.466 @xmath11 0.008 years , and @xcite , who found p = 2.469 @xmath11 0.001 years using hst fine guidance sensor data .
however , we determine an eccentricity of 0.06 , which is inconsistent with the value of 0.45 found in both of the hst studies .
we utilized the orbit - fitting code described in @xcite and set starting eccentricities of 0.05 to 0.95 in increments of 0.05 ; regardless , our data converged to the e = 0.06 value each time .
lrrrrrrr gj 748ab & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + lhs 1582ab & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + lhs 3738ab & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 130.5@xmath11 5.1 + [ tab3 ] the discrepancy between our eccentricity and that of the hst studies is likely due to our observations of gj 748 ab having been taken at the two different @xmath8 filters discussed in @xmath2 2.1 .
while the two filters are photometrically identical within measurable errors @xcite , they are not astrometrically identical .
we have analyzed the astrometric residuals for over 500 targets without detectable perturbations in the three different filters ( @xmath46 and @xmath14 ) over the length of our observing program .
the r and i filters are stable , but astrometric offsets in the v filters are evident over the time period when the problematic `` new '' @xmath8 filter was used .
these offsets have been mitigated as discussed in @xmath2 2.1 , allowing data from both @xmath8 filters to be used to produce reliable parallax results . in the case of gj 748 ab , we are able to recover the correct period , but the offsets in the residuals are likely contributing to the errant eccentricity .
we presently do not possess enough observations to perform a reduction of gj 748 ab without the problematic @xmath8 filter data .
we include the current solution because it is our only system for which an accurate period has been published , against which to compare our results .
in contrast to our photocenter data , the fgs observations resolve the system into two components at 15 epochs over 1.8 years .
they are exquisitely sensitive to the separation and position angle of the secondary from the primary , and are to be preferred to our ground - based results for the eccentricity .
a clever suggestion by hugh harris of usno has been suggested to solve this dilemma . because in an unresolved system the center of mass location is unknown , the zero points for the residuals in r.a . and
are unknown . by shifting the zero points and fitting the residuals ,
a different eccentricity may be derived .
we await the acquisition of resolved data for several more systems before exploring this technique so that a robust analysis can be accomplished . at present errors on the eccentricities in table [ tab3 ] should be treated with caution .
the orbital solutions for lhs 1582ab and lhs 3738ab are updated and improved over those presented in @xcite , which were the first orbits presented for each system . in addition to demonstrating the astrometric detection and characterization of unresolved companions , these results can provide additional dynamically determined masses for m dwarfs , once the systems have been resolved .
in narrow - angle field astrometry programs such as the one outlined here , the lowest mass companions detectable by astrometry are dependent on a number of factors . these include the apparent brightness of the host star and availability of suitable reference stars , which affect the precision of the astrometric measurements . additionally , the target star mass , companion mass , the luminosity ratio , orbital separation , and system distance determine the size of the astrometric perturbation .
the m dwarfs with planets that have been discovered so far , including the six in this paper , are not the most favorable to our astrometric observations .
they are among the brighter red dwarfs , with the latest type stars at m3.5v for the radial velocity detections included in this paper .
we are rarely sensitive to planets around these stars .
however , we are able to rule out the presence of a large fraction of potential brown dwarf companions with masses of @xmath47 @xmath1
. as observations continue , we will be able to rule out companions at longer periods , particularly in the cases of gj 317 and gj 1214 , which we have so far only observed for 3 4 years .
in addition , we have demonstrated that we are sensitive to planets with masses of @xmath48 @xmath1 around m dwarfs that are more favorable to astrometric observations .
of particular interest , we find that proxima centauri has no companions more massive than 2 @xmath1 with periods of 2 5 years and more massive than 1 @xmath1 for 5 12 years .
this is to be contrasted to the results of other companion searches of proxima centauri . using hst fine guidance sensor data , @xcite report a companion detection sensitivity of 1 @xmath1 at a 60 day period .
based on 7 years of rv observations , @xcite preclude the presence of companions with @xmath49 at periods @xmath50 years .
together these studies eliminate all jupiter mass planets around proxima centauri for orbital periods out to 12 years .
gj 163 b & 04 09 15.6 & @xmath053 22 25 & 66.59 @xmath11 1.79 & 2 & 1,2 & 0.03 & 0.024 & 9 & + gj 163 c & & & & & & 0.02 & 0.070 & 9 & + gj 163 d & & & & & & 0.09 & 1.654 & 9 & +
gj 176 b & 04 42 55.7 & @xmath3918 57 29 & 110.00 @xmath11 2.00 & 2 & 1,2 & 0.03 & 0.024 & 10 &
+ gj 179 b & 04 52 05.7 & @xmath3906 28 35 & 80.82 @xmath11 3.78 & 2 & 1,2 & 0.82 & 6.264 & 11 & + gj 317 b & 08 40 55.7 & @xmath023 28 00 & 65.34 @xmath11 0.39 & 3 & 1,3,4 & 1.81 & 1.895 & 4 & a + lhs 2335 b & 10 58 35.0 & @xmath031 08 38 & 50.55 @xmath11 1.55 & 1 & 5 & 0.02 & 0.007 & 12 & + gj 433 b & 11 35 26.9 & @xmath032 32 23 & 112.09 @xmath11 1.43 & 2 & 1,2 & 0.02 & 0.020 & 13 & a + gj 1148 b & 11 41 44.6 & @xmath3942 45 07 & 88.81 @xmath11 2.14 & 2 & 1,2 & 0.30 & 0.113 & 14 & + gj 436 b & 11 42 11.0 & @xmath3926 42 23 & 98.95 @xmath11 2.07 & 2 & 1,2 & 0.07 & 0.007 & 15 & + gj 581 b & 15 19 26.0 & @xmath007 43 20 & 159.28 @xmath11 1.32 & 3 & 1,2,3 & 0.05 & 0.015 & 16 & + gj 581 c & & & & & & 0.02 & 0.035 & 16 & + gj 581 d & & & & & & 0.02 & 0.183 & 16 & b + gj 581 e & & & & & & 0.01 & 0.009 & 16 & + lp 804 - 027 b & 16 12 41.7 & @xmath018 52 31 & 69.46 @xmath11 3.12 & 1 & 2 & 2.10 & 0.306 & 17 & + gj 649 b & 16 58 08.8 & @xmath3925 44 39 & 97.28 @xmath11 1.32 & 2 & 1,2 & 0.33 & 1.638 & 18 & a + gj 1214 b & 17 15 18.9 & @xmath3904 57 50 & 69.04 @xmath11 0.54 & 3 & 1,3,6 & 0.02 & 0.004 & 19 & c + gj 667c b & 17 18 57.1 & @xmath034 59 23 & 138.24 @xmath11 0.57 & 3 & 1,3,7 & 0.02 & 0.020 & 13 & a + gj 667c c & & & & & & 0.01 & 0.077 & 13 & + gj 674 b & 17 28 39.9 & @xmath046 53 42 & 220.11 @xmath11 1.39 & 2 & 1,2 & 0.03 & 0.013 & 20 & + gj 832 b & 21 33 33.9 & @xmath049 00 32 & 202.03 @xmath11 1.00 & 2 & 1,2 & 0.64 & 9.353 & 21 & + gj 849 b & 22 09 40.3 & @xmath004 38 26 & 113.46 @xmath11 1.32 & 3 & 1,2,3 & 0.90 & 5.241 & 22 & a + gj 876 b & 22 53 16.7 & @xmath014 15 49 & 214.45 @xmath11 0.57 & 4 & 1,2,3,8 & 1.95 & 0.167 & 23 & + gj 876 c & & & & & & 0.61 & 0.082 & 23 & + gj 876 d & & & & & & 0.02 & 0.005 & 23 & + gj 876 e & & & & & & 0.04 & 0.342 & 23 &
+ in the broader context , these results are consistent with recently published searches for jovian companions to m dwarfs at shorter orbital periods .
transit searches are unlikely to detect companions at jovian orbits , due to the narrow range of detectable inclinations .
@xcite found no jupiter - sized planets in their transit search , which is most sensitive to companions at orbital periods less than 10 days , and conclude that such planets rarely orbit m dwarfs .
based on the first two years of their astrometric search , @xcite find no planetary mass companions to the 20 m and l dwarfs they observed .
they determine the occurrence rate of planets more massive than @xmath105 @xmath1 to have an upper limit of 9% .
the longer time coverage of the astrometric results presented in this work overlap most closely with radial velocity results .
@xcite report detection limits based on radial velocity measurements for 102 nearby m dwarfs , of which six are featured in this paper . among the 102 m dwarfs searched ,
they confirm only two planets with orbital periods longer than 100 days . at an orbital period of 1000 days ( 2.7 years ) , they report a detection sensitivity of @xmath51 around 90% of stars observed , and @xmath52 at 10000 days ( 27 years ) .
@xcite report that 6.5% @xmath11 3.0% of m dwarfs host a 1 to 13 @xmath1 planet at a separation less than 20 au , based on their radial velocity survey and high resolution imaging .
the results presented here are among the first astrometric searches for jovian companions at jovian orbits , and fill in relatively unexplored mass and period parameter space models of how m dwarf planetary systems form and evolve must now explain the lack of massive companions with long period orbits like those in our solar system .
looking forward , the _ gaia _ mission may detect up to 2600 planets within 100 pc @xcite .
however , its ability to determine accurate masses and orbits will be limited to orbital periods less than 6 years , a fraction of the time coverage of our ground - based astrometric observations .
finding nearby m dwarfs with planets remains an important challenge , as the closest planets are the brightest and most easily studied , and m dwarfs dominate the stellar population . as part of its mission to characterize the solar neighborhood , recons is developing a database of all objects with accurate trigonometric parallaxes placing them within 25 pc ( @xmath53 mas with an error @xmath54 mas ) .
for an extrasolar planet to be included in the recons database it must orbit a star that meets the above criteria , or be a free - floating object with a comparable parallax , and be listed in _ both _ the extrasolar planets encyclopaedia ( _ exoplanet.eu _ ) and the exoplanet orbit database ( _ exoplanets.org _ , @xcite ) .
the recons database currently contains 1074 systems having m dwarf primaries ( @xmath55 @xmath56 ) within 25 pc , only 17 of which have detected exoplanets as of 2014 july 1 , listed in table [ tab4 ] .
the error - weighted mean parallax for each system is given in column 4 , including the parallaxes in this work and published values .
minimum masses and orbital periods for planets with references are listed in columns 7 9 .
note that only 3 of the 26 reported planets have masses greater than jupiter s .
based on the 36 m dwarf primaries within 5 pc @xcite , we anticipate that there are 4500 m dwarf primaries within 25 parsecs , yet only 17 so far have been found to host planets . clearly , many planets lurk undetected in the solar neighborhood
. discovering these planets will require a wide variety of survey techniques , and as sensitivities are improved , astrometry will continue to play an important role .
we wish to thank cassy davison for her assistance developing the simulations , and sergio dieterich for helpful conversations regarding brown dwarf luminosities , and his assistance in surveying the existing literature .
we also wish to thank hugh harris for his helpful suggestion regarding the eccentricities of unresolved binaries .
this effort has been supported by the national science foundation via grants ast 05 - 07711 and ast 09 - 08402 , and the long - term cooperative efforts of the national optical astronomy observatories and the members of the smarts consortium .
this research has made use of the exoplanet orbit database and the exoplanet data explorer at _
exoplanets.org_ , as well as the extrasolar planet encyclopaedia at _ exoplanet.eu_. this research has also made use of aladin and the simbad database , operated at cds , strasbourg , france .
anglada - escud , g. , boss , a. p. , weinberger , a. j. , et al .
2012 , apj , 746 , 37 anglada - escud , g. , rojas - ayala , b. , boss , a. p. , weinberger , a. j. , & lloyd , j. p. 2013
, , 551 , a48 anglada - escud , g. , arriagada , p. , tuomi , m. , et al . 2014 , arxiv:1406.0818 apps , k. , clubb , k. i. , fischer , d. a. , et al .
2010 , , 122 , 156 bailey , j. , butler , r. p. , tinney , c. g. , et al .
2009 , , 690 , 743 bartlett , j. l. , et al . in preparation benedict , g. f. , mcarthur , b. , chappell , d. w. , et al .
1999 , aj , 118 , 1086 benedict , g. f. , mcarthur , b. e. , franz , o. g. , et al .
2001 , aj , 121 , 1607 benedict , g. f. , mcarthur , b. e. , forveille , t. , et al .
2002 , apj , 581 , l115 berta , z. k. , irwin , j. , & charbonneau , d. 2013 , arxiv:1307.3178 bertin , e. , & arnouts , s. 1996 , a&as , 117 , 393 bonfils , x. , mayor , m. , delfosse , x. , et al .
2007 , , 474 , 293 bonfils , x. , gillon , m. , forveille , t. , et al .
2011 , a&a , 528 , a111 bonfils , x. , delfosse , x. , udry , s. , et al .
2013 , a&a , 549 , a109 bonfils , x. , lo curto , g. , correia , a. c. m. , et al .
2013 , , 556 , a110 burt , j. , vogt , s. s. , butler , r. p. , et al .
2014 , , 789 , 114 delfosse , x. , bonfils , x. , forveille , t. , et al .
2013 , a&a , 553 , a8 dieterich , s. b. , henry , t. j. , jao , w .- c . , et al . 2013 ,
arxiv:1312.1736 dressing , c. d. & charbonneau , d. 2013 , apj , 767 , 95 endl , m. & krster , m. 2008 , , 488 , 1149 fabricius , c. , & makarov , v. v. 2000 , aa&s , 144 , 45 forveille , t. , bonfils , x. , delfosse , x. , et al .
2009 , a&a , 493 , 645 franz , o. g. , wasserman , l. h. , benedict , g. f. , et al .
1999 , working on the fringe : optical and ir interferometry from ground and space , 194 , 28 frescura , f. a. m. , engelbrecht , c. a. , & frank , b. s. 2007 , arxiv : 0706.2225 graham , j. a. 1982 , pasp , 94 , 244 haghighipour , n. , vogt , s. s. , butler , r. p. , et al .
2010 , apj , 715 , 271 harpse , k. b. w. , hardis , s. , hinse , t. c. , et al .
2013 , , 549 , a10 hartkopf , w. i. , mcalister , h. a. , & franz , o. g. 1989 , , 98 , 1014 hawley , s. l. , gizis , j. e. , & reid , i. n. 1996 , aj , 112 , 2799 henry , t. j. , & mccarthy , d. w. , jr .
1993 , aj , 106 , 773 henry , t. j. , ianna , p. a. , kirkpatrick , j. d. , & jahrei , h. 1997 , aj , 114 , 388 henry , t. j. , franz , o. g. , wasserman , l. h. , et al . 1999 ,
apj , 512 , 864 henry , t. j. , subasavage , j. p. , brown , m. a. , beaulieu , t. d. , jao , w. , & hambly , n. c. 2004 , aj , 128 , 2460 henry , t. j. , jao , w .- c . , subasavage , j. p. , beaulieu , t. d. , ianna , p. a. , costa , e. , & mndez , r. a. 2006 , aj , 132 , 2360 henry , t.j .
2013 , in royal astronomical society of canada observer s handbook 2013 , ed .
david m.f .
chapman .
toronto : royal astronomical society of canada howard , a. w. , johnson , j. a. , marcy , g. w. , et al .
2010 , apj , 721 , 1467 jao , w .- c .
, henry , t. j. , subasavage , j. p. , bean , j. l. , costa , e. , ianna , p. a. , & mndez , r. a. 2003 , aj , 125 , 332 jao , w .- c . , henry , t. j. , subasavage , j. p. , brown , m. a. , ianna , p. a. , bartlett , j. l. , costa , e. , & mndez , r. a. 2005 , aj , 129 , 1954 jao , w .- c . ,
henry , t. j. , subasavage , j. p. , et al .
2011 , , 141 , 117 johnson , j. a. , howard , a. w. , marcy , g. w. , et al .
2010 , , 122 , 149 landolt , a. u. 1992 , aj , 104 , 340 landolt , a. u. 2007 , , 133 , 2502 maness , h. l. , marcy , g. w. , ford , e. b. , et al .
2007 , , 119 , 90 mayor , m. , bonfils , x. , forveille , t. , et al .
2009 , a&a , 507 , 487 montet , b. t. , crepp , j. r. , johnson , j. a. , howard , a. w. , & marcy , g. w. 2013 , arxiv:1307.5849 reid , i. n. , hawley , s. l. , & gizis , j. e. 1995 , aj , 110 , 1838 riedel , a. r. , subasavage , j. p. , finch , c. t. , et al .
2010 , , 140 , 897 rivera , e. j. , laughlin , g. , butler , r. p. , et al .
2010 , , 719 , 890 robertson , p. , mahadevan , s. , endl , m. , & roy , a. 2014 , arxiv:1407.1049 sahlmann , j. , lazorenko , p. f. , segransan , d. , et al .
2014 , arxiv:1403.1275 skrutskie , m. f. , cutri , r. m. , stiening , r. , et al .
2006 , , 131 , 1163 sozzetti , a. , giacobbe , p. , lattanzi , m. g. , et al . 2013 , arxiv:1310.1405 subasavage , j. p. , jao , w .- c . , henry , t. j. , bergeron , p. , dufour , p. , ianna , p. a. , costa , e. , & mndez , r. a. 2009 , aj , 137 , 4547 van altena , w. f. , lee , j. t. , & hoffleit , d. 1995 , in the general catalogue of trigonometric stellar parallaxes ( 4th ed .
; new haven , ct : yale univ .
obs . ) van de kamp , p. , 1967 ,
principles of astrometry ( san francisco : w. h. freeman and company ) van leeuwen , f. 2007 , a&a , 474 , 653 winters , j. g.,henry , t. j. , jao , w .- c . , subasavage , j. p. , finch , c. t. , & hambly , n. c. 2011 , aj , 141 , 21 wright , j. t. , fakhouri , o. , marcy , g. w. , et al .
2011 , pasp , 123 , 412 zechmeister , m. , krster , m. 2009 , a&a , 496 , 577 | astrometric measurements are presented for seven nearby stars with previously detected planets : six m dwarfs ( gj 317 , gj 667c , gj 581 , gj 849 , gj 876 , and gj 1214 ) and one k dwarf ( bd @xmath010 3166 ) .
measurements are also presented for six additional nearby m dwarfs without known planets , but which are more favorable to astrometric detections of low mass companions , as well as three binary systems for which we provide astrometric orbit solutions .
observations have baselines of three to thirteen years , and were made as part of the recons long - term astrometry and photometry program at the ctio / smarts 0.9 m telescope .
we provide trigonometric parallaxes and proper motions for all 16 systems , and perform an extensive analysis of the astrometric residuals to determine the minimum detectable companion mass for the 12 m dwarfs not having close stellar secondaries . for the six m dwarfs with known planets , we are not sensitive to planets , but can rule out the presence of all but the least massive brown dwarfs at periods of 2 12 years .
for the six more astrometrically favorable m dwarfs , we conclude that none have brown dwarf companions , and are sensitive to companions with masses as low as 1 @xmath1 for periods longer than two years .
in particular , we conclude that proxima centauri has no jovian companions at orbital periods of 2 12 years .
these results complement previously published m dwarf planet occurrence rates by providing astrometrically determined upper mass limits on potential super - jupiter companions at orbits of two years and longer . as part of a continuing survey ,
these results are consistent with the paucity of super - jupiter and brown dwarf companions we find among the over 250 red dwarfs within 25 pc observed longer than five years in our astrometric program . |
observations of the anisotropy of the cosmic microwave background ( cmb ) radiation have given us unprecedented knowledge of the very early universe and dramatically confirmed the picture of large - scale structure as arising from the gravitational amplification of small perturbations in a universe with a significant cold dark matter component @xcite .
in this model the ionization history of the universe has two main events , a ` recombination ' at @xmath0 in which it went from ionized to neutral and a ` reionization ' during @xmath1 in which the radiation from early generations of collapsed objects was able to ionize the intergalactic medium .
the former event is strongly constrained by the cmb . a new generation of instruments will soon allow us to probe this second event : `` the end of the dark ages '' ( for reviews of reionization see e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . since at reionization a very small fraction of the mass affected each and every baryon in the universe
, reionization is particularly sensitive to the distribution and behavior of collapsed structure .
we expect that the ionizing sources are situated in large ( @xmath2k or @xmath3 ) dark matter halos where the gas can cool efficiently to form stars .
models for the sources of reionization thus often start with estimates of the number and properties of virialized dark matter halos at high redshift , the focus of this paper . at @xmath4 ,
halos with @xmath5 are expected to be biased similarly to very massive clusters ( @xmath6 ) today , with the most massive and recently formed halos growing rapidly and merging frequently .
we explore some properties of these collapsed halos at a high redshift using a suite of high resolution , collisionless , n - body simulations .
we pay particular attention to merger rates and mass accretion histories with an eye to applications for reionization .
we also compare the n - body results with the predictions of the oft - used @xcite formalism . if halo mergers are accompanied by a temporary increase in photon production ( due either to starbursts or increased black hole accretion e.g. @xcite )
we expect reionization to be influenced by the merger and accretion history of dark matter halos , beyond just the fact that more massive halos emit more photons . with a simple model of star formation
we show that merger - induced scatter in photon production may be significant , with the production rates acquiring a substantial tail to large photon production rates .
since the massive halos are relatively rare , this individual halo scatter is expected to translate into a scatter of photon production rates inside ionized regions , changing the bubble distribution .
the outline of the paper is as follows . in
[ sec : sim ] we describe the n - body simulations .
the basic halo properties are described in [ sec : halo ] along with the results for mergers and mass gains and the comparison to press - schechter .
the consequences of this merging in a simple model for photon production are elucidated in
[ sec : reion ] and we summarize and conclude in [ sec : conclusions ] .
we base our conclusions on 5 dark matter only n - body simulations of a @xmath7cdm cosmology with @xmath8 , @xmath9 , @xmath10 , @xmath11 and @xmath12 , in agreement with a wide array of observations .
the initial conditions were generated at @xmath13 using the zeldovich approximation applied to a regular , cartesian grid of particles .
our two highest resolution simulations employed @xmath14 equal mass particles ( @xmath15 and @xmath16 ) in boxes of side @xmath17 and @xmath18mpc with plummer equivalent smoothings of @xmath19 and @xmath20kpc .
they were evolved to @xmath4 using the _ treepm _ code described in @xcite ( for a comparison with other codes see * ? ? ?
we ran 3 additional , smaller simulations in a @xmath21mpc box , one with @xmath22 particles and two with @xmath23 particles ( each started at @xmath24 ) .
a comparison of the boxes allows us to check for finite volume , finite mass and finite force resolution effects .
we shall comment on each where appropriate .
the phase space data for the particles were dumped at @xmath25 outputs spaced by @xmath26myr from @xmath27 to @xmath4 for all but the largest box .
the lower resolution of the largest box makes it less useful for merger trees , so it was sampled for only subset of these output times , ending at @xmath4 . for each output
we generate a catalog of halos using the friends - of - friends ( fof ) algorithm @xcite with a linking length , @xmath28 , of @xmath29 times the mean inter - particle spacing .
this partitions the particles into equivalence classes , by linking together all particle pairs separated by less than @xmath28 .
the halos correspond roughly to particles with @xmath30 times the background density .
we also made catalogs using a linking length of @xmath31 times the mean inter - particle spacing , which we shall discuss further below .
we found that the fof algorithm with a larger linking length had a tendency to link together halos which we would , by eye , have characterized as separate ( see also * ? ? ?
* for similar discussion ) .
this problem is mitigated with our more conservative choice of @xmath28 .
for each halo we compute a number of properties , including the potential well depth , peak circular velocity , the position of the most bound particle ( which we take to define the halo center ) and @xmath32 , the mass interior to a radius , @xmath33 , within which the mean density is @xmath34 times the background density .
our @xmath32 masses should thus be comparable to the sum of the particles in an @xmath35 group a common definition that employs both the @xmath35 halo finder and definition of mass . ] . as discussed in @xcite and @xcite ,
the choice of halo mass is problematic and ultimately one of convention .
we shall return to this issue in the next section .
merger trees are computed from the set of halo catalogs by identifying for each halo a `` child '' at a later time .
the child is defined as that halo which contains , at the later time step , more than half of the particles in the parent halo at the earlier time step ( weighting each particle equally ) .
for the purposes of tracking halos this simple linkage between outputs suffices ( note that we do not attempt to track subhalos within larger halos , which generally requires greater sophistication ) .
two examples of the halo merger trees are given in fig .
[ fig : treeplot ] , where we see a rich set of behaviours , including major and minor mergers and many body mergers . from the merger trees it is straightforward to compute the time when a halo ` falls in ' to a larger halo , the number and masses of the progenitors etc . due to finite computational resources
, all n - body simulations must trade - off computational volume for mass resolution . by running multiple simulations we can overcome this to some extent , but not entirely .
we have chosen to slightly under - resolve the low mass ( @xmath36k ) halos in order to simulate a slightly larger volume , since our focus will be on the more massive halos which have more frequent major mergers . under reasonable assumptions ( see below ) between @xmath37 off all photon production occurs in halos more massive than @xmath38 at @xmath4 , and we easily resolve these objects with the @xmath39mpc simulation which we use for the bulk of the paper .
the highest mass objects in our volume have mass @xmath40 and radii of several tens of kpc . at @xmath4
these halos are analogous to rich clusters today , being recently formed and rare : fig .
[ fig : comparenu ] shows the mass as a function of peak height , @xmath41 , at @xmath4 , 7 , 4 and 0 . the threshold @xmath42 is defined as @xmath43 , where @xmath44 is the linear growth factor normalized to unity at @xmath45 and @xmath46 is the variance of the mass computed using linear theory at @xmath45 . in our cosmology @xmath47 . due to the flatness of the dimensionless power on the scales of interest , the slightly red initial spectrum and the low clustering amplitude , the characteristic mass , @xmath48 , where @xmath49 , is @xmath50 at @xmath4 , so all of the halos we consider are @xmath51 .
one of the most basic and useful quantities we can derive from the simulations is the mass function , the spatial abundance of halos as a function of mass .
high redshift mass functions have been studied by many groups ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and @xcite offer a comprehensive summary of recent work .
most previous work finds mass functions which are better fit by the @xcite , @xcite or @xcite form .
we find that the appropriate mass function to use depends primarily on the definition of mass chosen and definitions which at @xmath52 give very similar mass functions can give quite different ones at @xmath4 .
we show the mass function(s ) from our three highest resolution simulations in fig .
[ fig : nofm ] .
if we use as our mass estimator the sum of the particle masses in the fof(0.2 ) groups ( open symbols ) then we find good agreement with the @xcite or @xcite forms .
this is the procedure followed by most of the groups above to 22% over the mass range plotted .
see , e.g. @xcite for further discussion . ] . however if we choose instead to use @xmath32 as our mass estimator ( filled symbols ) we find a different mass function .
although this mass function shows a marked excess of high mass halos compared to the @xcite form , it is a better fit than the alternate forms mentioned above .
agreeably , for the scales plotted , the @xmath32 mass function is independent of the initial fof group catalog used to define the centers about which @xmath32 is determined .
this is not too surprising as the group centers hardly change and the number of `` small '' groups which split off of larger fof groups as the linking length is decreased is tiny compared to the number of low - mass `` field '' halos .
the differences in mass functions then comes primarily from the definition of the masses of the found objects . comparing halo by halo the fof(0.2 )
masses are almost twice @xmath32 , though the difference depends on mass .
a similar difference was also noted by @xcite as a shift to lower abundance at fixed mass when comparing an fof(0.2)-based mass function to that of a different halo finder .
we believe the primary issue is not the halo finder , but the mass definition .
their second halo finder assigns masses which are essentially our @xmath32 .
the mass discrepancy is much larger for these halos at @xmath4 than it is for group and cluster - sized halos at the present day ( e.g. figure 11 in * ? ? ?
the mass differences are quite interesting .
the historical argument for choosing fof(0.2 ) was that the fof group finder selects particles approximately within a density @xmath53 times the mean density . for a singular isothermal sphere profile ( @xmath54 ) and a critical density universe
the mean enclosed density is thus @xmath55 , in accord with arguments based on spherical top - hat collapse ( e.g. * ? ? ?
* ) . at @xmath4
the universe is close to critical density so we might expect the fof(0.2 ) and @xmath32 mass functions to agree better than at lower @xmath56 where @xmath57 .
however , we are focusing on very high mass halos which have only recently formed at @xmath4 .
they are therefore less centrally concentrated for halos of the form proposed by @xcite .
] than a ` typical ' halo .
this can also be seen in fig .
[ fig : profile ] , where halo profiles are compared to the isothermal sphere profile . as the halo profiles are less steep than the isothermal sphere form assumed in the argument above , this leads to the differences in mass between the fof(0.2 ) and @xmath35 definitions .
by contrast , we find that the fof(0.168 ) mass function is very similar to the @xmath32 points plotted , and a halo by halo comparison shows that the two masses agree to within 20 - 30 per cent .
as we go down the mass function , to more concentrated , lower mass halos , we expect fof(0.2 ) to better match @xmath32 ( e.g. * ? ? ?
. in general , given the strong dependence of the mass function upon the mass definition , and the ambiguity in this quantity in many analytic treatments , significant care must be taken when making predictions for the abundance of halos . even if we decide to treat all halos as a simple 1-parameter family , it is likely preferable to make comparisons with some quantity more directly related to observables ( such as circular velocity , halo virial temperature or potential well depth ) or to discuss statistics as a function of number density rather than mass .
like rich clusters we expect that these massive halos , in the process of formation , will not lie on the usual ` vacuum ' virial relation 2ke@xmath58pe , where ke and pe refer to the potential and kinetic energy respectively .
in fact we find that 2ke / pe@xmath59 for halos in the range @xmath60 , very similar to the value found for rich clusters today @xcite .
a similar ` excess ' kinetic energy was also found by @xcite for lower mass halos . the ratio is larger than unity because of the steady accretion of material onto the cluster @xcite .
[ fig : bias ] shows the clustering of the dark matter and the halos from our @xmath18mpc run .
we plot the auto - correlation function of the dark matter and the cross - correlation of the halo centers with the dark matter respectively .
the latter is both less subject to noise , and jackknife errors on the cross - correlation , @xmath61 , are a few percent for the samples shown .
jackknife drastically underestimates the errors from finite volume however . ] from our small sample of massive halos and more applicable to understanding how radiation from the halos would influence the surrounding mass .
the ratio of the cross- to auto - correlation functions defines the scale dependent bias , @xmath62 .
the mass auto - correlation function is in good agreement between the @xmath17 and @xmath18mpc boxes up to @xmath63mpc , with @xmath64 from the @xmath39mpc box falling below that of the @xmath18mpc box beyond this scale .
the @xmath21mpc box has noticeably less power over a wide range of scales . for the masses where we can compare and for the range of linear scales plotted , the halo - mass cross - correlation functions of the @xmath17 and @xmath18mpc boxes are in excellent agreement
, so we have shown the results only for the @xmath18mpc box
. our halo samples are mass thresholded , however by using number density as our marker we largely avoid the issues of mass definition discussed earlier .
the differences in bias at fixed @xmath65 for the different mass choices , arising from the scatter between different mass definitions , is only a few percent .
taking @xmath66 as the asymptotic value , the large - scale bias is in good agreement with the models of @xcite and @xmath67 higher than that of @xcite .
those of @xcite and @xcite lie in between .
( the model of @xcite only extends up to 100 times the non - linear mass , where @xmath68 , and it not applicable to our results . ) to make contact with the earlier literature we plot in fig .
[ fig : b_vs_nu ] the bias as a function of peak height , @xmath69 , obtained from @xmath65 using the @xcite mass function . when computing @xmath70 in the simulation we rank order the halos by fof(0.2 ) mass in order to best match the chosen mass function
. this mass function is then used when analytically computing the halo - weighted bias @xmath70 from each of the analytic forms which provide @xmath71 . because of this the @xcite bias function is the only one which would give an average bias of unity when integrated over @xmath69 .
similar trends for rare halos to have larger bias than the modern fits predict have been seen at lower redshift ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for recent work ) but we must also remember that @xmath66 is likely higher than @xmath72 so the degree of overshoot is hard to quantify precisely .
as expected , the clustering strength is an increasing function of mass @xcite , or a decreasing function of halo abundance .
we now consider the hierarchical assembly of the dark matter halos through merging and accretion .
we shall use the @xmath14 , @xmath39mpc simulation since it provides both high mass resolution and a representative volume . since
our progenitor relationships are based on particles in the fof groups , we use the fof(0.168 ) masses for consistency .
as discussed earlier , for our massive halos these masses are within @xmath73 per cent of @xmath32 and none of our conclusions depend sensitively on this choice .
[ fig : mrates ] shows the fraction of halos with @xmath74 which have experienced at least one major merger as function of lookback time , in intervals of @xmath26myrs .
we show three different definitions of ` major ' merger , where the largest two progenitors of the halo have ratios below 1:10 , 1:5 or 1:3 .
mergers are frequent but not ubiquitous not all halos have had a major merger within @xmath75myrs , but many have .
the fraction decreases for smaller mass ratios and for lower mass halos , as expected .
we can also consider mass gains between time steps , often denoted in the literature as @xmath76 where @xmath77 is the mass of the largest progenitor at the earlier time and @xmath78 is the mass of the halo under consideration .
mass gains are sometimes used as a proxy for mergers .
[ fig : mrates ] shows those halos whose mass increased by at least 10 , 20 or 33 per cent as a function of lookback time .
the top and bottom panels of fig .
[ fig : mrates ] would be identical if all mergers were two body and there was no smooth accretion . as can be seen in fig .
[ fig : treeplot ] this is not the case ; fig .
[ fig : mrates ] quantifies this difference for major mergers .
the press - schechter model predicts the evolution of the mass function , and it can be extended to make predictions for the time history of halos .
this `` excursion set formalism '' is often called extended press - schechter @xcite and denoted eps
see @xcite for a recent review .
although it is analytically tractable , it has many inconsistencies and does not compare particularly well to n - body simulations ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
for example , @xcite found that with eps halos of mass @xmath79 at @xmath52 formed later than in n - body simulations ( but see * ? ? ?
* for a slightly different quantity ) . in fig .
[ fig : mah ] we compare the n - body mass accretion histories for massive halos at @xmath4 to a model by @xcite based on eps which predicts almost exponential growth with redshift .
( other analytic models also exist , see e.g. @xcite for a summary and comparison , there are some discrepancies between these which are not yet fully understood . )
we find that eps predicts mass growth which is too rapid also for the high mass , high redshift regime studied here .
the n - body mass accretion histories _ are _ relatively well fit by an exponential in @xmath56 a growth model also proposed by @xcite on the basis of n - body simulations of galaxy - sized halos at low @xmath56 but the coefficient predicted by @xcite is larger than measured in the simulations . perhaps the most common use of eps is to predict merger rates , and eps has been used in this context in several recent models of reionization . to compare the eps predictions with our simulations we computed merger rates using only our last ( @xmath26myr ) time step , taking for any halo with @xmath4 mass within @xmath80 to @xmath81 the distribution of progenitors , @xmath82 .
the eps prediction can be found in the appendix .
we show a representative example of @xmath83 for @xmath80 in the range @xmath84 in fig .
[ fig : epscomp ] . for most of the range
the agreement is reasonably good . at the low mass end
eps significantly underpredicts the number of predecessors found in our simulations ( see also * ? ? ?
at the high mass end the eps rate starts to climb rapidly , eventually diverging unphysically .
these trends are independent of the final mass chosen , or the definition of mass used . the eps formula as progenitor mass goes to zero also diverges , which we could not approach due to our finite mass resolution , but the mass weighted eps calculation is finite at both ends .
there is another notable difference between eps and our simulation .
though it is relatively small , our time step is still too large for all mergers to be truly 2-body ( see fig . [
fig : treeplot ] ) , as implicitly assumed by eps .
a large fraction ( 20 - 50 per cent , depending on @xmath80 ) of the halos are actually produced in 3 ( or more)-body mergers .
finally we also looked for evidence that recently merged halos clustered differently than randomly chosen halos of the same mass .
the correlation function of 1:2 or 1:3 mergers appeared to be slightly ( @xmath85 ) enhanced at @xmath86mpc compared to the random sample , but the number of merged halos was too small for this to be statistically meaningful .
the effect thus appears to be modest , if present at all , just as was found for lower redshift , high - mass halos ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this suggests that the clustering of massive halos does not depend strongly upon their recent merger history .
this in turn significantly eases the modeling of merger - related processes , such as enhanced photon production during reionization which we now discuss .
the rate of photon production in a galaxy can be enhanced by mergers , which can trigger starbursts or possibly accretion onto a black hole which may be present ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
it is reasonable to anticipate that the mergers of large dark matter halos could have similar effects on the photon production rate of the sources within them .
we will make this assumption , and then consider the consequences of the merger rates computed above for the photon production distribution .
we frame our discussion in terms of a simple but promising model for reionization proposed by @xcite , though our result is true more generally . in these models , a halo of a given mass @xmath87 ( in units of some reference mass )
is considered a source of photons with rate @xmath88 usually @xmath89 is taken to be mass independent , scale as @xmath90 or transition from @xmath90 to @xmath91 at @xmath92 ( @xcite , motivated by @xcite ) .
a region around these halos is taken to be ionized if the photons within it are sufficient to ionize all the interior mass .
some extensions also give recombinations spatial and/or temporal dependence and incorporate this into finding the bubble properties @xcite , or incorporate eq .
( [ eqn : dndt ] ) into n - body simulations @xcite . under these assumptions
the morphology of ionized regions can be computed from the photon production rate and spatial distribution of dark matter halos . a first step at including halo mergers within the above formalism ( and its generalizations )
was presented in @xcite .
those calculations were based on the press - schechter formalism , and so could only provide average numbers of mergers for halos in a given mass range ; scatter was computed by assuming that the mergers had a poisson distribution . with our simulations we are able to check these assumptions and significantly extend this work because we have access to the detailed merger history of each halo .
this allows us to go beyond their analytic estimates to explicitly calculate the full distribution of photon production for a halo of mass @xmath87 , taking into account the distribution of histories and their associated ( and different ) photon production rates for a fixed @xmath87 . from the merger tree for each halo at @xmath4 ( @xmath93 )
we identify which progenitors had at least one major merger ( greater than 1:3 or 1:10 ) , and the time @xmath94 they occurred .
we include all of the mergers in the tree and we place @xmath94 at random within the @xmath26myr interval between the relevant outputs .
each of these mergers is allowed to contribute `` excess '' photons beyond those which would automatically be assigned to the halo on the basis of its @xmath4 mass , @xmath95 , but the number of photons contributed is exponentially attenuated with an @xmath96-folding time @xmath97 .
the `` excess '' photon production is thus proportional to @xmath98 where the sum is over all halos which have undergone a major merger and we take @xmath99 or @xmath100 .
the exponential decay is motivated by modeling of starbursts , e.g. @xcite , hence the subscript @xmath101 .
we also consider another variant , including all halos with major mergers within @xmath97 of @xmath102 , with no attenuation : @xmath103 where @xmath104 if @xmath105 , @xmath106 if @xmath107 and zero otherwise .
we denote this by a subscript @xmath108 , to indicate photon production by black holes , which might have their photon production rate increase over time and then decay once the fuel is exhausted . assuming a step - like function is a crude first approximation to this uncertain physics . in all cases
we take the quiescent photon production to depend on the @xmath4 halo mass with the same index , @xmath109 , as @xmath110 .
we note this prescription might cause some overcounting if many mergers occur within a short time period and the gas becomes depleted from the earliest ones .
a more refined model would account for the evolving baryon budget within the halo , but our treatment is sufficient for the purpose of illustration .
the relative amplitudes of these two modes of photon production depend on a number of different factors ( see e.g. @xcite for discussion and summary of estimates at these redshifts ) but a factor @xmath111 is not unreasonable for starbursts and could be even larger for black holes .
the total photon production is thus enhanced by a factor @xmath112 for the `` starburst '' prescription , or its analogue @xmath113 for the `` black hole '' prescription . in principle both can contribute .
we considered the two effects separately , their combination is straightforward .
figure [ fig : enhance ] shows a typical example of the cumulative distribution of enhancement factors , eq .
( [ eqn : enhance ] ) .
we took the starburst form , 1:3 mergers , @xmath114 , @xmath115myr and @xmath116 , but other cases are very similar .
the enhancement distribution is extended , with a long tail to high @xmath117 and a peak at those halos which have not merged .
about half of the halos have twice the photon production , while 20 per cent have no enhancement .
choosing a larger @xmath118 increases the size of the enhancement , but does not qualitatively change the form of the distribution .
similarly , changing @xmath109 or @xmath97 changes the detailed form of the distribution but not its character .
halos down to @xmath119 show a very similar distribution of enhancements .
by contrast the model for black hole accretion produces a bimodal distribution , as the `` early '' mergers contribute relatively more than in the case of the starbursts , leading to a second peak .
even though the scatter in photon contributions from halo to halo is large for a given mass , if a large number of such halos are found in a bubble , their contributions to the photon numbers will tend to the mean , allowing the distribution to be replaced by the average . precisely counting
the number of halos of a given mass and the combined photon scatter inside a typical bubble is unfortunately self - referential : changing the ionization properties ( including mergers ) changes the bubble sizes and thus the number of halos within .
different assumptions about the nature of the sources and their feedback can give drastically different bubble sizes , and the relative importance of high vs. lower mass halos ( e.g. * ? ? ?
* ; * ? ? ?
given these uncertainties we consider properties in an average volume , for illustration .
for quiescent photon production and @xmath120 , analytic estimates such as press - schechter give that halos with @xmath121 contribute between @xmath37 of all photons .
even choosing @xmath122 , such halos contribute @xmath12310 per cent of the photons .
the number density of such halos is @xmath124 .
bubble radii in different models range over several orders of magnitude .
a middle - of - the - road estimate is @xmath125mpc , which would contain about 3 halos with @xmath121 .
the bubble radius would also be larger than the correlation length of our halos , so clustering is only expected to change this number by a factor of order unity .
a small number of halos contributing a large fraction of the photons means that scatter in their photon production should affect the properties of the bubbles .
our calculation is relatively crude , but it suggests that the inclusion of mergers into a more refined model of reionization could alter the distribution of ionized regions . for models based on approximate dynamics ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , a possible first step would be to assign a merger history to the sources at random .
this is accurate to the extent that recently merged halos are not spatially biased with respect to a random sample of halos of the same mass . for models which marry the analytic model to dark matter simulations
the merger history is known , so only the photon production rate needs to be modified .
more complex simulations involving radiative transfer will need to follow the photon production history as the halos evolve , perhaps using a semi - analytic model ( such as in * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
a full - blown simulation including radiative transfer and n - body in a large enough volume is still out of reach ( but see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for recent progress ) .
using 5 n - body simulations with different sized boxes and particle loads we considered the abundance , clustering and assembly histories of high mass halos at high redshift .
we present results specifically for @xmath4 , but the evolution of the populations is smooth and the results will be similar at slightly higher and lower redshift . like the halos of rich groups or clusters today the halos we consider are in the process of forming , growing rapidly through accretions and mergers .
we found that they had larger velocity dispersions than a naive application of the virial theorem would predict , due to a surface pressure from infalling material . being recently formed , the halos were not very centrally concentrated , leading to a factor of two difference between fof(0.2 ) masses and @xmath32 .
when measured against @xmath32 we found our halo abundances were closer to the press - schechter fitting formula than that of @xcite , though the simulations had more high mass halos than the analytic form .
if fof(0.2 ) masses were used instead , the mass function approached that of @xcite , in agreement with earlier work .
this discrepancy indicates that analytic models which assign an observable to halos of a given size need to pay particular attention to the marker of halo size employed .
the high mass halos were significantly clustered , and we calculated the halo bias by taking the halo - mass cross correlation and dividing by the matter auto - correlation function .
our rare halos were more clustered than the recent models of @xcite and closer to the models of @xcite .
merging is common , though not ubiquitous , in high mass halos at @xmath4 .
major mergers , with progenitor mass ratios less than 1:3 , occurred within @xmath75myr of @xmath4 for more than half of halos with @xmath121 .
we looked at the fraction of halos undergoing mergers for a variety of lookback times and progenitor ratios , finding more mergers for more massive halos , longer lookback times or less extreme merger events .
mass gains , parameterized by @xmath76 , showed similar trends even though not all merger events were two body within our @xmath26myr time step .
the eps model provides a reasonable description of the progenitor mass distribution , though it underpredicts the number of low mass progenitors and diverges as @xmath126 .
the mass accretions histories predicted by eps , as calculated by @xcite , provide only a qualitative guide to the mean mass accretion histories seen in our simulations . at @xmath4 reionization
is expected to be underway due to photon production from astrophysical objects which formed in collapsed halos . within the context of a simple model which associates mergers with an increase in photon production rate the photon production distribution developed a high rate tail due to recently merged halos . including these photon production enhancements
will likely drive scatter in photon production at fixed halo mass . since the number of massive halos within a typical ionized bubble can be small , this scatter in photon production could well translate into additional scatter in bubble sizes and it would be very interesting to include this effect in approximate models of reionization . if the recently merged halos are not spatially biased with respect to other halos of the same mass , including these effects in models , even those without merger trees , should be straightforward .
jdc would like to thank d. holz and s. koushiappas for conversations .
we thank o. zahn for helpful comments on an early draft , and s. furlanetto , c.p .
ma , l. miller , e. neistein and j. zhang for comments on the final draft .
the simulations were analyzed on the supercomputers at the national energy research scientific computing center .
mw was supported by nasa .
we thank the referee for encouraging us to run the @xmath18mpc box .
we compared the progenitor mass distribution from the simulations with the predictions of the extended press - schechter formalism . to compute the distribution in this formalism
we need the number of progenitors with @xmath127 at time @xmath128 for a given range of final masses , @xmath129 , at time @xmath130 .
the eps theory gives @xmath131 where the number of halos of mass @xmath132 , @xmath133 , is given by the usual mass function @xcite : @xmath134\ ] ] with @xmath135 the background density . to get eq .
( [ eqn : eps ] ) the number of halos at a given time is multiplied by the fraction of halos that have mass @xmath132 at @xmath130 and have jumped from mass @xmath136 within @xmath137 : @xmath138
|\frac{d\sigma^2(m_i)}{d m_i}| dm_i \delta t \ ; . \end{array } \label{eq : pdot}\ ] ] where @xmath139 . to produce the eps predictions discussed in
[ sec : merge ] , we take @xmath140myr , assuming that this time is small . | we use a suite of high - resolution n - body simulations to study the properties , abundance and clustering of high mass halos at high redshift , including their mass assembly histories and mergers .
we find that the analytic form which best fits the abundance of halos depends sensitively on the assumed definition of halo mass , with common definitions of halo mass differing by a factor of two for these low concentration , massive halos . a significant number of massive halos are undergoing rapid mass accretion , with major merger activity being common .
we compare the mergers and mass accretion histories to the extended press - schechter formalism .
we consider how major merger induced star formation or black hole accretion may change the distribution of photon production from collapsed halos , and hence reionization , using some simplified examples . in all of these ,
the photon distribution for a halo of a given mass acquires a large scatter .
if rare , high mass halos contribute significantly to the photon production rates , the scatter in photon production rate can translate into additional scatter in the sizes of ionized bubbles .
-1 cm |
polarimetric imagery consists in forming an image of the state of polarization of the light backscattered by a scene .
we consider in this paper that the scene is artificially illuminated with coherent light ( laser ) .
for example , this illumination is used in active imagery in order to combine night vision capability and to improve image resolution for a given aperture size . in practice , using a coherent illumination produces speckle noise that deteriorate the image @xcite .
however , the backscattered light gives information about the capability of the scene to polarize or depolarize the emitted light and thus allows one to determine the medium that compose the scene .
these information can be described by a scalar parameter : the degree of polarization of light .
this quantity is obtained in standard configurations of polarimetric systems using four pair of angular rotations of both a compensator and a polarizer .
four transmittance are thus recorded @xcite that lead to the estimation of the degree of polarization . however , this system is complex and it is interesting to develop methods to estimate the degree of polarization that could reduce the number of images to register . in @xcite ,
the authors proposed to estimate the degree of polarization with only two intensity images , however this method relies on the assumption that the measurements of the two components are uncorrelated which can be in some cases a too restrictive hypothesis .
this paper extends the work of @xcite by taking into account the correlation of the different components .
+ let us first introduce the context of the study .
the electric field of the light at a point of coordinates * r * ( vector of 3 component ) in a 3d space and at time @xmath0 can be written , if we assume the light to propagate in a homogeneous and isotropic medium , as @xmath1 .
e^{- i 2 \pi \nu t}\ ] ] where @xmath2 is the central frequency of the field and @xmath3 are unitary orthogonal vectors ( in the following bold letters represent vectors ) .
+ the terms @xmath4 and @xmath5 are complex and define the random vector called jones vector @xmath6.\ ] ] the state of polarization of light corresponds to the properties of @xmath7 at a particular point of space .
it can be described by the covariance matrix @xmath8 @xmath9\ ] ] where @xmath10 and @xmath11 define respectively the statistical average and the complex conjugate . for sake of brevity ,
the following notations for @xmath8 are introduced @xmath12 = \frac{1}{c_1c_4-|c_2|^2 } \left[\begin{array}{ll } c_4 & -c_2 \\ -c_2^ * & c_1 \\
\end{array}\right];\ ] ] @xmath13\ ] ] let us note that this matrix can be diagonalized since it is hermitic .
+ in the case of coherent light , the electric field is represented by the complex jones vector @xmath14 which follows a gaussian circular law @xcite @xmath15 where @xmath16 stands for the adjoint of the vector @xmath17 .
+ the degree of polarization is defined by @xcite @xmath18 where @xmath19 and @xmath20 are the eigenvalues of @xmath8 ( @xmath21 ) .
+ the degree of polarization is a scalar parameter that characterizes the state of polarization of the light : if @xmath22 , the light is said to be totally depolarized , and if @xmath23 , the light is said to be totally polarized . in the intermediate cases ,
the light is partially polarized .
+ with the notations introduced in ( [ cov ] ) one can show that @xcite @xmath24 the knowledge of this quantity allows one to study the way the illuminated scene polarized or depolarized the emitted light .
this degree gives information about the nature of the medium in the scene . in the standard configuration ,
four measurements are needed to estimate it . in the case of uncorrelated measurements , two intensity images
gives a good estimation of this degree using the osci @xcite .
however , in some cases the uncorrelation of the measurements may be not valid .
+ in this paper , an original estimation method that both uses a pair of images and that accounts for correlation in the components is proposed .
we recall in the following the method proposed in @xcite and we extend it to correlated measurements .
we then compare them through statistical measures using simulated data .
the results are also presented when the standard estimation of @xmath25 is used ( four images ) as this case is expected to give the best results .
finally we conclude and give the perspectives of this work .
in the case of two uncorrelated images , the orthogonal state contrast image ( osci ) is determined from @xcite , @xcite , @xcite @xmath26 where @xmath27 and @xmath28 are intensity measurements at the pixel site @xmath29 , assuming a lexicographic order for the pixels .
+ these two images are obtained with simple polarimetric systems .
first , the scene is illuminated by coherent light with single elliptical polarization state .
then the backscattered light is analysed in the polarization state parallel ( which leads to @xmath30 and orthogonal ( which leads to @xmath31 to the incident one .
+ the osci is an estimation of @xmath25 in each pixel provided that the materials of the scene modify the degree of polarization of incident light without modifying its principal polarized state ( . ] ) .
let us recall that this kind of material is called pure depolarizer . for such materials ,
the covariance matrix @xmath8 is diagonal and , since the diagonal terms represent the intensity images , the osci gives an estimation of @xmath25 with @xmath32 in the case of non pure depolarizer objects ( i.e. @xmath8 is non diagonal ) , the osci still reveals interesting contrast image but no more defines @xmath25 .
this leads us to a new method that considers the cases of correlated measurements .
this is the object of the following part .
in the case of correlated measurements , the covariance matrix is non diagonal and of the form ( [ cov ] ) . in its standard estimation ,
the degree of polarization needs four measurements , however , two images are sufficient to get an estimation of @xmath25 .
indeed , the coefficient @xmath33 is obtained from one measurement , as the coefficient @xmath34 and the squared modulus of @xmath35 can be estimated from the cross - correlation coefficient @xmath36 between two measurements @xmath37 and @xmath38 .
we have @xmath39 @xmath36 can be calculated by using the joined density probability function @xmath40 assuming that @xmath41 is gaussian circular ( i.e. the speckle is supposed to be fully developped ) .
it can be shown that the correlation coefficient is obtained with @xmath42 where @xmath43 stands for the modulus of @xmath44 . calculating the centered correlation coefficient defined by @xmath45 after some simple algebra
, we get @xmath46 thus the coefficient @xmath47 can be obtained from two measurements with @xmath48 .
this remark leads to write the following property . + * property a : * _ for fully developped speckle fluctuations , the degree of polarization can be obtained from only two intensity images . _ + + one can easily note that the degree of polarization can be written as a function of the osci with @xmath49 from ( [ posci ] ) , the degree of polarization estimated from the osci is clearly under - estimated .
thus , we can correct the osci in order to get an estimation of the degree of polarization using the correlation coefficient @xmath50 .
+ in the following part , the different estimation of the degree of polarization are compared to the estimation with four images through simulated data .
we generated @xmath51 experiments of @xmath52 samples of complex jones vectors which follow a gaussian circular law .
the covariance matrix is known and thus @xmath25 is also known . under the assumption that the statistical average can be estimated by spatial averages in homogenous regions ,
the coefficients @xmath33 can be estimated from a single image ( @xmath53 ) like the coefficient @xmath34 ( @xmath54 ) since @xmath55 @xmath56 where @xmath57 represents the component of the jones vector for the sample @xmath58 .
+ the estimation of @xmath25 differs in the studied cases by the way @xmath47 is estimated .
three different methods are used : * case of four images + in this situation , we have both the real and the imaginary part of the coefficient @xmath35 .
the quantity @xmath47 is estimated by + @xmath59 * case of two images with the osci + in this case @xmath35 is assumed to be equal to zero .
* case of two images with the proposed approach + the coefficient @xmath47 is estimated using ( [ cross_coef0 ] ) with + @xmath60 in the three cases , @xmath25 was estimated from the relation ( [ deg_pol ] ) with the estimated parameters . +
in order to characterize the precision of the estimation , one considers six examples of matrix @xmath8 .
@xmath61 ; \gamma_2=\left [ \begin{array}{cc } 16 & 0\\ 0 & 3.6 \end{array}\right];\ ] ] @xmath62 ; \gamma_4=\left [ \begin{array}{cc } 18 & 7 + 8i\\ 7 - 8i & 11 \end{array}\right];\ ] ] @xmath63 ; \gamma_6=\left [ \begin{array}{cc } 1.25 & 5.5i\\ -5.5i & 26 \end{array}\right];\ ] ] these matrices were chosen such as the degree of polarization are approximatively in @xmath64 .
+ the simulations are performed for @xmath65 realisations of @xmath66 samples for the six covariance matrices . for the two matrices @xmath67 and @xmath68 , supplementary cases have been studied when @xmath52 @xmath69 @xmath70 .
the results are presented in figures [ fig1 ] , [ fig2 ] , [ fig3 ] , [ fig4 ] , [ fig5 ] , [ fig6 ] .
several points are important to notice .
first of all , as expected , the best estimations of the degree of polarization , regarding all the cases tested , was achieved with four images .
however the proposed approach that relies on two correlated images produces good estimations of the degree of polarization whatever the covariance matrix is used ( fig.[fig1 ] ) as soon as @xmath71 ( fig.[fig3 ] and fig.[fig5 ] ) . note that the estimation with the osci gives results which can not be used if the term @xmath72 is too high ( for example if @xmath47 is non negligible ) .
fig.[fig2 ] , [ fig4 ] and [ fig6 ] show that the experimental variance using four measurements or the osci are comparable whatever @xmath8 and @xmath52 are , whereas the variance obtained with the proposed approach is larger
. however this precision should be sufficient for some practical applications .
this point should be studied in details in a future work .
we have proposed a new approach to estimate the degree of polarization on polarimetric images degraded by speckle noise . assuming that the speckle is fully developped , this method allows one to estimate this degree with only two intensity images whereas four images are needed in a standard experimental setup .
this presents a great interest in term of reduction of cost of the imagery system since the original setup can be simplified .
the proposed approach has been tested on simulated data and compared to the standard estimation techniques that requires either 4 images or 2 independant images ( osci ) .
the results show that the proposed method gives good approximation of the degree of polarization .
this study needs to be extended with a theoritical analyses in order to precise the conditions of validity of the proposed approach . | active polarimetric imagery is a powerful tool for accessing the information present in a scene . indeed
, the polarimetric images obtained can reveal polarizing properties of the objects that are not avalaible using conventional imaging systems @xcite .
however , when coherent light is used to illuminate the scene , the images are degraded by speckle noise .
the polarization properties of a scene are characterized by the degree of polarization . in standard polarimetric imagery system , four intensity images are needed to estimate this degree @xcite .
if we assume the uncorrelation of the measurements , this number can be decreased to two images using the orthogonal state contrast image ( osci ) @xcite .
however , this approach appears too restrictive in some cases .
we thus propose in this paper a new statistical parametric method to estimate the degree of polarization assuming correlated measurements with only two intensity images .
the estimators obtained from four images , from the osci and from the proposed method , are compared using simulated polarimetric data degraded by speckle noise . |
segue 1 was discovered by @xcite as an overdensity of resolved stars in imaging data from the sloan digital sky survey ( sdss ) .
it is located at equatorial coordinates @xmath1 and has a heliocentric distance of @xmath2 kpc .
this corresponds to galactic coordinates @xmath3 , @xmath4 and a galactocentric distance of @xmath5 kpc .
@xcite suggested that segue 1 was an extended globular cluster , possibly associated with the sagittarius stream . with its initially determined half - light radius of approximately @xmath6 pc
, it would be amongst the largest milky way globular clusters such as palomar 5 .
the object is also unusually faint ( @xmath7 ) for its size .
segue 1 has some points in common with two other discoveries made with sdss namely willman 1 @xcite and bootes ii @xcite .
all three objects have similar absolute magnitudes and half - light radii , intermediate between the dwarf spheroidal galaxies , which are dark matter dominated , and the globular clusters , which show no evidence for dark matter ( see figure @xmath8 in @xcite ) . determining the true nature of these three objects is important as it may shed light on the important question of the size of the smallest dark matter haloes in which baryons collapsed to form galaxies .
the interpretation of segue 1 as a globular cluster has recently been contested by @xcite . using keck / deimos spectroscopy , they measured the radial velocities of 24 stars in segue 1 with a mean heliocentric velocity of @xmath9 kms@xmath10 and a velocity dispersion of @xmath11 kms@xmath12 , leading to claims that segue 1 is a dwarf galaxy rather than a globular cluster . assuming these stars are gravitationally bound to segue 1 and are in dynamical equilibrium ,
then the implied mass - to - light ratio is @xmath13 @xcite , which would make segue 1 the most dark matter dominated galaxy detected to date .
however , both assumptions are questionable
. if segue 1 is a globular cluster that is undergoing tidal disruption , then extra - tidal stars may not be so easy to distinguish from gravitationally bound members and the hypothesis of dynamical equilibrium may be a poor one . more seriously ,
if segue 1 is immersed in the sagittarius stream , then contamination of any sample of segue 1 stars by stream stars may be hard to avoid . in this paper , we analyze sdss and auxiliary canada - france - hawaii telescope ( cfht ) photometry of segue 1 .
we use the optimal filter techniques pioneered by @xcite to identify tidal features in 2 and 3 .
a stellar population embedded deep in a massive halo would not be expected to show visible signs of battering by the galactic tides .
we compute segue 1 s structural parameters in 4 . finally ,
in 5 , we extract from the sdss spectral database blue horizontal branch stars and used them to identify the kinematical signal from the sagittarius stream .
we show that contamination of kinematically selected segue 1 stars by sagittarius stream stars is a serious problem , and can lead to artificially inflated velocity dispersions .
the sloan digital sky survey ( sdss ; york et al .
2000 ) is an imaging and spectroscopic survey that covers one - quarter of the celestial sphere .
the sdss data are described in the data release papers ( adelmann - mccarthy 2008 for the sixth release , dr 6 ) and documented at `` http://www.sdss.org . ''
we select stars in a @xmath14 box centered on segue 1 from sdss dr 6 data , using the sdss clean photometry flags and a magnitude cut ( @xmath15 ) to remove data reduction artifacts from the field .
in addition , we correct for extinction using the maps of @xcite and apply the ubercal correction as described in @xcite .
figure [ fig : density]a shows the density of all stars ( fainter than @xmath16 ) selected by dividing the area in @xmath17 pixels and smoothing the field with a gaussian kernel fwhm of 2 pixels ( this is applied to all density plots in this work ) .
segue 1 is too faint to stand out in this figure .
figure [ fig : density]b shows a density plot of stars that we have removed from the sample using our magnitude cut .
structural artifacts due to problems in data reduction are clearly visible and correspond to the sdss scan patterns on the sky .
figure [ fig : density]c shows our estimate of the field star density with extinction contours overplotted .
the estimate is determined by first replacing segue 1 within @xmath18 with a representative patch of the field star distribution taken at @xmath19 . by a similar cloning method
, we remove the other obvious overdensity in the field , the leo i dwarf galaxy ( @xmath20 , at a galactocentric distance of 250 kpc ) .
we then compute the density and smooth the resulting distribution with a gaussian kernel with a fwhm of 5 pixels and a box - car smoothing over 25 pixels . finally , since classification of galaxies and stars especially at faint magnitudes can be uncertain in the sdss , it is important to verify that possible structures are not influenced by such misclassifications .
figure [ fig : density]d shows that there are no obvious galaxy overdensities in the immediate vicinity of segue 1 .
there does seem to be an underdensity , however , this would not generate spurious tidal structures near segue 1 except in the highly unlikely event that almost all galaxies were misclassified as stars .
the main dataset used in this paper is from the sdss .
however , we also have access to a set of megacam pointings taken on the canada - france - hawaii telescope ( cfht ) on 17 - 25th january 2007 .
data were taken in @xmath21 and @xmath22 filters , typically three 200 s exposures using the measured zero - points for megacam .
the megacam @xmath21 and @xmath22 magnitudes are converted to sdss magnitudes by matching the stars in the cfht data to the sdss observations .
we find 21952 matches between the two data sets .
we then plot the difference in magnitudes in the two systems versus cfht color to find the relations @xmath23 there are 10 megacam fields each covering 1 deg@xmath24 , straddling segue 1 at fixed declination . a rectangle bounding the 10 fields is shown in figure [ fig : density]a .
although deeper than the sdss data by about a magnitude , the cfht data cover considerably less area and are not so useful for diagnosing tidal features . nonetheless , it is the best dataset for estimating the surface brightness profile and hence the structural parameters of segue 1 .
.k - s tests comparing the hess diagrams of different above average signal regions and segue 1 .
a high k - s probability indicates a high confidence that the two samples are drawn from the same parent population .
the number of stars is the count of stars inside the cmd mask in a given box .
the count of sagittarius stars ( i.e. stars falling inside the cmd mask but belonging to the sagittarius stream rather than segue 1 ) is estimated using the methods described in 5.1 . [ cols="<,<,^,^,^,^,^,^,^,^ " , ] [ tab : params ] ) plane integrated forwards ( dark grey ) and backwards ( light gray ) for 2 gyr .
the position of segue 1 is close to crossings of the rosette orbit .
the potential is defined by equations ( 1)-(3 ) in fellhauer et al .
the dashed line corresponds to an orbit starting at right ascension @xmath25 , declination @xmath26 and a heliocentric distance of 25 kpc ( the location of the sagittarius ) with initial velocities @xmath27 kms@xmath12 , @xmath28 mas yr@xmath12 , @xmath29 mas yr@xmath12 .
the solid line starts at the same location but has initial velocities @xmath30 kms@xmath12 , @xmath31 mas yr@xmath12 , @xmath32 mas yr@xmath12 . ] and black representing @xmath33 . to guide the eye , the one - dimensional density profile of the sagittarius stream ( dashed curve from fig [ fig : selection ] )
is shown connecting the two panels .
the dashed lines mark the boundaries of the sagittarius stream .
lower : radial velocity distributions for stars in the declination range @xmath34 of the sagittarius stream .
the stream is most clearly identifiable in the right panel in which the velocities are galactocentric .
note the two features at @xmath35 ( the a and b streams ) and @xmath36 kms@xmath10 ( the c and d streams ) .
we also show as dotted lines the characteristic velocity distributions of the galactic halo and thick disk to emphasise that the features can not be ascribed to these components .
finally , the two black distributions are gaussian fits to the stream velocities . ] whilst the grey line represents a contaminating population with dispersion of 15 kms@xmath12 .
the value reported by geha et al .
( 2008 ) is shown as a horizontal dotted line , with the dash - dotted lines giving the reported error . ]
segue 1 lies in a very busy area of the sky with multiple wraps of the sagittarius stream as well as other debris streams , as shown in the `` field of streams '' ( belokurov et al .
it has been known since the work of ibata et al .
( 2001 ) that the sagittarius tidal debris is wrapped around the galaxy a number of times .
there are at least 4 wraps corresponding to two streams , leading and trailing ( see e.g. , * ? ? ?
a schematic plot is shown in figure [ fig : rosette ] , in which the orbit of the sagittarius dwarf is integrated for @xmath37 gyr backwards and forwards in time .
segue 1 is located close to an intersection of the trailing and leading arms .
the `` field of streams '' traces the young leading ( a ) and old trailing ( b ) wraps , which are closely matched in distance around the north galactic cap .
simulations suggest the existence of further streams corresponding to old leading and young trailing wraps in the same field of view .
so far , observational evidence for these wraps is sparse . at the location of segue 1 , the wraps have similar distances and
are composed of similar stars .
the only way to identify them is through their different kinematics . to this end
, we interrogate the sdss spectral database to pick out blue horizontal branch stars .
figure [ fig : selection ] shows the bhb branch of m92 with an enclosed box , which is used for selection in @xmath38 versus @xmath39 . to eliminate false positives
, we simultaneously impose the cut @xmath40 @xcite .
this box selects bhb stars at the approximate distance of segue 1 .
our rationale is to trace with bhbs a slice through the sagittarius stream at constant right ascension ( @xmath41 ) .
any features that stand out in the same range of declination can be attributed to the sagittarius stream .
the right panel tells us what this range of declination is .
the density of turn - off stars is shown as the uppermost solid line , with the turn - off stars selected via the mask shown in the left panel .
the mask is wide enough to include all stars ( 22 , excluding 2 bhbs ) in the spectroscopically confirmed sample of @xcite .
although the mask is centered on the cmd of segue 1 , because of its width , it will also pick out stars in the sagittarius stream . to make sagittarius stand out more clearly
, we subtract the galactic foreground .
the number density of sagittarius stars depends on the model adopted for the foreground . in one method
, we take advantage of the apparent linearity of the foreground ( shown as a dotted line ) .
once subtracted , this gives the lower dashed profile , which peaks at @xmath42 stars per deg@xmath43 .
however , at low declinations , the fit could be in error due to a contribution from the `` virgo overdensity '' ( * ? ? ?
* ; * ? ? ?
* see e.g. ) . as an alternative
, we examine the density of stars at the same galactic latitude but with the sign of longitude reversed .
this results in a lower foreground estimate and a higher peak density for the sagittarius of @xmath44 stars per deg@xmath45 .
the sagittarius stream is limited to declinations satisfying @xmath46 . figure [ fig : signal ] shows the distribution of bhb stars in the plane of radial velocity versus declination . once the gradients due to the projection of the local standard of rest have been removed , the two overdensities coincident with the sagittarius stream in declination stand out at @xmath47 kms@xmath12 and @xmath36 kms@xmath12 . at this location , it is unlikely that these features can be due to the galactic thick disk or halo , as evidenced by the velocity histograms in the lower panel . in fact , they nicely match the velocities of the wraps of the sagittarius stream at this location shown in figure 3 of fellhauer et al . ( 2006 ) .
segue 1 is in the sagittarius stream and appears to be moving with the same velocity as one its wraps . in the lower panels in figure [ fig : signal ] , we show gaussian fits to each of the overdensities associated with sagittarius .
the dispersions are @xmath48 kms@xmath12 . assuming typical radial velocity errors of @xmath49 kms@xmath12
gives the intrinsic dispersion in each wrap to be @xmath50 kms@xmath12 .
of course , this is approximate , as both measured dispersion and the errors could be somewhat larger .
this has immediate consequences for estimates of the velocity dispersion of segue 1 , as the problem of contamination by sagittarius stream stars is substantial .
let us estimate the number of sagittarius stars in the sample of @xcite .
as shown in figure [ fig : selection ] , our mask already wraps around the candidate stars of geha et al .
hence , the total number of sagittarius stars at the location of segue 1 can be read off the dashed curve in the right - hand panel .
it is between @xmath51 and @xmath52 stars per deg@xmath53 depending on the model adopted for foreground subtraction . to estimate how many of these move with @xmath54 kms@xmath12 , we note that the two velocity peaks in the lower panel of figure [ fig : signal ] integrate to roughly the same numbers of stars
so , allowing a factor of 0.5 to to account for the fact that both leading and trailing arm stars occur at this location and scaling to the @xmath55 deg@xmath53 field of view used by geha et al , we estimate that the final contamination in their sample is at least @xmath51 stars deg@xmath56
@xmath57 deg@xmath53 , which comes to 4 stars .
this might be raised as high as 8 stars , if the lower value for the foreground is adopted .
we caution that such estimates necessarily involve extrapolation from comparatively small numbers of bhbs .
could contaminating stars be responsible for the high velocity dispersion of @xmath58 kms@xmath12 reported by @xcite ?
figure [ fig : mc ] shows the results of monte carlo simulations to gauge the importance of this contamination .
the sagittarius stream stars are assumed to have a velocity dispersion of @xmath59 kms@xmath12 and segue 1 stars to have a velocity dispersion of @xmath60 kms@xmath12 ( typical for a globular cluster ) . in the simulations , samples of 24 stars
are generated , with the number of contaminating sagittarius stream stars allowed to vary between 0 and 10 .
the simulations are run for 10000 iterations .
the points on the plot are the average dispersion measured in the 10000 samples of 24 stars and the error bars represent the spread .
the number of contaminating sagittarius stars required to inflate the velocity dispersion from its true value ( in the simulation ) of @xmath60 kms@xmath12 to the value reported by geha et al is surprisingly small perhaps even one contaminating star is enough ! recall that we estimate the actual number of contaminants in the geha et al sample is probably between 4 and 8 .
although the kinematic selection of geha et al is sufficient to exclude stars from one of the wraps of the sagittarius , it is not sufficient to exclude stars from the wrap moving with @xmath61 kms@xmath12 .
a very modest contamination from this wrap can cause an anomalously high velocity dispersion to be reported . assuming that segue 1 is on a sagittarius - like orbit , its radial and tangential velocities are sufficiently well constrained to ask what effect galactic tides would have on the object .
figure [ fig : orbit_info ] shows the results of orbit integrations , in which a test particle is placed at the location of segue 1 with a variety of initial velocities .
the particle travels through the milky way potential as defined in fellhauer et al . (
2006 ) for roughly 2 periods .
we then measure the perigalactic distance of the orbit as well as its eccentricity .
the contours show the regions in which the points lie in the plane of perigalactic distance and eccentricity .
we then calculate the tidal radius for a cluster of mass @xmath62 travelling on such an orbit using @xcite @xmath63 here , @xmath64 is the perigalactic distance , @xmath65 is the eccentricity , and @xmath66 is the mass of the galaxy interior to @xmath64 . in both panels of the figure , the radial velocity is @xmath67 kms@xmath12 ( heliocentric ) or 118 kms@xmath12 ( galactocentric ) .
but , panel ( a ) shows the results with initial tangential velocities corresponding to a typical galactic range ( @xmath68 kms@xmath12 to @xmath69 kms@xmath12 ) , and with cluster masses of @xmath70 and m=@xmath71 .
panel ( b ) assumes the tangential velocity is chosen to correspond approximately to that of the old leading arm 215 kms@xmath12 galactocentric ) . from both panels
, we find that if segue 1 is a globular cluster with mass around @xmath72 , then its tidal radius would be smaller than its observed half - light radius which implies that it could not survive on such an orbit for long and would be in the throes of destruction .
if segue 1 is a dwarf galaxy with mass of @xmath71 , the estimated tidal radius is much larger than the measured half - light radius and we would not expect to see any signs of tidal disruption
the distinction between star clusters and dwarf galaxies is largely based on size .
there is a factor of @xmath73 difference in the characteristic sizes of clusters and galaxies . despite differences in the velocity dispersion ,
this translates into an order of magnitude difference in the estimated mass - to - light ratios , under the assumption of virial equilibrium .
the example of pal 5 shows the dangers of assuming virial equilibrium .
pal 5 has a half - light radius of @xmath48 pc , one of the largest of the milky way globular clusters .
it is also well - known to be disintegrating under the galactic tidal field , and so its half - light radius exceeds its tidal radius .
the most distant stars are therefore already unbound and belong properly speaking to pal 5 s tidal tails .
the assumption of virial equilibrium would lead to a mass - to - light ratio that is in serious error .
as noted by dehnen ( 2004 ) , the final disruption of pal 5 happens very quickly , and most mass is lost during the final few percent of its lifetime .
another example of a dissolving milky way satellite is provided by ursa major ii ( uma ii ) . in the model of fellhauer
et al . ( 2007 ) , the orbit of uma ii has pericentric and apocentric distance similar to that of the sagittarius .
the fate of uma ii is very similar to that of pal 5 , but the final disruption happens on a longer timescale , comparable to the orbital period of 1 gyr ( see figure 6 of fellhauer et al .
suppose we were to observe pal 5 or uma ii at a future time equal to a fraction of their orbital period .
then , just a few faint stars will mark the nucleus , whilst any streams will become blurred and of lower contrast as they diffuse in the milky way potential .
in other words , we will see something like segue 1 . of course
, to prove the hypothesis of extra - tidal stars for segue 1 , what is needed is wide - field kinematical data . at the moment ,
the only published study is by @xcite .
they measured the radial velocities of 24 stars and concluded that segue 1 has a heliocentric velocity of @xmath9 kms@xmath10 and a velocity dispersion of @xmath74 kms@xmath12 .
the systemic velocity therefore coincides with both the predicted velocity of the sagittarius stream @xcite and the measured velocity as recorded in this paper .
this not only strengthens the argument that segue 1 is a star cluster , originating from the sagittarius , but also raises the question of the levels of contamination in kinematically selected datasets around segue 1 s location .
in particular , such contaminants may be inflating the velocity dispersion , and hence the inferred mass - to - light ratio , of segue 1 .
taking advantage of the wide field of view of the anglo - australian telescope/ aaomega fiber - fed spectrograph combination , gilmore and collaborators have observed a @xmath75 area around segue 1 .
observing parameters were the same as those for the related study of bootes 1 ( see norris et al .
targets are selected from the sdss photometric data to be broadly consistent with the segue 1 cmd published by belokurov et al .
( 2007 ) and a radial velocity accuracy of @xmath76 kms@xmath12 is attained . analysis of this dataset is ongoing and will be presented elsewhere . here ,
we are primarily concerned with existence of extratidal stars .
the left panel of figure [ fig : aat ] shows the dataset plotted in the plane of heliocentric radial velocity and angular distance from segue 1 .
applying a kinematic cut of @xmath77 kms@xmath12 picks out the 8 likely members , which are reasonably well separated from the galactic foreground populations .
the right panel shows the location of the likely members as compared to the density distribution inferred from the optimal filter .
not only do the candidates extend well beyond the half - light radius marked by the thin circle , but they also follow in broad outline the distribution suggested by the photometric analysis .
this confirms the wide spatial distribution of stars with kinematics like segue 1 in this field .
accordingly , we favour the scenario in which segue 1 is a star cluster stripped early on from the sagittarius galaxy .
this provides a natural explanation for the fact that all four of the phase space coordinates that are measured coincide with those of the sagittarius stream .
what is needed to make the argument complete is a demonstration that the metallicity of segue 1 is compatible with an origin in the sagittarius galaxy .
@xcite measured the metallicity of one star as [ fe / h ] = -3.3 , and the metallicity of a further 13 as [ fe / h ] = -1.8 .
the latter is consistent with the metallicity of sagittarius .
the former is very surprising , as there are no sagittarius clusters with a metallicity lower than -2.2 .
however , the declination and radial velocity of segue 1 place it in the very oldest wrap of material , and so it must have been stripped first and have resided in the very outskirts of the sagittarius progenitor .
this could explain variations in the metallicity of segue 1 and other parts of the sagittarius stream .
regardless of whether the metal - poor star of @xcite is a member of segue 1 or not , it signifies the detection of a substantial metallicity gradient in the sagittarius progenitor , although this is more extreme than has been measured so far @xcite . in any case , the claimed status of segue 1 as the least luminous galaxy is very uncertain .
we thank the referee , rodrigo ibata , for a careful reading of the paper .
mno is funded by the gates cambridge trust , the isaac newton studentship fund and the science and technology facilities council ( stfc ) , whilst vb acknowledges financial support from the royal society .
funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org/. | we apply the optimal filter technique to sloan digital sky survey photometry around segue 1 and find that the outer parts of the cluster are distorted .
there is strong evidence for @xmath0 elongations of extra - tidal stars , extending both eastwards and southwestwards of the cluster .
the extensions have similar differential hess diagrams to segue 1 and a kolmogorov - smirnov test suggests a high probability that both come from the same parent distribution .
the location of segue 1 is close to crossings of the tidal wraps of the sagittarius stream . by extracting blue horizontal branch stars from sloan s spectral database , two kinematic features
are isolated and identified with different wraps of the sagittarius stream .
we show that segue 1 is moving with a velocity that is close to one of the wraps . at this location
, we estimate that there are enough sagittarius stars , indistinguishable from segue 1 stars , to inflate the velocity dispersion and hence the mass - to - light ratio .
all the available evidence is consistent with the interpretation that segue 1 is a star cluster , originally from the sagittarius galaxy , and now dissolving in the milky way .
-.6 in [ firstpage ] galaxies : kinematics and dynamics globular clusters : individual ( segue 1 ) |
in the last fifteen years , power - law distributions , with their heavy - tailed and scale - invariant features , have become ubiquitous in scientific literature in general @xcite .
heavy - tailed distributions have been fitted to data from a wide range of sources , including ecological size spectra @xcite , dispersal functions for spores @xcite , seeds @xcite and birds @xcite , and animal foraging movements @xcite . in the latter case ,
the fit of a heavy - tailed distribution has been used as evidence that the optimal foraging strategy is a lvy walk with exponent @xmath0 @xcite .
however , the appropriateness of heavy - tailed distributions for some of these data sets , and the methods used to fit them , have recently been questioned by edwards _
et al _ @xcite .
in a literature search of studies on foraging movements , ecological size spectra and dispersal functions , the first 24 papers that fitted power - law distributions to data were chosen @xcite . in only six of these 24 papers was it clear that the authors correctly fitted a statistical distribution to the data , whereas 15 used a flawed fitting procedure ( and in the remaining three it was unclear ) .
ten of the studies compared the goodness - of - fit of the power - law distribution to other types of distribution , but at least two of these ten used comparison methods that were invalid in this context . correctly fitting a power - law distribution to data is not difficult but requires a little more statistical knowledge than the standard linear regression and linear correlation techniques used in most cases . in this report ,
the most common methods currently used to fit a power law to data are reviewed , and some of the drawbacks of these methods are discussed . a method that avoids these drawbacks , based on maximum likelihood estimation ,
is then described in a step - by - step guide ( the reader is also referred to @xcite ) .
two case studies are presented illustrating the advantages of this method .
the probability density function ( pdf ) for a power - law distribution has the form @xmath1 this function is not defined on the range 0 to @xmath2 , but on the range @xmath3 to infinity , otherwise the distribution can not be normalised to sum to @xmath4 .
this condition implies that the exponent @xmath5 is related to @xmath3 and @xmath6 by @xmath7 a common problem is to determine whether a particular sample of data is drawn from a power - law distribution and , if so , to estimate the value of the exponent @xmath5 . a standard approach to
this problem ( used by 14 of the papers in the literature sample ) is to bin the data and plot a frequency histogram on a log log scale .
if the sample is indeed drawn from a power - law distribution and the correct binning strategy is used , there is a linear relationship between @xmath8 and @xmath9 , so the frequency plot will produce a straight line of slope @xmath10 @xcite .
alternatively , one may use the cumulative distribution function ( cdf ) @xmath11 , which shows a linear relationship between @xmath8 and @xmath12 .
so plotting @xmath13 ( _ i.e. _ the relative frequency of observations with a value greater than @xmath14 ) against @xmath14 ( called a rank frequency ( rf ) plot ) on a log log scale will give a straight line of slope @xmath15 .
all data sets are subject to statistical noise , particularly in the tail of the distribution .
hence a frequency chart will never produce a perfectly straight line , so some kind of fitting procedure is necessary .
the most common method , used by 16 of the 24 studies surveyed , is to estimate the power - law exponent by using linear regression to find the line of best fit on the frequency histogram or rf plot .
linear regression assumes that one is free to vary both the slope and intercept of the line in order to obtain the best possible fit .
however , this is not true in the case of fitting a probability distribution , because of the constraint that the distribution must sum to @xmath4 .
once the range of the data has determined @xmath3 , the power - law distribution has only one degree of freedom @xmath5 , as opposed to the two afforded by the naive linear regression or line of best fit approach .
unless this constraint is explicitly acknowledged and respected , the fitting procedure will produce an incorrect estimate of the exponent @xmath5 and the fitted distribution will not be a pdf on the appropriate @xmath14 range ( see for example case study 1 ) .
furthermore , the linear regression method does not offer a natural way to estimate the size of the error in the fitted value of @xmath5 ( very few of the studies surveyed made any attempt to do this ) .
a point value for @xmath5 on its own is of questionable merit .
a third criticism of the linear regression method of fitting a power law to some sample is that very rarely is any meaningful attempt made to judge the goodness - of - fit of the proposed model .
the most common approach , taken by twelve of the surveyed works , is to provide the correlation coefficient @xmath16 .
this is a measure of the strength of linear correlation between @xmath8 and @xmath9 , but is not a measure of the goodness - of - fit of a proposed model such as a power law .
of course , in the absence of any _ a priori _ knowledge of the distributions of errors in the data , measuring goodness - of - fit of a single model is extremely difficult .
however , it is possible to compare goodness - of - fit of two or more candidate models . in the work surveyed
, nine papers explicitly compared the goodness - of - fit of the power - law distribution with other candidate models , but at least two of these nine used a comparison based on the flawed linear regression method and on @xmath16 , so in reality added little to the analysis . finally , although it is not necessary to bin the data to use a linear regression method , 14 of the 24 surveyed papers did so in their analyses , and nine used bins of equal width .
it should be noted that , when a sample that does follow a power - law distribution is binned using fixed width bins , the log log plot of the pdf is not linear . to see a linear relationship one must use logarithmic width bins .
this fitting approach is described in detail in @xcite , but the statistical results are often sensitive to the binning strategy ( in particular the logarithmic base ) used @xcite and the occurrence of empty bins is problematic .
case study 1 illustrates some of the nonsensical conclusions that can result from binning the data . by far the best approach to fitting a power law
is not to bin the data in any way , but to use the raw data whenever possible . a simple technique for fitting a power - law distribution is to use maximum likelihood estimation ( mle ) .
this method provides an unbiased estimate of the exponent @xmath5 @xcite .
equally importantly , it provides an estimate of the error in the fitted value of @xmath5 , and allows the goodness - of - fit of different candidate models to be directly compared @xcite .
furthermore , the method does not require the data to be binned in any way . in the step - by - step guide
below , the method is used to compare the fit of a power - law distribution and an exponential distribution , @xmath17 , on the same range @xmath18 , but can be easily adapted to compare with other candidate distributions ( _ e.g. _ normal distribution , gamma distribution ) .
the method proceeds in the following steps . 1 .
* draw a rf plot . * to gain a picture of the distribution of the data , plot the fraction of observations greater than @xmath14 against @xmath14 on a log log plot by : * ordering the data from largest to smallest so that @xmath19 ; * plotting @xmath20 against @xmath21 ( for @xmath22 ) .
2 . * choose @xmath3 .
* in some cases , it may be desirable to examine only the ` tail ' of the distribution by first discarding from the sample all values less than some cut - off @xmath23 .
typically , this is chosen so as to disregard the curved part of the rf plot on its left - hand side . if it is required to fit a distribution to the entire sample , take @xmath3 to be the smallest observation .
3 . * calculate mle parameters and likelihoods : * * mle power - law exponent @xmath24 * log - likelihood for power - law model @xmath25 * mle exponential parameter @xmath26 * log - likelihood for exponential model @xmath27 + where @xmath28 is the number of data points in the ( truncated ) sample .
see @xcite for a derivation of these formulae .
* select the best model .
* for each model , calculate the akaike information criterion ( aic ) and akaike weight ( @xmath29 ) , which are defined , for model @xmath30 , by @xcite : @xmath31 where @xmath32 is the number of parameters in model @xmath30 ( @xmath33 for the power - law and exponential models ) , @xmath34 is the total number of models being compared ( in this case , comparing a power - law model and an exponential model means that @xmath35 ) and @xmath36 is the smallest of the aic across all @xmath34 models .
+ the akaike weight gives a measure of the likelihood that a particular model provides the best representation of the data .
if @xmath37 then a power - law distribution provides a better fit to the data than an exponential distribution , and the estimated value of the exponent of the power law is @xmath38 .
if @xmath39 then an exponential distribution is favoured over a power law on that range .
calculate the error in the estimate of @xmath5 . * the standard deviation @xmath40 of the estimated power - law exponent @xmath38 may also be calculated : @xmath41 this gives the average size of the error in the estimated value of @xmath5 .
six of the 24 surveyed studies correctly fitted the data and compared different candidate models using a method similar to the one described above .
( it is interesting to note that these six were all found in the dispersal literature . )
austin _ et al _ @xcite fitted power - law distributions to the movement lengths of grey seals ; figure [ fig : seals ] shows an example data set consisting of 96 observations .
the original data are already binned , so it was necessary , for the re - analysis , to assume that all observations in the sample were at the midpoint of their respective bin . using linear regression
, the line of best fit has equation @xmath42 ( in log coordinates ) , leading to an estimated value of the exponent of @xmath43 .
however , it was omitted from @xcite that , in order for the distribution to sum to @xmath4 , this line of best fit implies an @xmath3 value of @xmath44 km . as the data are in the range @xmath45 km to @xmath46 km , it is nonsensical to fit a distribution with range @xmath47 .
these data were re - analysed in @xcite .
this work used logarithmic binning but , again , used linear regression to find the line of best fit and estimated the exponent to be @xmath48 .
this exponent can not be correct as any power law with an exponent of less than @xmath4 can not be normalised to sum to @xmath4 and hence is not a probability distribution .
this illustrates the major drawbacks of using linear regression on binned data : the fitted line may be sensitive to the size and number of bins used and to the occurrence of empty bins , and may result in a power law that is not an appropriate probability distribution . using the maximum likelihood method outlined above
, the estimated exponent is @xmath49 ( with an expected error of @xmath50 ) .
this explicitly presumes a minimum data value of @xmath51 km ( the smallest observation in the sample ) .
however , comparing to an exponential distribution as described above shows that the exponential distribution ( with @xmath52 ) provides a much better model for this data set than a power law ( @xmath53 ) .
hence , the power - law hypothesis for this data set can be confidently rejected .
( in km ) on a log log scale , together with the line of best fit @xmath54 .
this line is not a probability distribution on an appropriate @xmath14 range.,width=302 ] the data set shown in appendix [ sec : appb ] contains the land area ( in hectares ) of 789 farms in the hurunui district of new zealand ; this data set was chosen for illustrative purposes . the rf plot for this data set
is shown in figure [ fig : farms](a ) .
we find the maximum likelihood estimate of the power - law exponent for a range of values of the minimum cut - off @xmath3 ( discarding all data less than @xmath3 ) using the method described above . by calculating the akaike weights , it is possible to determine whether an exponential or a power - law distribution provides the better fit to the data , for any given value of @xmath3 .
a power law is favoured over an exponential ( @xmath37 ) for all choices of @xmath23 between @xmath55 and @xmath56 , but an exponential is favoured ( @xmath39 ) for @xmath57 .
figure [ fig : farms](b ) shows the estimated power - law exponent @xmath38 for the different @xmath3 values : it is clear that the estimated exponent has a wide range of values from @xmath45 to @xmath58 , depending on the range of data used .
it should also be noted that , if the aim is to find the best statistical model describing the entire data set ( _ i.e. _ @xmath3 equal to the smallest observation in the sample ) , then an exponential distribution ( with @xmath59 ) is clearly favoured over a power law ( @xmath60 ) .
the dependence of the outcome of the analysis on the arbitrary selection of @xmath3 further illustrates the potential problems with fitting power - law distributions . against the cut - off @xmath3 . a power law provides a better fit to the data than an exponential distribution for value of @xmath3 between @xmath55 and @xmath56 ; an exponential provides a better fit for values of @xmath3 outside this range.,title="fig:",width=302 ] + ( a ) + against the cut - off @xmath3 .
a power law provides a better fit to the data than an exponential distribution for value of @xmath3 between @xmath55 and @xmath56 ; an exponential provides a better fit for values of @xmath3 outside this range.,title="fig:",width=302 ] + ( b )
some of the statistical procedures commonly used to fit power - law distributions to ecological data often lead to misleading or incomplete conclusions @xcite . in this paper ,
a simple step - by - step method for fitting a power law to a data set has been described .
this method , based on maximum likelihood estimation , provides an unbiased estimate of the power - law exponent , as well as the expected error in this value , and assesses the goodness - of - fit of a power - law model compared to alternative candidate models such as the exponential distribution .
the authors are grateful to environment canterbury for providing the data set on farm sizes , and to richard duncan and two anonymous referees for useful comments that helped to improve this paper .
the following data set ( courtesy of environment canterbury ) contains the sizes ( in hectares ) of 789 farms in the hurunui district of new zealand .
0.06 , 0.08 , 0.08 , 0.08 , 0.1 , 0.1 , 0.1 , 0.1 , 0.1 , 0.3 , 0.5 , 0.5 , 0.53 , 0.6 , 0.9 , 1 , 1 , 1 , 1 , 1 , 1 , 1 , 1.3 , 1.5 , 1.9 , 1.9 , 2 , 2 , 2 , 2 , 2 , 2 , 2 , 2 , 2 , 2 , 2.3 , 2.9 , 2.9 , 2.9 , 3 , 3 , 3 , 3 , 3 , 3 , 3 , 3 , 3.2 , 3.3 , 3.5 , 3.8 , 3.9 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4.2 , 4.2 , 4.3 , 4.6 , 4.8 , 4.9 , 5 , 5 , 5 , 5 , 5 , 5.4 , 5.7 , 6 , 6 , 6 , 6 , 6 , 6 , 6.2 , 6.3 , 6.3 , 7 , 7 , 7 , 7 , 7 , 7.2 , 7.5 , 7.9 , 8 , 8 , 8 , 8 , 8 , 8 , 8 , 8.3 , 8.8 , 9 , 9 , 9 , 9 , 9.3 , 9.9 , 10 , 10 , 10 , 10 , 10 , 10.2 , 10.4 , 11 , 11 , 11 , 11 , 11 , 11.8 , 12 , 12 , 12 , 12 , 12 , 12 , 12 , 12 , 12.1 , 13 , 13 , 1 , 13 , 13.8 , 13.9 , 14 , 14 , 14 , 14 , 14.4 , 14.5 , 15 , 15 , 15 , 15 , 15 , 15 , 15.2 , 15.7 , 15.8 , 16 , 16 , 16 , 16 , 16 , 17 , 17 , 17 , 17 , 17 , 17.1 , 17.1 , 17.5 , 17.6 , 18 , 18 , 18 , 18 , 18 , 18 , 18 , 18 , 19 , 19 , 19 , 19 , 19 , 19 , 19 , 19 , 19 , 20 , 20 , 20 , 21 , 21 , 21 , 21.9 , 22 , 22 , 23 , 23 , 23 , 24 , 24 , 24 , 25 , 25 , 26 , 27 , 27 , 27 , 28 , 28 , 29 , 29 , 30 , 31 , 31 , 32 , 32 , 32 , 32 , 32 , 32 , 32.9 , 34 , 34 , 35 , 35 , 35 , 36 , 36 , 36 , 37 , 38 , 38 , 39 , 39 , 39 , 40 , 41 , 43 , 43 , 44.1 , 45 , 45 , 46 , 46 , 46 , 46 , 47 , 47 , 47 , 47 , 48 , 49 , 50 , 53 , 54 , 55 , 56 , 56.5 , 57 , 58 , 58 , 60 , 63 , 67 , 68 , 69 , 70 , 71 , 72 , 72 , 72 , 73 , 73 , 74 , 75 , 75 , 75 , 75 , 77 , 78 , 78 , 78 , 78 , 79 , 80 , 81 , 82 , 83 , 85 , 85 , 87 , 87 , 90 , 90 , 91 , 91 , 91 , 92 , 93 , 93 , 96 , 9 , 97 , 99 , 99 , 99 , 100 , 100 , 100 , 101 , 101 , 102 , 103 , 103 , 104 , 104 , 105 , 106 , 107 , 109 , 109 , 113 , 115 , 117 , 118 , 120 , 121 , 121 , 121 , 124 , 124 , 124 , 124 , 125 , 126 , 12 , 128 , 128 , 131 , 131 , 132 , 133 , 135 , 135 , 136 , 136 , 137 , 137 , 137 , 138 , 141 , 142 , 143 , 143 , 143 , 143 , 144 , 144 , 145 , 145 , 148 , 148 , 149 , 149 , 150 , 151 , 153 , 155 , 15 , 156 , 157 , 157 , 158 , 160 , 161 , 162 , 162 , 162 , 163 , 164 , 164 , 165 , 167 , 167 , 169 , 169 , 170 , 171 , 171 , 172 , 173 , 173 , 173 , 175 , 176 , 177 , 181 , 182 , 182 , 183 , 183 , 18 , 184 , 185 , 186 , 188 , 188 , 190 , 191 , 192 , 192 , 192 , 193 , 193.5 , 194 , 195 , 196 , 197 , 198 , 198 , 200 , 200 , 200 , 202 , 202 , 203 , 206 , 207 , 209 , 210 , 210 , 213 , 216 , 216 , 219 , 221 , 223 , 223 , 226 , 226 , 226 , 228 , 229 , 229 , 230 , 232 , 233 , 233 , 234 , 237 , 239 , 242 , 245 , 247 , 247 , 250 , 250 , 252 , 254 , 256 , 258 , 263 , 267 , 267 , 267 , 268 , 269 , 269 , 269 , 270 , 271 , 273 , 273 , 273 , 275 , 277 , 277 , 279 , 280 , 282 , 283 , 284 , 284 , 284 , 288 , 289 , 289 , 295 , 295 , 298 , 298 , 298 , 299 , 301 , 305 , 306 , 308 , 309 , 311 , 31 , 313 , 313 , 316 , 316 , 320 , 322 , 324 , 327 , 332 , 333 , 333 , 339 , 341 , 342 , 345 , 346 , 348 , 349 , 349 , 352 , 353 , 354 , 354 , 355 , 357 , 361 , 362 , 364 , 365 , 365 , 365 , 366 , 36 , 367 , 372 , 372 , 375 , 376 , 379 , 380 , 384 , 384 , 385 , 388 , 388 , 388 , 390 , 391 , 392 , 394 , 395 , 395 , 396 , 398 , 399 , 401 , 401 , 401 , 404 , 405 , 410 , 410 , 415 , 416 , 423 , 42 , 428 , 430 , 431 , 431 , 432 , 440 , 443 , 443 , 448 , 448 , 450 , 453 , 456 , 457 , 461 , 461 , 464 , 466 , 471 , 476 , 478 , 480 , 496 , 497 , 498 , 502 , 504 , 504 , 505 , 511 , 512 , 513 , 51 , 515 , 517 , 518 , 520 , 524 , 526 , 529 , 533 , 540 , 542 , 546 , 550 , 550 , 552 , 552 , 553 , 555 , 555 , 558 , 563 , 565 , 565 , 567 , 567 , 571 , 571 , 576 , 578 , 580 , 581 , 583 , 587 , 58 , 591 , 597 , 597 , 597 , 598 , 602 , 611 , 612 , 614 , 614 , 616 , 630 , 635 , 635 , 640 , 640 , 645 , 650 , 652 , 652 , 661 , 662 , 667 , 681 , 685 , 695 , 699 , 719 , 720 , 731 , 732 , 741 , 75 , 752 , 760 , 763 , 764 , 769 , 771 , 774 , 776 , 785 , 790 , 792 , 792 , 804 , 807 , 814 , 824 , 843 , 844 , 849 , 859 , 874 , 886 , 892 , 892 , 892 , 923 , 923 , 932 , 935 , 938 , 942 , 952 , 95 , 955 , 958 , 975 , 980 , 984 , 989 , 996 , 999 , 1002 , 1004 , 1013 , 1040 , 1044 , 1054 , 1059 , 1064 , 1082 , 1094 , 1110 , 1117 , 1124 , 1153 , 1172 , 1191 , 1196 , 1201 , 1216 , 1219 , 1226 , 1261 , 1262 , 126 , 1278 , 1303 , 1312 , 1336 , 1339 , 1364 , 1371 , 1372 , 1379 , 1380 , 1387 , 1401 , 1403 , 1405 , 1431 , 1431 , 1439 , 1459 , 1477 , 1489 , 1491 , 1521 , 1534 , 1537 , 1544 , 1560 , 176 , 1593 , 1594 , 1607 , 1608 , 1631 , 1690 , 1765 , 1778 , 1780 , 1780 , 1798 , 1812 , 1815 , 1899 , 1917 , 1986 , 2081 , 2152 , 2200 , 2205 , 2288 , 2335 , 2440 , 2576 , 2758 , 2975 , 3378 , 3512 , 4126 , 4136 , 4300 , 4466 , 5299 , 5967 , 6700 , 8596 , 10336 .
bartumeus , f. , peters , f. , pueyo , s. , marras , c. and catalan , j. , 2003 .
helical lvy walks : adjusting searching statistics to resource availability in microzooplankton .
, 100 , 1277112775 .
bertrand , s. , burgos , j. m. , gerlotto , f. and atiquipa , j. , 2005 .
lvy trajectories of peruvian purse - seiners as an indicator of the spatial distribution of anchovy ( _ engraulis ringens _ ) .
_ ices j. marine sci .
_ , 62 , 477482 .
chapman , c. a. , chapman , l. j. , vulinec , k. , zanne , a. and lawes , m. j. , 2003 . fragmentation and alteration of seed dispersal processes : an initial evaluation of dung beetles , seed fate , and seedling diversity .
_ bioone _ , 35 , 382393 .
edwards , a. m. , phillips , r. a. , watkins , n. w. , freeman , m. p. , murphy , e. j. , afanasyev , v. , buldyrev , s. v. , da luz , m. g. e. , raposo , e. p. , stanley , h. e. and viswanathan , g. m. , 2007 .
revisiting lvy flight search patterns of wandering albatrosses , bumblebess and deer .
_ nature _ , 449 , 10441048 .
katul , g. g. , porporato , a. , nathan , r. , siqueira , m. , soons , m. b. , poggi , d. , horn , h. s. and levin , s. a. , 2005 .
mechanistic analytical models for long - distance seed dispersal by wind . _ am . naturalist _
, 166 , 368381 .
klein , e. t. , lavigne , c. , picault , h. , renard , m. and gouyon , p .- h . , 2006 .
pollen dispersal of oilseed rape : estimation of the dispersal function and effects of field dimension .
_ j. appl .
_ , 43 , 141151 . , g. , mateos , j. l. , miramontes , o. , cocho , g. , larralde , h. and ayala - orozco , b. , 2004 .
lvy walk patterns in the foraging movements of spider monkeys ( _ ateles geoffroyi _ ) .
_ , 55 , 223230 . | heavy - tailed or power - law distributions are becoming increasingly common in biological literature . a wide range of biological data has been fitted to distributions with heavy tails .
many of these studies use simple fitting methods to find the parameters in the distribution , which can give highly misleading results .
the potential pitfalls that can occur when using these methods are pointed out , and a step - by - step guide to fitting power - law distributions and assessing their goodness - of - fit is offered . |
the recent discovery of extrasolar giant planets orbiting around nearby solar type stars ( marcy & butler 1998 , 2000 ; mayor & queloz 1995 ) has stimulated renewed interest in the theory of planet formation .
the objects observed so far have masses , @xmath2 , that are characteristic of giant planets , i.e. @xmath3 , m@xmath4 denoting a jupiter mass .
the orbital semi - major axes are in the range @xmath5 , and orbital eccentricities in the range @xmath6 ( marcy & butler 2000 ) .
it is a challenge to formation theories to explain the observed masses and orbital element distributions .
there are two main theories of giant planet formation ( see papaloizou , terquem & nelson 1999 and references therein ) .
one is the core instability scenario . in this , a solid core of several earth masses is built up in the protostellar disc , at which point it is able to begin to accrete gas and evolve to become a giant planet .
once massive enough it is able to open a gap in the disc and undergo orbital migration through disc protoplanet interactions ( e.g. lin & papaloizou 1993 ) .
it has been suggested that the high orbital eccentricities for extrasolar planets might be explained by disc - protoplanet interactions ( artymowicz 1992 ) .
recent simulations of protoplanets in the observed mass range ( kley 1999 , bryden et al . 1999 ,
lubow , seibert & artymowicz 1999 ) interacting with a disc with parameters thought to be typical of protoplanetary discs , but constrained to be in circular orbit , indicate gap formation and upper mass limit consistent with the observations .
however , simulations by nelson et al .
( 2000 ) that relaxed the assumption of fixed circular orbits found inward migration and that the disc - protoplanet interaction leads to strong eccentricity damping .
thus the observed eccentricities of apparently isolated extrasolar planets are so far unexplained by this scenario .
the other possible formation mechanism is through fragmentation or gravitationnal instability in a protostellar disc ( e.g. cameron 1978 , boss 2000 ) .
this may occur early in the life of a protostellar disc surrounding a class 0 protostar on a dynamical timescale .
such discs have been observed ( see , e.g. , pudritz et al . 1996 ) and the characteristic size is about 100 au .
it is unlikely that such a process would operate at distances smaller than about 50 au from the central star as , in the optically thick parts of the disc , non axisymmetric density waves redistibute mass and angular momentum before fragmentation can proceed ( eg .
papaloizou & savonije 1991 , laughlin & bodenheimer 1994 ) .
fragmentation is more likely when cooling is efficient , as may occur in the optically thin parts of the disc , beyond about 50 au ( papaloizou et al 1999 ) .
however , the detailed conditions required for it to occur are unclear and may require constraining influences from the external environment ( pickett et al 2000 ) .
note that fragmentation may also occur before a disc is completely formed , during the initial collapse of the protostellar envelope .
such opacity limited fragmentation has been estimated to produce objects with a lower mass limit of 7 jupiter masses ( masunaga & inutsuka 1999 ) , but there is no definitive argument to rule out somewhat smaller masses ( bodenheimer et al .
it is possible that both a disc and fragments may form simultaneously out of the envelope , the relative importance of the two processes depending for instance on the angular momentum content of the envelope , on the strength of any magnetic field ( so far neglected in disc fragmentation calculations ) and possibly on the initial clumpiness .
note that large scale observations of class 0 envelopes so far do not rule out the presence of clumps with masses smaller than about 10 jupiter masses ( motte & andr 2001 ) .
it is the purpose of this paper to investigate the evolution under gravitational interactions of a distribution of @xmath7 massive planets which we assume to have been formed through a fragmentation process rapidly enough that their orbits can undergo subsequent dynamical relaxation on a time scale of hundreds of orbits . in common with related work on orbital evolution
ocuring after assummed formation in a disc ( eg .
weidenschilling & mazari 1996 , rasio & ford 1996 , lin & ida 1997 ) we shall neglect the effects of any remnant disc gas so that apart from tidal interactions with the central star there are only gravitational interactions .
thus this work complements studies of the initial fragmentation process in a gaseous medium .
it turns out that the resulting evolution leads to similar end states independently of whether the initial configuration is assumed to be in the form of a spherical shell or a disk like structure .
the motivations for this work are firstly the suggestion by black ( 1997 ) and stepinski & black ( 2000 ) that massive extrasolar planets on highly eccentric orbits could actually be the low mass tail of the low mass companion distribution to solar like stars produced by fragmentation processes . here
we wish to investigate to what extent planets with orbital elements similar to those observed can be produced and in particular whether hot jupiters orbiting close to the star can be formed .
secondly we consider the recently detected population of free - floating planets and its relationship to that of planets orbiting solar type stars ( lucas & roche 2000 , zapatero osorio et al . 2000 ) .
it is of interest to know to what extent free - floating planets could be produced as a result of ejection from the neighbourhood of a star .
we have considered the orbital evolution of @xmath7 bodies with masses in the giant planet range , which are assumed to be formed rapidly , using up the gas in a protostellar disc or envelope around a solar mass star , so that they can undergo subsequent dynamical relaxation on a timescale @xmath1 orbits .
we have performed calculations with @xmath8 in all the runs we performed , most of the planets where ejected from the system and at most 3 planets remained bound to the central star .
we found that close encounters with the central star occured for about 10% of the planets for all values of @xmath7 considered .
such close encounters early in the evolution tended to result in collisions .
these tended to be avoided at later times so that tidal interaction might then result in orbital circularization at fixed pericentre distance leading to the formation of a very closely orbiting giant planet .
typically the runs ended up with either ( i ) one potential hot jupiter plus up to two external companions , i.e. planets orbiting near the outer edge of the initial distribution ; ( ii ) one or two external planets or even none at all ; ( iii ) one planet on an orbit with a semi major axis 10 to a 100 times smaller than the outer boundary radius of the inital distribution together with an external companion .
apart from the potential hot jupiters , all these objects are on highly eccentric orbits .
we found that , apart from the close orbiters , the probability of ending up with a planet orbiting at a given distance from the central star increases with the distance .
the objects that become unbound may contribute to a population of freely floating planets ( lucas & roche 2000 , zapatero osorio et al .
2000 ) which could be several times larger than that of giant planets found close to the central star .
thus the dynamical relaxation process considered here may operate in some cases to produce giant planets with high orbital eccentricity at several astronomical units from their central star as well as closely orbiting planets . in all cases a population of loosely bound planets
is also expected .
the plan of this paper is as follows . in section [ mod ]
we describe the model and basic equations used . in section [ ini - bc ] the initial conditions are formulated and in section [ relax2 ] the physics of the relaxation process is discussed . in section [ results ]
we present our numerical results and in section [ disc ] we summarize and discuss them .
we consider a system consisting of @xmath7 planets and a primary star moving under their gravitational attraction . as we are interested in possible close approaches to , or collisions with , the central star we adopt a spherical polar coordinate system @xmath9 with origin at the stellar centre .
the planets and central star are treated as point masses .
however , to take into account possible losses of orbital energy and angular momentum to the stellar material , a simple model for taking into account the tidal interaction between the star and a closely approaching planet is also included .
the equations of motion are @xmath10 here @xmath11 and @xmath12 denote the mass of the central star , the mass of planet @xmath13 the position vector of planet , @xmath13 and @xmath14 respectively .
the acceleration of the coordinate system based on the central star ( indirect term ) is @xmath15 and that due to tidal interaction with planet @xmath16 is @xmath17 in the situation envisaged here , tidal interactions occur when a planet has a close encounter with the star .
when this occurs , the planet approaches from large distances on an almost parabolic orbit .
the time between a subsequent encounters will then be long compared to that for the tidal interaction itself .
accordingly we approximate the process as a sequence of independent energy and angular momentum transfers that occur at each periastron passage .
we utilize the results of press & teukolsky ( 1977 ) who calculated these transfers in the small perturbation limit for a non rotating star modelled as a polytrope and a perturber on a parabolic orbit .
we shall neglect the effects of tides acting on the planet itself .
accordingly our model is simplified .
however , it does enable us to include tidal effects and demonstrate how they start to lead to orbital circularization and gravitational decoupling of an inner planet from the others as it moves onto a close orbit .
however , it is only applicable while the planet orbit has high eccentricity .
we adopt a form for @xmath18 that is able to approximately give the correct angular momentum and energy exchanges with the star on a close approach but which is negligible at larger distances from the central star .
@xmath19 here @xmath20 is the specific angular momentum of planet @xmath13 @xmath21 is the stellar radius , @xmath22 is the distance of closest approach of planet @xmath16 assuming a parabolic orbit , @xmath23 and @xmath24 the equations of motion are here integrated using the bulirsch - stoer method ( eg .
press et al 1993 ) . using ( [ ftidal ] )
we can derive the energy lost to the star during a close encounter of planet @xmath16 , assumed on a parabolic orbit as @xmath25 where , because of the rapid convergence of the integral , the limits are extended to @xmath26 this gives @xmath27 which gives values coinciding approximately with values given by press & teukolsky(1977 ) .
we comment that acording to ( [ de ] ) a star grazing encounter on a parabolic orbit results in a final semi - major axis @xmath28 for a central solar mass with @xmath29 and @xmath30 @xmath31 au .
thus bound orbits with @xmath32 au are significantly modified if they have a close encounter with the central star .
note too that the energy exchange rates are small for the planetary mass objects considered here giving some justification for the linear approximation used to calculate them .
the simulations performed here were begun by placing @xmath7 planets in some specified volume in a random location chosen to give a monte - carlo realization of a prescribed density distribution .
we considered both the case of a uniform density spherical shell with @xmath33 and that of a thick annulus with @xmath33 occupying @xmath34 with a density @xmath35 , where @xmath36 is the spherical radius and @xmath37 the co latitude .
for the calculations presented here , @xmath38 and @xmath39 for the annulus and the spherical shell , respectively .
the planets were then given the local circular velocity in the azimuthal direction .
note that because of the spatial placements the initial orbits are not coplanar . in some cases
the planets were taken to have equal mass @xmath2 while in others each planet was allocated a mass @xmath40 where @xmath41 was a random number between zero and one . in the latter case
the mass distribution of the protoplanets does not correspond to the specified density distribution indicating some redistribution of mass among the embryos . in general
the various initial conditions we have considered lead to the same qualitative evolution .
however , it must be emphasized that the systems are chaotic with the consequence that very small detailed changes to the initial conditions or the integration procedure will in general lead to very different results in detail . to deal with this issue
one can adopt the notion of shadow orbits ( quinlan & tremaine 1992 ) . according to this
, one can expect that although inevitable small errors lead to a significant deviation of the numerical solution from the actual one , there is another solution of the real system obtained with slightly different initial conditions that remains close to the numerical one .
thus we should be able to identify qualitative trends in the evolution of an ensemble of systems of the type we consider and we restrict ourselves to that .
we comment that the equations we solve , incorporating tidal effects , have a radial scale invariance .
thus all radii may be mutiplied by some factor @xmath42 while the timescales are multiplied by @xmath43 the size of the central star has to be scaled by the factor @xmath42 also . here
we shall take the unit of length to be @xmath44 .
then the stellar radius is specified through @xmath45 the time unit is the period of a circular orbit at @xmath46 @xmath47 the interactions amongst the planets lead to some escaping the system
. objects with positive energy which had reached a distance @xmath48 were considered to be escapers . we have considered @xmath49 and @xmath50 which both lead to the same qualitative picture .
even the systems which have a small number of bodies are found to interact strongly and to undergo relaxation like a stellar system ( binney & tremanine 1987 ) .
for such a system the realaxation time is @xmath51 here the root mean square velocity is @xmath52 @xmath53 is the mass density of interacting bodies assumed for simplicity to have equal mass @xmath2 , and @xmath54 using @xmath55 @xmath56 and adopting the orbital period @xmath57 this becomes @xmath58 thus for @xmath59 and @xmath60 the relaxation time is about @xmath61 orbits . for systems with @xmath62 in the @xmath61 au to @xmath63 au range this time is around @xmath64 y which is within the estimated lifetimes of protostellar discs .
the evolution we obtain is similar to that undergone by spherical star clusters ( binney & tremaine 1987 ) .
the relaxation , due to binary encounters , causes some objects to attain escape velocity while others become more bound .
eventually all either escape or end up in extended orbits , except for one body which takes up all the binding energy .
this is a generic result provided that close encounters with the star can be avoided .
however , for the parameter range of interest such encounters are likely .
the situation resembles that of accretion of stars from a spherical star cluster by a black hole ( eg
. frank & rees 1976 ) . at a location with radius @xmath65 the fractional volume of phase space containing orbits that would collide with the star if unperturbed is @xmath66 ( this being the ratio of the square of the angular momentum below which an impact is expected to the square of the mean angular momentum ) . if an object can not diffuse out of this volume before impact , the mean time before diffusion into the effective volume produces an impact , for a particular object , is @xmath67 however , the time to diffuse out of the effective volume is @xmath68
if this time is less than the crossing time @xmath69 the expected mean time to impact is @xmath70 this time is just the crossing time divided by the probability of being in the effective volume of phase space which is regularly sampled because of the effective diffusion ( see frank & rees 1976 ) . in our calculations the innermost most tightly bound object undergoes relaxation or phase space diffusion which can lead to close encounters . for @xmath71
au , @xmath72 y , and @xmath73 cm , @xmath74 y. thus we can expect close encounters to occur within the general relaxation process .
in some cases these can lead to a strong tidal interaction which can circularize the orbit and potentially lead to the formation of a hot jupiter.
we here describe a sample of our results which illustrate the characteristic behaviour exhibited by the systems we consider .
the calculations presented are listed in table [ tab1 ] .
we note that if @xmath44 = 100 au , @xmath75 or @xmath76 corresponds to @xmath77 or 3 r@xmath78 , respectively , which is the radius of a protostar with a mass around 1 m@xmath78 in the early stages .
lllll @xmath79 & @xmath7 & @xmath80 & @xmath81 & @xmath82 + + 1 & 5 & @xmath83 & i2 & @xmath84 + 2 & 10 & @xmath85 & i1 & @xmath84 + 3 & 10 & @xmath86 & i2 & @xmath84 + 4 & 8 & @xmath86 & i2 & @xmath84 ( r ) + 5 & 40 & @xmath85 & i1 & @xmath84 + 6 & 100 & @xmath85 & i1 & @xmath87 ( r ) + run 1 corresponds to a system of @xmath60 planets each having a mass @xmath88 figure [ fig1 ] shows plots of @xmath89 in logarithmic scale , where @xmath90 denotes the semi - major axis , for each planet in the system versus time ( measured in units of @xmath91 ) .
each line corresponds to a different planet . during the run 3 planets
escape , a line terminating just prior to an escape .
for this case @xmath92 so that there was no tidal interaction with the central star .
the initial relaxation time for this and other similar cases is on the order of @xmath93 , in agreement with equation ( [ relax1 ] ) . the evolution of @xmath94 @xmath95 being the pericentre distance and @xmath96 being the eccentricity , is also shown in figure [ fig1 ] .
we see that approaches to within @xmath97 of the central star occur for the innermost object on a timescale of the same order of magnitude as the relaxation time .
this is a common feature of the simulations presented here .
note that for @xmath98 au , the closest approach is to within @xmath99 cm , which is comparable to a solar radius .
after about @xmath100 , the main relaxation is over with nearly all the binding energy being contained within one object with @xmath101 and @xmath102 however , after about @xmath103 , the two innermost planets have a close approach which results in their position being exchanged and one of them being ejected .
the end result of this run is then two planets on well separated orbits .
the innermost planet has @xmath101 and @xmath104 a plot of the evolution of the semi major axes and pericenter distances versus time ( measured in units of @xmath91 ) for run 2 in table [ tab1 ] is given in figure [ fig2 ] .
for this run the initial number of planets is @xmath105 and @xmath106 so that tidal effects operate .
after about @xmath107 , there are only two planets still bound to the star .
the others have been ejected .
the innermost planet which is left has @xmath108 and @xmath109 .
if @xmath110 au , this corresponds to @xmath111 au .
figure [ fig3 ] shows the evolution of the semi major axes and pericenter distances versus time ( measured in units of @xmath91 ) for run 3 in table [ tab1 ] . for this run @xmath105 and @xmath112 the plot terminates at about @xmath113 , just before the innermost planet has a close encounter with the star .
the evolution toward the end of the run is zoomed on in figure [ fig3b ] .
we see that the semi major axis decreases significantly , down to about @xmath114 , whereas the pericenter distance varies much less .
this indicates a tidal interaction with the star .
the run if continued much further becomes innaccurate because of our crude treatment of tides , as mentionned in section [ mod ] .
we expect at that point the orbit to circularize at fixed pericentre distance .
evolution of this type which is also quite common in our simulations could potentially lead to a hot jovian mass planet. figure [ fig4 ] shows the evolution of the semi major axes and pericenter distances versus time ( measured in units of @xmath91 ) for run 4 in table [ tab1 ] . in this case @xmath115,@xmath116 and the masses were selected uniformly at random in the interval @xmath117 here again , after about @xmath118 there are only two planets still bound to the star .
the innermost planet has @xmath119 and @xmath120 , which would correspond to 4 au in a 100 au annulus .
we now consider runs with larger @xmath121 figure [ fig5 ] shows the evolution of semi - major axes and pericenter distances versus time ( measured in units of @xmath91 ) for run 5 in table [ tab1 ] . in this case @xmath122 and @xmath123 . in that case
also there are only two planets bound to the star after about @xmath124 .
the others have either been ejected or have collided with the central star .
our treatment of tides is too crude to be certain about the orbital evolution of planets on very eccentric orbits which have grazing approaches to the star . in our simulations ,
such approaches result in the planet being lost .
as far as the subsequent evolution of the system is concerned however , it does not matter whether the planet indeed hits the star of gets circularized on a close orbit . in this run
, evolution occurs on a shorter timescale due to the larger number of planets . here
the innermost planet has @xmath125 and @xmath126 , which is 3 au if @xmath110 au .
figure [ fig6 ] shows the evolution of the semi major axes versus time ( measured in units of @xmath91 ) for the only 2 planets in the system which have not either been ejected or collided with the central star after about 180 @xmath91 , for run 6 in table [ tab1 ] .
also shown are the pericenter distance and the apocenter distance for these planets . here
@xmath127 , @xmath123 and the masses were selected uniformly at random in the interval @xmath128 we see from figure [ fig6 ] that at a time of about @xmath129 the apocenter and the pericenter of the innermost and outermost planets , respectively , are very close to each other .
the innermost planet then suffers a gravitational scattering by the more massive outermost planet which results in an increase of its eccentricity and eventually a collision with the central star .
we are then left with one planet with @xmath130 and @xmath131 .
we have considered the orbital evolution of @xmath7 bodies with masses in the giant planet range , which are assumed to have formed rapidly enough that they can undergo subsequent dynamical relaxation on a timescale @xmath1 orbits .
we have considered @xmath132 we assume that rapid formation may occur as a result of fragmentation triggered by gravitational instabilities or clumping , either in a spherical envelope during the initial protostellar collapse phase or in a disc like configuration ( e.g. pudritz et al 1996 ) as that forms . but note that as magnetic fields may play a role the disc may not be entirely centrifugally supported .
we have considered initial distributions of planets with masses in the range 5 to 10 jupiter masses that are in the form of a sphericalshell or are disc like .
however , final outcomes are independent of this . in our calculations , along with previous work related to the core accretion model ( weidenschilling & mazari 1996 , rasio & ford 1996 , lin & ida 1997 ) we neglect the effects of gas , thus the efficiency of fragmentation is assumed to be maximal . although our results can be applied to different radial scales , we focus the discussion below on distributions with an outer radius of 100 au this is the dimension of observed protostellar discs in class 0 objects ( eg . pudritz et al 1996 )
. this work has been partly motivated by the suggestion by black ( 1997 ) and stepinski & black ( 2000 ) that some massive extrasolar planets could actually belong to the low mass tail of the distribution of low mass companions to solar like stars .
that suggestion is derived from the observation that extrasolar planets far enough from the star that tidal circularization does not operate tend to have highly eccentric orbits and that these could be produced by a relaxation process of the type we consider .
altogether we have run about 25 cases with @xmath133 and many more with @xmath134
. we have described six representative runs in detail in the paper .
independently of how many planets we began with , the initial set up and the technical details of the computation , we obtained the same sets of characteristic behaviour and end states . in every case
most of the planets where ejected from the system and at most 3 planets remained bound to the central star after a time typically on the order of a few @xmath135 outer periods @xmath91 .
the dynamical relaxation phase is shorter when the initial number of planets is larger , but when the number of planets is reduced to less than 10 the system evolves in the same way as the systems which had a smaller intital number of planets .
we found that close encounters with the central star often occured ( for about 10% of the planets ) for all values of @xmath7 considered . at an early stage these tended to result in direct collisions .
when a direct collision is avoided , tidal interaction between the star and planet on a very ecentric orbit may result in orbital circularization at fixed pericentre distance which might ultimately lead to the formation of a very closely orbiting giant planet .
as far as we could monitor the runs presented here , we did not find any physical collisions between the planets themselves .
this is consistent with lin & ida ( 1997 ) who found that such collisions were very rare when , as here , mutual tidal interactions between the planets were assumed to be ineffective .
if physical collisions were to occur , they would not be expected to affect the typical outcome of our runs .
typically the runs ended up with either ( i ) one potential hot jupiter plus up to two external companions , i.e. planets orbiting near the edge of the initial distribution ; ( ii ) one or two external planets or even none at all since a gravitational scattering between two planets may result in one being ejected and the other one colliding with the star ; ( iii ) one planet on an orbit with a semi major axis 10 to a 100 times smaller than the inital distribution , e.g. @xmath13610 au for a 100 au distribution , plus one external companion .
apart from the potential hot jupiters , all these objects are on highly eccentric orbits .
we found that , apart for the hot jupiters , the probability of ending up with a planet orbiting at a given distance from the central star increases with the distance .
thus , this scenario produces a decreasing number of planets as we go from 100 au down to 0.1 au .
this is expected since a planet on a highly eccentric orbit has more chance of colliding with the central star as it gets closer to it .
paradoxically the action of the central star tends to clear out a large cavity around it .
we now turn to the characteristics of some of the extrasolar planets detected so far .
we restrict our attention to those with a projected mass @xmath137 m@xmath4 , consistent with the larger masses expected for formation through fragmentation .
there are 7 such isolated objects around hd 190228 , ua3 , hd 2222582 , hd 10697 , 70 vir , hd 89744 and hd 114762 .
all have an eccentricity larger than 0.12 , and for 6 of them @xmath138 .
major axis is in the range 0.882.5 for 5 of the objects , while the others have @xmath139 and 0.43 , respectively .
the characteristics of these planets might be accounted for by our scenario though the two cases with small @xmath90 would only occur with a small probability . amongst the hot jupiters detected so far , @xmath140 boo is particularly massive with @xmath141 this planet is clearly a candidate for being produced by a mechanism of the type we consider . the objects expelled as a result of the type of relaxation process we consider may produce a population of freely floating planets ( lucas & roche 2000 , zapatero osorio et al .
2000 ) which is several times larger than that of giant planets close to the central star .
this population would be expected to be typically at least 10 times larger than the population of massive planets orbiting around the star , and depends on the initial number of planets in the distribution .
however , note that the population of planets orbiting central stars that went through a relaxation process of the type we consider may be significantly smaller than the total if planets are also independently formed through a core accretion process .
the model we have considered in this paper has of course some limitations .
in particular , we have not included the effect of gas which is probably still significant in the early stages of evolution of these systems , and which may result in more planets orbiting at smaller distances from the central star .
also we have not considered the effects of roche lobe overflow for the planets which have a close encounter with the central star ( trilling et al .
1998 ) which may significantly reduce their mass .
the considerations above lead to the suggestion that there may be two populations amongst the extrasolar planets detected so far .
some of the more massive objects may be produced through fragmentation of an envelope or a disc like structure followed by dynamical relaxation .
other predominantly lower mass objects could be produced in a disc as a result of the core accretion model .
we would of course expect to also have hybrid systems , in which both processes have occured .
the relative importance of these processes may depend for instance on the physical parameters , such as the angular momentum content , of the parent cloud
. if the scenario presented here is indeed effective in planet formation , we would expect additional massive planets to be detected further from the central star , and an important distribution of loosely bound , or free - floating objects associated with these systems .
j.c.b.p . acknowledges visitor support from the university paris 6 and the iap where part of this work was carried out .
the authors thank the participants of the eara workshop disks , extrasolar planets and brown dwarfs held at the iap in july 2000 for useful discussions . | following the suggestion of black(1997 ) that some massive extrasolar planets may be associated with the tail of the distribution of stellar companions , we investigate a scenario in which @xmath0 planetary mass objects are assumed to form rapidly through a fragmentation process occuring in a disc or protostellar envelope on a scale of 100 au .
these are assumed to have formed rapidly enough through gravitational instability or fragmentation that their orbits can undergo dynamical relaxation on a timescale @xmath1 orbits . under a wide range of initial conditions and assumptions
the relaxation process ends with either ( i ) one potential hot jupiter plus up to two external companions , i.e. planets orbiting near the outer edge of the initial distribution ; ( ii ) one or two external planets or even none at all ; ( iii ) one planet on an orbit with a semi major axis 10 to a 100 times smaller than the outer boundary radius of the inital distribution together with an external companion .
most of the other objects are ejected and could contribute to a population of free floating planets .
apart from the potential hot jupiters , all the bound objects are on highly eccentric orbits .
we found that , apart from the close orbiters , the probability of ending up with a planet orbiting at a given distance from the central star increases with the distance .
this is because of the tendency of the relaxation process to lead to collisions with the central star .
we discuss the application of these results to some of the more massive extrasolar planets .
giant planet formation extrasolar planets dynamical relaxation
orbital elements |
this work falls into a general framework which consists of observing the behavior of patterns and structures that can be formed after instability onset in an evaporating liquid layer . in previous work , we studied theoretical instability thresholds in pure fluids [ 1,2 ] and in binary mixtures [ 3,4 ] .
what is of interest here , is a two - dimensional numerical simulation study of the transient temperature and fluid motion in the liquid for a liquid evaporating into a nitrogen gas flow .
the chosen liquid is hfe7100 ( an electronic liquid produced by 3 m ) .
the numerical ( cfd ) simulations are performed using the software comsol ( finite elements method ) . the evaporation causes the instability and the gas flow evacuates the liquid vapor .
the setup used for this numerical simulation is represented in fig . [ scheme ] and is inspired from the cimex experimental setup of esa [ 5 ] .
the gas flow is maintained at 100 ml / min in a channel of 3 mm height , while three different liquid thicknesses are considered : 2 , 4 and 8 mm .
the width of the whole setup is 50 mm .
the cover between the liquid and gas channel is @xmath0 thick . at the middle of this cover
, there is an opening with a width of 10.6 mm , allowing contact between the liquid and gas channel .
these items are taken into the geometry of the numerical software comsol .
the boundaries of the whole system are kept at an ambient temperature and pressure of respectively 298 k and 1 atm , except for the gas channel outlet where only the ambient pressure is respected . also , the whole system is surrounded by walls except for the gas flow inlet and outlet .
the interface is kept at a constant height , since in the esa experimental setup the liquid is to be replenished at the same rate as the evaporation rate . at the interface , flux conservation is maintained and a tangential stress balance is considered .
furthermore , a no - slip condition is assumed at the interface .
the assumption of local thermodynamic equilibrium at the interface allows us the use of raoult s law , in which the temperature dependence of the saturation pressure is determined via the clausius - clapeyron relation .
the results present the temperature in the liquid and gas phase as well as the fluid motion in the liquid ( caused by the evolution of the temperature via surface - tension and buoyancy effects ) by means of streamlines as a function of time .
the real total elapsed time is 10 seconds .
two videos are shown , presenting the same results in the following urls : 1 .
link : doi[video 1 - high resolution ] 2 .
link : doi[video 2 - low resolution ] note that in the videos the inner streamlines represent the highest velocity values .
the red color represents the highest observed temperature ( that of the ambient one ) , 298 k. the blue color represents the lowest observed temperature , around 285 k.
from the results in the videos we can observe that first several small rolls are formed near the surface , caused by the surface - tension effect as fig .
[ comparisont1 ] shows for the three liquid layer thicknesses at time @xmath1 . due to buoyancy and as time proceeds ,
the rolls grow towards the bottom of the liquid layer .
then the rolls also grow in horizontal direction merging with each other until a steady configuration is obtained . for a higher liquid layer thickness , the merging occurs earlier and less rolls are left .
furthermore , the temperature gradients decrease as the liquid thickness increases , which is caused by the higher mixing efficiency when the liquid is less confined . moreover , the rolls extend more horizontally under the cover towards the side walls as the liquid layer thickness increases . for smaller liquid layer thicknesses ,
the rolls reach the bottom where a constant temperature of 298 k is maintained . therefore the rolls stay concentrated close to the interface . as the liquid layer thickness increases , the rolls have more time to increase in size towards the side walls before they reach the bottom of the liquid layer .
[ comparisont10 ] shows this at the time @xmath2 .
this work yields valuable information about the supercritical instability behavior of an evaporating liquid and the qualitative influence of its confinement by means of fluid dynamics .
the authors gratefully acknowledge financial support of belspo and esa .
[ 1 ] b. haut and p. colinet , j. colloid interface sci . , 285 : 296 - 305 , 2005 . [ 2 ] f. chauvet , s. dehaeck and p. colinet , europhys .
lett . , 99 : 34001 , 2012 .
[ 3 ] h. machrafi , a. rednikov , p. colinet , p.c .
dauby , j. colloid interface sci . , 349 : 331 - 353 , 2010 .
[ 4 ] h. machrafi , a. rednikov , p. colinet , p.c .
dauby , eur .
j. , 192 : 71 - 81 , 2011 .
[ 5 ] esa , http://www.esa.int/specials/hsf_research/semlvk0yduf_0.html[cimex experimental setup ] , accessed 12 octobre 2012 | this work presents fluid dynamics videos obtained via numerical ( cfd ) calculations using comsol ( finite elements method ) software , showing the evaporation of hfe7100 ( 3 m company refrigerant ) into a nitrogen gas flow along the liquid interface .
the overall temperature evolution and liquid motion , which is caused by surface - tension ( marangoni ) and buoyancy ( rayleigh ) instability mechanisms , are shown as well .
flow behavior in the liquid caused by the aforementioned instability mechanisms can be nicely seen .
finally , these observations are made for three liquid thicknesses in order to appreciate the qualitative influence of confinement . |
seasonal unit roots and seasonal heterogeneity often coexist in seasonal data . hence , it is important to design seasonal unit root tests that allow for seasonal heterogeneity . in particular ,
consider quarterly data @xmath5 , @xmath6 generated by @xmath7 where @xmath8 are seasonally varying autoregressive ( ar ) filters , and @xmath9 have seasonally varying autocovariances . for more information on seasonal time
series , see ghysels and osborn ( 2001 ) , and franses and paap ( 2004 ) .
now suppose @xmath10 is a weakly stationary vector - valued process , and for all @xmath11 , the roots of @xmath8 are on or outside the unit circle .
if for all @xmath12 , @xmath8 have roots at @xmath0 , @xmath1 , or @xmath4 , then respectively @xmath13 has stochastic trends with period @xmath14 , @xmath15 , or @xmath16 . to remove these stochastic trends , we need to test the roots at 1 , @xmath1 , or @xmath4 . to address this task , franses ( 1994 ) and
boswijk , franses , and haldrup ( 1997 ) limit their scope to finite order seasonal ar data and apply johansen s method ( 1988 ) .
however , their approaches can not directly test the existence of a certain root without first checking the number of seasonal unit roots . as a remedy ,
ghysels , hall , and lee ( 1996 ) designs a wald test that directly tests whether a certain root exists .
however , in their own simulation , the wald test turn out less powerful than the augmented hegy test .
does hegy test work in the seasonally heterogeneous setting ? to the best of our knowledge , no literature has offered a satisfactory answer .
burridge and taylor ( 2001a ) analyze the behavior of augmented hegy test when only seasonal heteroscadasticity exists ; del barrio castro and osborn ( 2008 ) put augmented hegy test in the periodic integrated model , a model related but different from model .
no literature has ever touched the behavior of unaugmented hegy test proposed by breitung and franses ( 1998 ) , the important semi - parametric version of hegy test . since unaugmented hegy test does not assume the noise having an ar structure , it may suit our non - parametric noise in better . to check the legitimacy of hegy tests in the seasonally heterogeneous setting
, this paper derives the asymptotic null distributions of the unaugmented hegy test and the augmented hegy test whose order of lags goes to infinity .
it turns out that , the asymptotic null distributions of the statistics testing single roots at 1 or @xmath1 are standard .
more specifically , for each single root at 1 or @xmath1 , the asymptotic null distributions of the augmented hegy statistics are identical to that of augmented dickey - fuller ( adf ) test ( dickey and fuller , 1979 ) , and the asymptotic null distributions of the unaugmented hegy statistics are identical to those of phillips - perron test ( phillips and perron , 1988 ) . however , the asymptotic null distributions of the statistics testing any combination of roots at 1 , @xmath1 , @xmath2 , or @xmath3 depend on the seasonal heterogeneity parameters , and are non - standard , non - pivotal , and not directly pivotable .
therefore , when seasonal heterogeneity exists , both augmented hegy and unaugmented hegy tests can be straightforwardly applied to single roots at 1 or @xmath1 , but can not be directly applied to the coexistence of any roots . as a remedy
, this paper proposes the application of bootstrap . in general , bootstrap s advantages are two fold .
firstly , bootstrap helps when the asymptotic distributions of the statistics of interest can not be found or simulated . secondly ,
even when the asymptotic distributions can be found and simulated , bootstrap method may enjoy second order efficiency .
for the aforementioned problem , bootstrap therefore serves as an appealing solution .
firstly , it is hard to estimate the seasonal heterogeneity parameters in the asymptotic null distribution , and to simulate the asymptotic null distribution .
secondly , it can be conjectured that bootstrap seasonal unit root test inherits second order efficiency from bootstrap non - seasonal unit root test ( park , 2003 ) .
the only methodological literature we find on bootstrapping hegy test is burridge and taylor ( 2004 ) .
their paper centers on seasonal heteroscadasticity , designs a bootstrap - aided augmented hegy test , reports its simulation result , but does not give theoretical justification for their test .
it will be shown ( remark [ re : seasonal iid bootstrap ] ) that their bootstrap approach is inconsistent under the general seasonal heterogeneous setting . to cater to the general heterogeneous setting
, this paper designs new bootstrap tests , namely 1 ) seasonal iid bootstrap augmented hegy test , and 2 ) seasonal block bootstrap unaugmented hegy test . to generate bootstrap replicates , the first test get residuals from season - by - season augmented hegy regressions , and then applies seasonal iid bootstrap to the whitened regression errors .
on the other hand , the second test starts with season - by - season unaugmented hegy regressions , and then handles the correlated errors with seasonal block bootstrap proposed by dudek , lekow , paparoditis , and politis ( 2014 ) .
our paper establishes the functional central limit theorem ( fclt ) for both bootstrap tests .
based on the fclt , the consistency for both bootstrap approaches is proven . to the best of our knowledge
, this result gives the first justification for bootstrapping hegy tests under .
this paper proceeds as follows .
section 2 formalizes the settings , presents the assumptions , and states the hypotheses .
section 3 gives the asymptotic null distributions of the augmented hegy test statistics , details the algorithm of seasonal iid bootstrap augmented hegy test , and establishes the consistency of the bootstrap .
section 4 presents the asymptotic null distributions of the unaugmented hegy test statistics , specifies the algorithm of seasonal block bootstrap unaugmented hegy test , and proves the consistency of the bootstrap .
section 5 compares the simulation performance of the two aforementioned tests .
appendix includes all technical proofs .
recall the quarterly data @xmath5 , @xmath6 generated by the seasonal ar model , @xmath17 where @xmath18 , @xmath19 .
if for all @xmath12 , @xmath8 has roots on the unit circle , we suppose that all @xmath8 share the same set of roots on the unit circle , this set of roots on the unit circle is a subset of @xmath20 , and @xmath21 ; otherwise , suppose our data is a stretch of the process @xmath22
, @xmath6 . let @xmath23 and @xmath24 be the regression errors and regression coefficients of , respectively .
more specifically , @xmath23 is the distance between @xmath25 and the vector space generated by @xmath26 , @xmath27 , and @xmath24 is the coefficient of the projection of @xmath23 on the aforementioned vector space . let @xmath28 , @xmath29 . denote by ar@xmath30 an autoregressive process with order @xmath31 , by vma@xmath32 a vector moving average process with infinite moving average order , and by varma@xmath33 a vector autoregressive moving average process with autoregressive order @xmath31 and moving average order @xmath34 . let @xmath35 be the real part of complex number @xmath36 .
let @xmath37 be the largest integer smaller or equal to real number @xmath38 , and @xmath39 be the smallest integer larger or equal to @xmath38 .
assump [ assump 1a ] assume @xmath40 where @xmath41 , @xmath42 ; the @xmath43 entry of @xmath44 , denoted by @xmath45 , satisfies @xmath46 for all @xmath47 and @xmath48 ; the determinant of @xmath49 has all roots outside the unit circle ; @xmath50 is a lower diagonal matrix whose diagonal entries equal 1 ; @xmath51 is a vector - valued white noise process with mean zero and covariance matrix @xmath52 ; and @xmath52 is diagonal .
assumption [ assump 1a ] assumes that @xmath53 is vma@xmath32 with respect to white noise innovation .
this is equivalent to the assumption that @xmath53 is a weakly stationary process with no deterministic part in the multivariate wold decomposition .
the assumptions on @xmath50 and the determinant of @xmath49 ensure the causality and the invertibility of @xmath53 and the identifiability of @xmath52 .
[ assump 1b ] assume @xmath54 where @xmath55 ; @xmath56 ; determinants of @xmath57 and @xmath58 have all roots outside the unit circle ; @xmath59 is the identity matrix ; @xmath60 is a lower diagonal matrix whose diagonal entries equal 1 ; @xmath51 is a vector - valued white noise process with mean zero and covariance matrix @xmath52 ; and @xmath52 is diagonal .
assumption [ assump 1b ] restricts @xmath53 to be varma@xmath33 with respect to white noise innovation .
compared to the vma@xmath32 model in assumption [ assump 1a ] , varma@xmath33 s main restraint is its exponentially decaying autocovariance .
again , the assumptions on @xmath59 , @xmath60 and the determinant of @xmath57 and @xmath58 in assumption [ assump 1b ] ensure the causality and the invertibility of @xmath53 and the identifiablity of @xmath52 . at this stage @xmath61
is only assumed to be a white noise sequence of random vectors .
in fact , @xmath61 needs to be weakly dependent as well .
assump [ assump 2a ] ( i ) @xmath61 is a fourth - order stationary martingale difference sequence with finite @xmath62 moment for some @xmath63 .
( ii ) @xmath64 , @xmath65 @xmath2 , @xmath47 , @xmath48 , and @xmath66 , @xmath67 .
[ assump 2b ] ( i ) @xmath61 is a strictly stationary strong mixing sequence with finite @xmath62 moment for some @xmath63 .
( ii ) @xmath61 s strong mixing coefficient @xmath68 satisfies @xmath69 .
notice the higher moment @xmath61 has , the weaker assumption we require on the strong mixing coefficient of @xmath61 in assumption [ assump 2b ] .
the strong mixing condition in assumption [ assump 2b ] actually guarantees ( ii ) of assumption [ assump 2a ] ( see lemma [ boundedness ] ) .
we tackle the following set of null hypotheses .
the alternative hypotheses are the complement of the null hypotheses .
indeed , the alternative hypotheses can be written as one - sided .
recall we suppose that for all @xmath11 , the roots of @xmath8 are either on or outside the unit circle . since @xmath71 , by the intermediate value theorem , @xmath72 implies @xmath73 , @xmath74 implies @xmath75 , and @xmath76 implies @xmath77 . to further analyze the roots of @xmath8 , hegy ( hylleberg , engle , granger , and yoo , 1990 ) propose the partial fraction decomposition @xmath78 thus @xmath79 + substituting into , we get @xmath80 where @xmath81 indeed
, @xmath82 relates to the root of @xmath83 , i.e. , @xmath84 ; hence the proposition below .
[ prop : hegy ] @xmath85 by proposition [ prop : hegy ] , the test for the null hypotheses can be carried on by checking the corresponding @xmath82 .
further , @xmath82 can be estimated by ordinary least squares ( ols ) .
unfortunately , ols can not be readily applied to season by season , because @xmath86 in are not asymptotically orthogonal for any fixed @xmath12 .
( see also ghysels and osborn , 2001 , p. 158 . ) on the other hand , @xmath87 in non - periodic regression equations and are asymptotically orthogonal ( see lemma [ le : unaug real1 ] ) .
so we wonder if the ols estimators based on and can be used to test the null hypotheses .
when we regress @xmath13 with non - periodic regression equations and , the seasonally heterogeneous sequence @xmath9 is fitted in seasonal homogeneous ar models .
consider , as an example , fitting @xmath9 in a misspecified ar(1 ) model @xmath88 .
then @xmath89 , where @xmath90 .
\label{eqn : tgamma}\ ] ] since @xmath91 is positive semi - definite , we can find a weakly stationary sequence @xmath92 with mean zero and autocovariance function @xmath93 .
we call @xmath92 a misspecified constant parameter representation ( see also osborn , 1991 ) of @xmath9 , and will refer to this concept in later sections .
in seasonally homogeneous setting @xmath94 where @xmath95 , the augmented hegy test detailed below copes with the roots of @xmath96 at @xmath0 , @xmath1 , and @xmath4 . by calculations similar to , hegy ( 1990 )
get @xmath97 where augmentations @xmath98 , @xmath99 , pre - whiten the time series @xmath100 up to an order of @xmath48 . as the sample size @xmath101 ,
let @xmath102 , so that the residual @xmath103 is asymptotically uncorrelated
. let @xmath104 be the ols estimator of @xmath105 , @xmath106 be the t - statistics corresponding to @xmath104 , and @xmath107 be the f - statistic corresponding to @xmath108 and @xmath109 .
other f - statistics @xmath110 , @xmath111 , @xmath112 , and @xmath113 can be defined similarly . in seasonally homogeneous configuration ,
hegy ( 1990 ) proposes to reject @xmath114 if @xmath115 is too small , reject @xmath116 if @xmath117 is too small , reject @xmath118 if @xmath107 is too large , and reject other composite hypotheses if their corresponding f - statistics are too large .
now we apply the augmented hegy test to seasonally heterogeneous processes .
namely , we run regression equation with @xmath13 generated by .
our results show that when testing roots at 1 or @xmath1 individually , the t - statistics @xmath119 , @xmath120 , and the f - statistics have standard and pivotal asymptotic distributions . on the other hand , when testing joint roots at 1 and @xmath1 , and when testing hypotheses that involve roots at @xmath4 , the asymptotic distributions of the t - statistics and the f - statistics are non - standard , non - pivotal , and not directly pivotable .
[ aug real ] assume that assumption 1.b and one of assumption 2.a or 2.b hold .
further , assume @xmath101 , @xmath121 , @xmath122 , and @xmath123 for some @xmath124 , @xmath125
. then under @xmath126 , the asymptotic distributions of @xmath106 , @xmath127 , and f - statistics are given by @xmath128 where @xmath129 , @xmath130 , @xmath131 , and @xmath132 , @xmath133 , @xmath134 , @xmath135 is a four - dimensional standard brownian motion .
the asymptotic distributions presented in theorem [ aug real ] degenerate to the distributions in burridge and taylor ( 2001b ) and del barrio castro , osborn and taylor ( 2012 ) when @xmath9 is a seasonally homogeneous sequence with homoscedastic noise , and to the distributions in burridge and taylor ( 2001a ) when @xmath9 is a seasonally homogeneous finite - order ar sequence with heteroscedastic noise .
[ pm1 ] notice @xmath136 s are standard brownian motions . when @xmath9 is seasonally homogeneous ( burridge and taylor , 2001b , del barrio castro et al . , 2012 ) ,
@xmath136 s are independent , so are the asymptotic distributions of @xmath119 and @xmath120 . on the other hand , when @xmath9 has seasonal heterogeneity , @xmath136 s are in general independent , so @xmath119 and @xmath120 are in general dependent , even asymptotically .
hence , when testing @xmath137 , it is problematic to test @xmath114 and @xmath116 separately and calculate the level of the test with the independence of @xmath119 and @xmath120 in mind . instead , the test of @xmath137 should be handled with @xmath110 .
further , because of the dependence of @xmath119 and @xmath120 , the asymptotic distribution of @xmath110 under heterogeneity is different from its counterpart when @xmath9 is seasonally homogeneous .
hence , the augmented hegy test can not be directly applied to test @xmath137 .
[ heter ] when @xmath9 is only seasonally heteroscedastic ( burridge and taylor , 2001a ) , @xmath138 does not occur in the asymptotic distributions of the f - statistics . on the other hand , when @xmath9 has generic seasonal heterogeneity , @xmath138 impacts firstly the correlation between brownian motions @xmath139 and @xmath140 , and secondly the weights @xmath141 and @xmath142 .
as burridge and taylor ( 2001a ) point out , the dependence of the asymptotic distributions on weights @xmath141 and @xmath142 can be expected .
indeed , @xmath143 is the partial sum of @xmath144 , while @xmath145 is the partial sum of @xmath146 . since these two partial sums differ in their variances , both @xmath147 and @xmath148 involve two different weights @xmath141 and @xmath142 .
[ root ] theorem [ aug real ] presents the asymptotics when @xmath149 has all roots at 1 , @xmath1 , and @xmath4 .
when @xmath13 has some but not all roots at 1 , @xmath1 , and @xmath4 , we let @xmath150 , @xmath151 , and calculate @xmath152 such that @xmath153 .
the asymptotic distributions can be expressed with respective to @xmath152 and end up having the same form with those given in theorem [ aug real ] , where @xmath13 has all roots .
[ power ] the preceding results give the asymptotic behaviors of the testing statistics under the null hypotheses . under the alternative hypotheses ,
we conjecture the powers of the augmented hegy tests tend to one , as the sample size goes to infinity . to see this , we can without loss of generality
assume that @xmath13 has root at none of @xmath0 , @xmath1 or @xmath4 .
then @xmath13 is stationary , and thus for @xmath154 , the @xmath155 corresponding to ( the misspecified constant parameter representation of ) @xmath13 are negative , due to proposition [ prop : hegy ] .
we conjecture that for @xmath154 , the ols estimators @xmath156 in converge in probability to @xmath155 , and as a result the powers of the tests converge to one .
see also theorem 2.2 of paparoditis and politis ( 2016 ) . to accommodate the non - standard , non - pivotal asymptotic null distributions of the augmented hegy test statistics ,
we propose the application of bootstrap . in particular , the bootstrap replications are created as follows .
firstly , we pre - whiten the data season by season to obtain uncorrelated noises .
although these noises are uncorrelated , they are not white due to seasonally heteroscadasticity .
hence secondly we resample season by season in order to generate bootstrapped noise , as in burridge and taylor ( 2001b ) .
finally , we post - color the bootstrapped noise .
the detailed algorithm of this seasonal iid bootstrap augmented hegy test is given below .
[ seasonal iid bootstrap ] step 1 : calculate the t - statistics @xmath119 , @xmath120 , and the f - statistics @xmath157 , @xmath158 and @xmath159 from the augmented hegy test regression @xmath160 step 2 : record ols estimators @xmath161 , @xmath162 and residuals @xmath163 from the season - by - season regression @xmath164 step 3 : let @xmath165 .
store demeaned residuals @xmath166 of the four seasons separately , then independently draw four iid samples from each of their empirical distributions , and then combine these four samples into the vector @xmath167 , with their seasonal orders preserved ; + + step 4 : set all @xmath161 corresponding to the null hypothesis to be zero . for example , set @xmath168 for all @xmath12 when testing roots at @xmath4 .
let @xmath169 be generated by @xmath170 step 5 : get t - statistics @xmath171 , @xmath172 , and the f - statistics @xmath173 from the regression @xmath174 step 6 : run step 3 , 4 , and 5 for @xmath175 times to get @xmath175 sets of statistics @xmath171 , @xmath172 , and the bootstrapped f - statistics @xmath176 .
count separately the numbers of @xmath171 , @xmath172 and @xmath176 than which @xmath119 , @xmath120 , and the f - statistics @xmath157 are more extreme .
if these numbers are higher than @xmath177 , then we consider @xmath119 , @xmath120 , and the f - statistics @xmath157 extreme , and reject the corresponding hypotheses .
it seems also reasonable to keep steps 1 , 2 , 3 , 5 , and 6 of the algorithm [ seasonal iid bootstrap ] , but change the generation of @xmath178 in step 4 to @xmath179 this new algorithm is in fact theoretically invalid for the tests of any coexistence of roots ( see remark [ pm1 ] , [ heter ] , and [ root ] ) , but it is valid for individual tests of roots at 1 or @xmath1 , due to the pivotal asymptotic distributions of @xmath119 and @xmath120 in theorem [ aug real ] . [ re : seasonal iid bootstrap ] if we keep steps 1 , 3 , 5 , and 6 of algorithm [ seasonal iid bootstrap ] , but run regression equations with seasonally homogeneous coefficients @xmath156 and @xmath180 in steps 2 and 4 , then this algorithm is identical with burridge and taylor ( 2004 ) .
however , this algorithm can not in step 2 fully pre - whiten the time series , and it leaves the regression error @xmath103 serially correlated .
when @xmath103 is bootstrapped by seasonal iid bootstrap , this serial correlation structure is ruined . as a result , @xmath181 differs from @xmath100 in its correlation structure , in particular @xmath138 , and the conditional distributions of the bootstrapped f - statistics @xmath176 differ from the distributions of the original f - statistics @xmath157 ( see remark [ pm1 ] and [ heter ] ) .
now we justify the seasonal iid bootstrap augmented hegy test ( algorithm [ seasonal iid bootstrap ] ) .
since the derivation of the real - world asymptotic distributions in theorem [ aug real ] calls on fclt ( see lemma [ le : unaug real1 ] ) , the justification of bootstrap approach also requires fclt in the bootstrap world . from now on ,
let @xmath182 , @xmath183 , @xmath184 , @xmath185 , @xmath186 be the probability , expectation , variance , standard deviation , and covariance , respectively , conditional on our data @xmath13 .
[ iid fclt ] suppose the assumptions in theorem [ aug real ] hold .
let @xmath187 @xmath188 where @xmath189 , & \sigma_{2}^{\star}&=std^{\circ}[\frac{1}{\sqrt{4t}}\sum _ { t=1}^ { 4 t } ( -1)^{t}{\epsilon}_{t}^{\star}],\\ \sigma_{3}^{\star}&=std^{\circ}[\frac{1}{\sqrt{4t}}\sum _ { t=1}^ { 4 t } \sqrt{2}\sin(\frac{\pi t}{2}){\epsilon}_{t}^{\star } ] , & \sigma_{4}^{\star}&=std^{\circ}[\frac{1}{\sqrt{4t}}\sum _ { t=1}^ { 4 t } \sqrt{2}\cos(\frac{\pi t}{2}){\epsilon}_{t}^{\star}].\end{aligned}\ ] ] then , no matter which hypothesis is true , @xmath190 in probability as @xmath101 , where @xmath191 is a four - dimensional standard brownian motion . by the fclt given by proposition [ iid fclt ] and the proof of theorem [ aug real ] , in probability the conditional distributions of @xmath192 , @xmath193 and @xmath176
converge to the limiting distributions of @xmath106 , @xmath193 and @xmath157 , respectively .
since conditional on @xmath13 , @xmath178 is a finite - order seasonal ar process , the derivation of the conditional distributions of @xmath192 , @xmath193 and @xmath176 turns out easier than that of theorem [ aug real ] , and in particular does not involve the fourth moments of @xmath178 .
hence the consistency of the bootstrap .
suppose the assumptions in theorem [ aug real ] hold .
let @xmath194 be the probability measure corresponding to the null hypothesis @xmath195 .
then , @xmath196 @xmath197
in the proceeding section our analysis focuses on the augmented hegy test , an extension of the adf test to the seasonal unit root setting .
an important alternative of the adf test is the phillips - perron test ( phillips and perron , 1988 ) .
while the adf test assumes an ar structure over the noise and thus becomes parametric , its semi - parametric counterpart , phillips - perron test , allows a wide class of weakly dependent noises .
unaugmented hegy test ( breitung and franses , 1998 ) , as the extension of phillips - perron test to the seasonal unit root , inherits the semi - parametric nature and does not assume the noise to be ar .
given seasonal heterogeneity , it will be shown in theorem [ unaug real ] that the unaugmented hegy test estimates seasonal unit root consistently under the very general vma@xmath32 class of noise ( assumption 1.a ) , instead of the more restrictive varma@xmath33 class of noise ( assumption 1.b ) , which is needed for the augmented hegy test .
now we specify the unaugmented hegy test .
consider regression @xmath198 let @xmath156 be the ols estimator of @xmath155 , @xmath199 be the t - statistic corresponding to @xmath156 , and @xmath107 be the f - statistic corresponding to @xmath108 and @xmath109 .
other f - statistics @xmath110 , @xmath111 , @xmath112 , and @xmath113 can be defined analogously .
similar to the phillips - perron test ( phillips and perron , 1988 ) , the unaugmented hegy test can use both @xmath156 and @xmath199 when testing roots at 1 or @xmath1 . as in the augmented hegy test , we reject @xmath114 if @xmath115 ( or @xmath119 ) is too small , reject @xmath116 if @xmath117 ( or @xmath120 ) is too small , and reject the joint hypotheses if the corresponding f - statistics are too large .
the following results give the asymptotic null distributions of @xmath104 , @xmath199 , @xmath200 , and the f - statistics .
[ unaug real ] assume that assumption 1.a and one of assumption 2.a or assumption 2.b hold .
then under @xmath126 , as @xmath201 , @xmath202 where @xmath129 , @xmath130 , @xmath131 , @xmath132 , @xmath203 , @xmath134 , @xmath135 is a four - dimensional standard brownian motion , @xmath204 are defined in , @xmath205 , @xmath206 , @xmath207 , and @xmath208 .
the results in theorem [ unaug real ] degenerate to the asymptotics in burridge and taylor ( 2001ab ) when @xmath9 is uncorrelated , and degenerate to the asymptotics in breitung and franses ( 1998 ) when @xmath9 is seasonally homogeneous .
when @xmath9 is seasonally homogeneous ( breitung and franses , 1998 ) , the asymptotic distributions of @xmath209 and @xmath210 are independent . on the other hand ,
when @xmath9 has seasonal heterogeneity , @xmath209 and @xmath210 are dependent , as what we have seen for augmented hegy test ( remark [ pm1 ] ) .
hence , when testing @xmath137 , it is problematic to test @xmath114 and @xmath116 separately and calculate the level of the test with the independence of @xmath209 and @xmath210 in mind . instead , the test of @xmath137 should be handled with @xmath110 .
the parameters @xmath211 have the same definition as in theorem [ aug hegy ] . since @xmath212 , and @xmath213 , the asymptotic distributions of @xmath104 and @xmath106 , @xmath127 , only depends on the autocorrelation function of @xmath92 , the misspecified constant parameter representation of @xmath9 .
since @xmath92 can be considered as a seasonally homogeneous version of @xmath9 , we can conclude that the asymptotic behaviors of the tests for single roots at 1 or @xmath1 are not affected by the seasonal heterogeneity in @xmath9 . on the other side , the asymptotic distributions of the f - statistics
do not solely depend on @xmath92 .
hence , the test for the concurrence of roots at 1 and @xmath1 and the tests involving roots at @xmath4 are affected by the seasonal heterogeneity . to remove the nuisance parameters in the asymptotic distributions , we notice that the asymptotic behaviors of @xmath104 and @xmath106 , @xmath193 have identical forms as in phillips and perron ( 1988 ) . in light of their approach
, we can construct pivotal versions of @xmath104 and @xmath106 , @xmath193 that converge in distribution to standard dickey - fuller distributions ( dickey and fuller , 1979 ) . more specifically ,
for @xmath193 we can substitute any consistent estimator for @xmath214 and @xmath215 below : @xmath216 however , there is no easy way to construct pivotal statistics for @xmath108 , @xmath217 , @xmath109 , @xmath218 , and f - statistics such as @xmath107 .
the difficulties are two - fold .
firstly the denominators of the asymptotic distributions of these statistics contain weighted sums with unknown weights @xmath219 and @xmath220 ; secondly @xmath139 and @xmath140 are in general correlated standard brownian motions as in theorem [ aug real ] . the result in theorem
[ unaug real ] can be generalized .
suppose @xmath13 is not generated by @xmath126 , and only has some of the seasonal unit roots .
let @xmath150 , and @xmath221
. then we can find @xmath222 such that @xmath223 .
the asymptotic distributions of @xmath104 , @xmath106 , @xmath193 and the f - statistics have the same forms as those in theorem [ unaug real ] , with @xmath138 substituted by @xmath224 , and @xmath93 based on @xmath225 . as for the asymptotic results under the alternative hyphothese
, we conjecture that the powers of the unaugmented hegy tests converge to one as sample size goes to infinity . as in remark
[ power ] , we can assume without loss of generality that @xmath13 has no root at @xmath0 , @xmath1 , or @xmath4 .
then for @xmath154 , the coefficient @xmath155 corresponding to ( the misspecified constant parameter representation of ) @xmath13 are negative , according to proposition [ prop : hegy ] .
we conjecture that for @xmath154 , the ols estimators @xmath156 in converge to @xmath155 , and as a result the power of the tests tend to one .
since many of the asymptotic distributions delivered in theorem [ unaug real ] are non - standard , non - pivital , and not directly pivotable , we propose the application of bootstrap . since the regression error @xmath9 of is seasonally stationary , we in particular apply the seasonal block bootstrap of dudek et al .
the algorithm of seasonal block bootstrap unaugmented hegy test is illustrated below .
[ seasonal block boot ] step 1 : get the ols estimators @xmath115 , @xmath117 , t - statistics @xmath119 , @xmath120 , and the f - statistics @xmath157 , @xmath158 and @xmath159 from the regression of the unaugmented hegy test @xmath226 step 2 : record residual @xmath227 from regression @xmath228 step 3 : let @xmath229 , choose a integer block size @xmath230 , and let @xmath231 . for @xmath232 , let @xmath233 where @xmath234 is a sequence of iid uniform random variables taking values in @xmath235 with @xmath236 and @xmath237 + + step 4 : set the @xmath161 corresponding to the null hypothesis to be zero .
for example , set @xmath168 for all @xmath12 when testing roots at @xmath4 .
generate @xmath238 by @xmath239 step 5 : get ols estimates @xmath240 , @xmath241 , t - statistics @xmath242 , @xmath243 , and f - statistics @xmath173 from regression @xmath244 step 6 : run step 3 , 4 , and 5 for @xmath175 times to get @xmath175 sets of statistics @xmath240 , @xmath241 , @xmath242 , @xmath243 , and @xmath173 .
count separately the numbers of @xmath240 , @xmath241 , @xmath242 , @xmath243 , and @xmath173 than which @xmath115 , @xmath117 , @xmath119 , @xmath120 , and @xmath157 are more extreme .
if these numbers are higher than @xmath177 , then consider @xmath115 , @xmath117 , @xmath119 , @xmath120 and @xmath157 extreme , and reject the corresponding hypotheses .
[ sbb fclt ] let @xmath245 @xmath246 where @xmath247 , & \sigma_{2}^{*}&=std^{\circ}[\frac{1}{\sqrt{4t}}\sum _ { t=1}^ { 4 t } ( -1)^{t}v_{t}^{*}],\\ \sigma_{3}^{*}&=std^{\circ}[\frac{1}{\sqrt{4t}}\sum _ { t=1}^ { 4 t } \sqrt{2}\sin(\frac{\pi t}{2})v_{t}^ { * } ] , & \sigma_{4}^{*}&=std^{\circ}[\frac{1}{\sqrt{4t}}\sum
_ { t=1}^ { 4 t } \sqrt{2}\cos(\frac{\pi t}{2})v_{t}^{*}].\end{aligned}\ ] ] if @xmath248 , @xmath101 , @xmath249 , then no matter which hypothesis is true , @xmath250 in probability , where @xmath191 is a four - dimensional standard brownian motion . by the fclt given by proposition [ sbb fclt ] , the proof of theorem [ unaug real ] , and the convergence of the bootstrap standard deviation @xmath251 ( dudek et al . , 2014 )
, we have that the conditional distribution of @xmath252 , @xmath253 , @xmath193 and @xmath173 in probability converges to the limiting distribution of @xmath104 , @xmath106 , @xmath193 and @xmath157 , respectively .
hence the consistence of the bootstrap .
[ coro : unaug real ] suppose the assumptions in theorem [ unaug real ] hold .
let @xmath194 be the probability measure corresponding to the null hypothesis @xmath195 .
if @xmath248 , @xmath101 , @xmath249 , then @xmath254 @xmath255 @xmath256
we focus on the hypotheses test for root at 1 ( @xmath114 against @xmath257 ) , root at @xmath1 ( @xmath116 against @xmath258 ) , and root at @xmath4 ( @xmath118 against @xmath259 ) . in each hypothesis test , we equip one sequence with all nuisance unit roots at 1 , @xmath1 , and @xmath4 , and the other with none of the nuisance unit roots .
the detailed data generation processes are listed in table [ dgp ] . to produce power curves , we let parameter @xmath2600 , 0.004 , 0.008 , 0.012 , 0.016 , and 0.020 .
notice that @xmath261 is set to be seasonally homogeneous for the sake of simplicity .
further , we generate six types of innovations @xmath9 according to table [ noise ] , where @xmath262 .
the values of @xmath263 are assigned so that the misspecified constant parameter representation ( see section [ sec : settings ] ) of the `` period '' sequence has almost the same ar structure as the `` ar '' sequence .
.data generation processes [ cols="^,^,^,^ " , ] now we present in figure [ fig : root1 ] , [ fig : root-1 ] , and [ fig : rooti ] the main simulation result of the seasonal iid bootstrap augmented hegy test and the seasonal block bootstrap unaugmented hegy test .
this simulation includes two cases of nuisance roots ( see table [ dgp ] ) and six types of noises ( see table [ noise ] ) , and sets sample size @xmath264 , number of bootstrap replicates @xmath265 , number of iterations @xmath266 , and nominal size @xmath267 . when our data have a potential root at 1 , but no other nuisance roots at @xmath1 or @xmath4 , the power curves of the both bootstrap tests almost overlap , according to ( a)-(f ) in figure [ fig : root1 ] .
further , both power curves start at the correct size , @xmath267 , and tend to one when @xmath261 departs from zero .
hence both tests work well when no nuisance root occurs .
when data have a potential root at 1 and all nuisance roots at @xmath1 and @xmath4 , the sizes of seasonal block bootstrap unaugmented hegy test are distorted in ( g ) , ( h ) , ( j ) , and ( l ) in figure [ fig : root1 ] .
these distortions may result from the errors in estimating @xmath82 and the need to recover @xmath13 with the estimated @xmath82 .
the size distortion in ( j ) is particularly serious , since the unit root filter @xmath268 is partially cancelled by the moving average ( ma ) filter @xmath269 , and this cancellation can not be handled well by block bootstrap ( paparoditis and politis , 2003 ) .
in contrast , in ( l ) the filter @xmath268 is enhanced by the ar filters @xmath270 , thus the size is distorted toward zero . on the other hand ,
seasonal iid bootstrap augmented hegy test is free of the size distortions when data have nuisance roots .
this is in part because the test recovers @xmath13 using the true values of @xmath82 , namely zero , instead of using the estimated values .
moreover , when both hegy tests have almost the correct sizes as in ( i ) and ( k ) , seasonal iid bootstrap augmented hegy test attains equal or higher powers .
therefore , when testing the root at 1 , seasonal iid bootstrap augmented hegy test is recommended .
( a)-(f ) have no nuisance roots ; ( g)-(l ) have all nuisance roots ; + blue dotted curve is for seasonal iid bootstrap ; red solid curve is for seasonal block bootstrap .
[ fig : root1 ]
now we come to the tests for root at @xmath1 .
when none of the nuisance root at 1 or @xmath4 exists , the power curves of the two tests are very close to each other , as ( a)-(f ) in figure [ fig : root-1 ] indicate .
this patterns of curves have been seen in ( a)-(f ) in figure [ fig : root1 ] , and indicate the nice performance of both tests .
when nuisance roots are present , sizes of seasonal block bootstrap unaugmented hegy test are distorted in nearly all scenarios in ( g)-(l ) in figure [ fig : root-1 ] .
in particular , the size distortion in ( i ) is the worst , because of the partial cancellation of the seasonal unit root filter @xmath271 and the ma filter @xmath272 .
however , the power curves of seasonal iid bootstrap augmented hegy test start around the nominal size 0.05 in all of ( g)-(l ) .
further , these curves tend to 1 , as @xmath261 grows larger .
therefore , we recommend seasonal iid bootstrap test for testing root at @xmath1 .
( a)-(f ) have no nuisance roots ; ( g)-(l ) have all nuisance roots ; + blue dotted curve is for seasonal iid bootstrap ; red solid curve is for seasonal block bootstrap .
[ fig : root-1 ] finally we discuss the tests for roots at @xmath4 .
with none of the nuisance root at 1 or @xmath1 ,
( a)-(f ) in figure [ fig : rooti ] illustrate that both tests achieve sizes that are close to the nominal size , and powers that tend to one .
when all of nuisance roots show up , both tests suffer from some size distortions .
the empirical sizes of seasonal iid bootstrap augmented hegy test are biased toward zero in ( g)-(l ) ; the sizes of seasonal block bootstrap unaugmented hegy test are biased toward zero in ( g ) and ( h ) , but are biased toward one in ( j)-(l ) .
on the other hand , seasonal block bootstrap unaugmented hegy test s empirical powers prevail throughout ( g)-(l ) , and therefore shall be recommended for testing roots at @xmath4 .
( a)-(f ) have no nuisance roots ; ( g)-(l ) have all nuisance roots ; + blue dotted curve is for seasonal iid bootstrap ; red solid curve is for seasonal block bootstrap .
[ fig : rooti ]
in this paper we analyze the augmented and unaugmented hegy tests in the seasonal heterogeneous setting . given root at 1 or @xmath1 , the asymptotic distributions of the test statistics are standard .
however , given concurrent roots at 1 and @xmath1 , or roots at @xmath4 , the asymptotic distributions are neither standard , pivotal , nor directly pivotable .
therefore , when seasonal heterogeneity exists , hegy tests can be used to test the single roots at 1 or @xmath1 , but can not be directly applied to any combinations of roots . bootstrap proves to be an effective remedy for hegy tests in the seasonal heterogeneous setting .
the two bootstrap approaches , namely 1 ) seasonal iid bootstrap augmented hegy test and 2 ) seasonal block bootstrap unaugmented hegy test , turn out both theoretically solid . in the comparative simulation study ,
seasonal iid bootstrap augmented hegy test has better performance when testing roots at 1 or @xmath1 , but seasonal block bootstrap unaugmented hegy test outperforms when testing roots at @xmath4 . therefore , when testing seasonal unit roots under seasonal heterogeneity , the aforementioned bootstrap hegy tests become competitive alternatives of the wald - test proposed by ghysels et al .
further study will be needed to compare the theoretical and empirical efficiency of the two bootstrap hegy tests and the wald - test by ghysels et al .
the appendix includes the proof of the theorems in this paper .
we first present the proof for the asymptotics of the unaugmented hegy test , then the asymptotics of the augmented hegy test , then the consistency of the seasonal iid bootstrap augmented hegy test , and finally the consistency of the seasonal block bootstrap unaugmented hegy test . thoughout the appendix ,
let @xmath273 , @xmath274 , @xmath275 $ ] , @xmath276 be a four - dimensional standard brownian motion , @xmath277 denotes @xmath278 , and @xmath279 denotes @xmath280 .
see hamilton ( 1994 , proposition 18.1 , pp .
547 - 548 ) for the proof with iid innovations , chan and wei ( 1988 ) for the proof under assumption 2.a , and de jong and davidson ( 2000 ) for the proof under assumption 2.b .
[ le : unaug real2 ] let @xmath282 , and @xmath283 , where @xmath284 stands for unaugmented hegy , and @xmath285 is defined in .
let @xmath286 , @xmath287 be the matrix generated by assigning zero to all entries of @xmath288 but those above the main diagonal . then , under @xmath126
, @xmath289 for the proof of part ( a ) , see the lemma 3.2(a ) of burridge and taylor ( 2001a ) and its proof . for part ( b ) , we only present the proof of the first statement .
other statements are proven in similar ways . by lemma [ le : unaug real1 ] , @xmath290 let @xmath291 , @xmath292 , @xmath293 , and @xmath294
. then @xmath295^{-1}(\xi_{1},\xi_{2},\xi_{3},\xi_{4 } ) ' \text { by lemma \ref{le : unaug real2}},\\ \hat{\sigma}^{2 } & = ( 4t)^{-1}(\bm{v}'\bm{v}+2({\bm{x}}_{u}\bm{\hat{\pi}}-{\bm{x}}_{u}\bm{\pi})'(\bm{v}-{\bm{x}}_{u}\bm{\pi})\\ & \mathrel{\phantom{= } } + ( { \bm{x}}_{u}\bm{\hat{\pi}}-{\bm{x}}_{u}\bm{\pi})'({\bm{x}}_{u}\bm{\hat{\pi}}-{\bm{x}}_{u}\bm{\pi}))\\ & = ( 4t)^{-1}\bm{v}'\bm{v}+o_{p}(1 ) \text { by the consistency of } \bm{\hat{\pi}}\\ & \stackrel{p}\rightarrow \text{tr}(\bm{\gamma}_{0})/4,\\ \bm{t}&=\hat{\sigma}^{-1}[\text{diag}({\bm{x}}_{u}'{\bm{x}}_{u})^{-1}]^{-1/2}({\bm{x}}_{u}'{\bm{x}}_{u})^{-1}{\bm{x}}_{u}'\bm{v}\\ & \rightarrow ( \text{tr}(\bm{\gamma}_{0})/4)^{-1/2}[\text{diag}(\eta_{1},\eta_{2},\eta_{3},\eta_{4})]^{-1/2}(\xi_{1},\xi_{2},\xi_{3},\xi_{4})'.\end{aligned}\ ] ] further , the asymptotic distributions of f - statistics are identical with the asymptotic distributions of the averages of the squares of the corresponding t - statsitics , i.e. , @xmath296 , due to the asymptotic orthogonality indicated by lemma [ le : unaug real2 ] ( a ) .
the proof follows the lines of said and dickey ( 1984 ) and contains two parts .
firstly , we show when @xmath297 and @xmath298 simultaneously , the statistic of interest tends to a limit free of @xmath48 , and then we prove this limit tends to a certain distribution as @xmath297 . to begin with ,
notice that when @xmath102 , the error term of regression tends to a limit .
surprisingly , this limit is in general not @xmath299 , because the regression falsely assumes seasonally homogeneous coefficients and thus in general can not find the correct residuals @xmath299 . to find the limit , recall that @xmath92 is defined as a misspecified constant parameter representation of @xmath9 . under assumption 1.b , the spectral densities of @xmath92 are finite and positive everywhere , so @xmath92 has ar@xmath32 and ma@xmath32 expressions @xmath300 where @xmath301 , @xmath302 .
let @xmath303 , and @xmath304 , where @xmath305 are the ar coefficients defined in .
since a misspecified constant parameter representation of @xmath306 is @xmath307 , which is exactly @xmath308 defined in , no ambiguity arises .
the following lemma provides two properties of @xmath309 , whose proof is left to the readers .
now we show when @xmath101 and @xmath311 simultaneously , the statistics of interest tend to certain limits .
let @xmath312 be the design matrix of regression equation , @xmath313 be the estimated coefficient vector of regression equation , @xmath314 , @xmath315 , and @xmath316 .
define the @xmath317 dimensional scaling matrix @xmath318 . then @xmath319 let @xmath320 be the @xmath321 induced norm of matrices . now we want to define a diagonal matrix @xmath322 such that @xmath323 converges to 0 in probability . by the multivariate beveridge - nielson decomposition ( see hamilton , 1994 , pp .
545 - 546 ) , since @xmath324 converges in probability to the seasonal average of autocovariance of @xmath325 of lag @xmath326 , we let @xmath327 where @xmath328 following the definition of @xmath322 , we make the following decomposition : @xmath329\bm{d}_{t}\bm{x}'\bm{\zeta}^{(k)}\\ & + \bm{r}^{-1}\bm{d}_{t}\bm{x}'(\bm{\zeta}^{(k)}-\bm{\zeta)}\\ & + \bm{r}^{-1}\bm{d}_{t}\bm{x}'\bm{\zeta}. \end{aligned } \label{said decomposition}\ ] ] notice the last term in the right hand side summation , @xmath330 , is free of @xmath48 .
later we will find out its asymptotic distribution as @xmath201 .
but now we need to prove the first two terms in the right hand side of converge to zero as @xmath201 and @xmath311 .
indeed , @xmath331.\end{aligned}\ ] ] notice that @xmath332 . under assumption 1.b
, @xmath9 is a varma sequence with finite orders , thus @xmath92 also has an arma expression with finite orders ( see osborn , 1991 ) , @xmath333 hence , @xmath334 has exponentially decaying coefficient @xmath335 .
it follows straightforwardly that @xmath336 . for , notice that @xmath337.\ ] ] by lemma [ zeta ] and the stationarity of @xmath61 , @xmath338\\ & = \frac{1}{4}\sum_{s=-3}^{0}\sum_{h=-\infty}^{\infty}cov(v_{4t+s - i}\zeta_{4t+s},v_{4t+s - h - i}\zeta_{4t+s - h})+o(1)\\ & = \frac{1}{4}\sum_{s=-3}^{0}\sum_{h=-\infty}^{\infty}cov(v_{s - i}\zeta_{s},v_{s - h - i}\zeta_{s - h})+o(1).\end{aligned}\ ] ] without loss of generality we can focus on @xmath339 and @xmath340 . by writing @xmath325 and @xmath306 as linear combinations of @xmath299 , @xmath341 the right hand side of this inequality is assumed to be bounded under assumption 2.a . on the other hand ,
the right hand side is also bounded under assumption 2.b , by the lemma below .
[ boundedness ] suppose ( i ) @xmath342 is a strictly stationary strong mixing time series with mean zero and finite @xmath62 moment for some @xmath63 , and ( ii ) @xmath343 s strong mixing coefficient @xmath344 satisfies @xmath345
. then @xmath64 such that for all @xmath346 and @xmath347 , @xmath348 we have proven that @xmath354=o(1)$ ] .
similarly , it can be shown that @xmath355=o(1)$ ] .
hence , follows . to justify , notice @xmath356 where @xmath357 indicates standard four - dimensional brownian motion . since @xmath358 @xmath359 such that @xmath360
. follows from the definition of @xmath361 . combining equations , , , and
, we have @xmath362 .
notice @xmath363 by .
the consistency follows from @xmath364 and @xmath122 .
further , the asymptotic distribution of @xmath365 can be derived with the asymptotic equivalence of @xmath366 and @xmath330 .
notice @xmath330 is free of @xmath48 .
as @xmath367 , @xmath368 converges in distribution to a functional of brownian motion , and the asymptotics of @xmath369 can be found with the following lemma .
@xmath371^{2}}{\frac{1}{4}({\bm{c}}_{4}'{\bm{\theta}}(1){\bm{\omega}}{\bm{\theta}}(1)'{\bm{c}}_{4}+{\bm{c}}_{3}'{\bm{\theta}}(1){\bm{\omega}}{\bm{\theta}}(1)'{\bm{c}}_{3})}\\ & \mathrel{\phantom{\rightarrow}}+\frac { var({\tilde{\zeta}}_{t})[\sqrt{\frac{1}{4}{\bm{c}}_{4}'{\bm{\theta}}(1){\bm{\omega}}{\bm{\theta}}(1)'{\bm{c}}_{4}\frac{1}{4}{\bm{c}}_{3}'{\bm{\theta}}(1){\bm{\omega}}{\bm{\theta}}(1)'{\bm{c}}_{3}}(\int_{0}^{1 } w_{3}(r)d w_{4}(r)-\int w_{4}(r)dw_{3}(r))]^{2}}{\frac{1}{4}({\bm{c}}_{4}'{\bm{\theta}}(1){\bm{\omega}}{\bm{\theta}}(1)'{\bm{c}}_{4}+{\bm{c}}_{3}'{\bm{\theta}}(1){\bm{\omega}}{\bm{\theta}}(1)'{\bm{c}}_{3})}.\end{aligned}\ ] ] firstly we focus on the convergence of @xmath372 .
the convergence of @xmath373 can be proven analogously .
define @xmath374 such that @xmath375 .
let @xmath376 , @xmath377 , @xmath378 , @xmath379 .
then @xmath380 , and @xmath381\\ & \rightarrow\frac{1}{4}{\tilde{\theta}}(1){\bm{c}}_{1}'{\tilde{\bm{\psi}}}(1){\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\tilde{\bm{\psi}}}(1)'{\bm{c}}_{1 } \ \text{(by lemma \ref{zeta } and
fclt)}\\ & \mathrel{\phantom{\rightarrow}}+\frac{1}{4}{\tilde{\theta}}(1)[\sum_{s=-3}^{0}\sum_{k=-3}^{s-1}e\zeta_{4t+k}\zeta_{4t+s}+{\bm{c}}_{1}'\sum_{i=1}^{\infty}e{\bm{\zeta}}_{t - i}{\bm{\zeta}}_{t}'{\bm{c}}_{1}]\\ & = \frac{1}{4}{\tilde{\theta}}(1){\bm{c}}_{1}'{\tilde{\bm{\psi}}}(1){\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\tilde{\bm{\psi}}}(1)'{\bm{c}}_{1 } \ \text{(since } \{{\tilde{\zeta}}_{t}\ } \text { is white noise)}\\ & = \frac{1}{4}{\tilde{\theta}}(1)({\tilde{\psi}}(1))^{2}{\bm{c}}_{1}'{\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\bm{c}}_{1 } \ ( \text{since } { \bm{c}}_{1}'{\tilde{\bm{\psi}}}(1)={\tilde{\psi}}(1){\bm{c}}_{1}')\\ & = var({\tilde{\zeta}}_{t})\tilde{\theta}(1)\int_{0}^{1 } w_{1}(r)dw_{1}(r)\\ & \text{(by osborn ( 1991 , p. 378 ) , } \frac{1}{4}{\bm{c}}_{1}'{\bm{\theta}}(1){\bm{\omega}}{\bm{\theta}}(1)'{\bm{c}}_{1}= var({\tilde{\zeta}}_{t})\tilde{\theta}(1)^{2}).\end{aligned}\ ] ] secondly we show the convergence of @xmath382 .
let @xmath383 , @xmath384 , @xmath385 , @xmath386 , and @xmath387 .
then @xmath388\\ & \mathrel{\phantom{=}}-\frac{1}{4t}\sum_{t=1}^{t}{\tilde{\theta}}_{b}[{\bm{c}}_{3}'{\bm{\xi}}_{t-1}{\bm{\zeta}}_{t}'{\bm{c}}_{4}-{\bm{c}}_{4}'{\bm{\xi}}_{t-1}{\bm{\zeta}}_{t}'{\bm{c}}_{3}-\sum_{s=-3}^{-1}\zeta_{4t+s}\zeta_{4t+s+1}+\zeta_{4t-3}\zeta_{4t}]\\ & \rightarrow \frac{1}{4}{\tilde{\theta}}_{a}[{\bm{c}}_{3}'{\tilde{\bm{\psi}}}(1){\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\tilde{\bm{\psi}}}(1)'{\bm{c}}_{3}\\ & \mathrel{\phantom{\rightarrow \frac{1}{4}{\tilde{\theta}}_{a}[}}+{\bm{c}}_{4}'{\tilde{\bm{\psi}}}(1){\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\tilde{\bm{\psi}}}(1)'{\bm{c}}_{4}]\\ & \mathrel{\phantom{=}}-\frac{1}{4}{\tilde{\theta}}_{b}[{\bm{c}}_{3}'{\tilde{\bm{\psi}}}(1){\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\tilde{\bm{\psi}}}(1)'{\bm{c}}_{4}\\ & \mathrel{\phantom{\mathrel{\phantom{=}}-\frac{1}{4}{\tilde{\theta}}_{b}[}}-{\bm{c}}_{4}'{\tilde{\bm{\psi}}}(1){\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\tilde{\bm{\psi}}}(1)'{\bm{c}}_{3}]\\ & ( \text{by lemma \ref{zeta } and fclt , the covariances of } \zeta_{t } \text { cancel out since } \{{\tilde{\zeta}}_{t}\ } \text { is white noise})\\ & = \frac{1}{4}{\tilde{\theta}}_{a}|{\tilde{\psi}}(i)|^{2}[{\bm{c}}_{4}'{\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\bm{c}}_{4}+{\bm{c}}_{3}'{\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\bm{c}}_{3}]\\ & \mathrel{\phantom{=}}-\frac{1}{4}{\tilde{\theta}}_{b}|{\tilde{\psi}}(i)|^{2}[{\bm{c}}_{3}'{\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\bm{c}}_{4}-{\bm{c}}_{4}'{\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\bm{c}}_{3}]\\ & ( \text{since } { \bm{c}}_{3}'{\tilde{\bm{\psi}}}(1)={\tilde{\psi}}_{b}{\bm{c}}_{4}'+{\tilde{\psi}}_{a}{\bm{c}}_{3 } ' , \ { \bm{c}}_{4}'{\tilde{\bm{\psi}}}(1)={\tilde{\psi}}_{a}{\bm{c}}_{4}'-{\tilde{\psi}}_{b}{\bm{c}}_{3 } ' , \ \text{and } { \tilde{\psi}}_{a}^{2}+{\tilde{\psi}}_{b}^{2}=|{\tilde{\psi}}(i)|^{2}).\end{aligned}\ ] ] similarly , @xmath389\\ & \mathrel{\phantom{=}}+\frac{1}{4}{\tilde{\theta}}_{a}|{\tilde{\psi}}(i)|^{2}[{\bm{c}}_{3}'{\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\bm{c}}_{4}-{\bm{c}}_{4}'{\bm{\theta}}(1){\bm{\omega}}^{1/2}\int { \bm{w}}d { \bm{w } } ' { \bm{\omega}}^{1/2}{\bm{\theta}}(1)'{\bm{c}}_{3}]\end{aligned}\ ] ] the lemma follows from @xmath390 and ( osborn , 1991 ) @xmath391 now we come to the asymptotic distribution of the t - statistics and the f - statistics .
notice , @xmath392_{ii}]^{-1/2}[(\bm{x}'\bm{x})^{-1}\bm{x}'\bm{\zeta}^{(k)}]_{i}\\ & = \hat{\sigma}^{-1}[[[(4t - k)^{-2}(\bm{x}'\bm{x})^{-1}]_{ii}]^{-1/2}-[[\bm{r}^{-1}]_{ii}]^{-1/2}](4t - k)[(\bm{x}'\bm{x})^{-1}\bm{x}'\bm{\zeta}^{(k)}]_{i}\\ & \mathrel{\phantom{=}}+\hat{\sigma}^{-1}[[\bm{r}^{-1}]_{ii}]^{-1/2}((4t - k)(\bm{x}'\bm{x})^{-1}\bm{x}'\bm{\zeta}^{(k)}-\bm{r}^{-1}(4t - k)^{-1}\bm{x}'{\bm{\zeta}})_{i}\\ & \mathrel{\phantom{=}}+\hat{\sigma}^{-1}[[\bm{r}^{-1}]_{ii}]^{-1/2}(\bm{r}^{-1}(4t - k)^{-1}\bm{x}'{\bm{\zeta}})_{i}\\ & = \hat{\sigma}^{-1}[[\bm{r}^{-1}]_{ii}]^{-1/2}(\bm{r}^{-1}(4t - k)^{-1}\bm{x}'{\bm{\zeta}})_{i}+o_{p}(1).\end{aligned}\ ] ] by the consistency of @xmath365 , we have @xmath393 .
the asymptotic distributions of the t - statistics follows straightforwardly from lemma [ asy zeta ] .
further , the asymptotic distributions of the f - statistics are identical with the asymptotic distributions of the averages of the squares of the corresponding t - statistics because of the asymptotic orthogonality of the regression .
hence , the proof of theorem [ aug real ] is complete .
define @xmath394 and @xmath234 such that @xmath395 and @xmath396 . by algorithm [ seasonal iid bootstrap ]
, @xmath394 is a sequence of independent but not identical random variables , while @xmath234 is a sequence of iid random variables .
recall @xmath397 where @xmath398 is the regression error .
let @xmath399 let @xmath400 be the partial sum of @xmath401 above .
formally , @xmath402 to justify theorem [ iid fclt ] , it suffices to show @xmath403 @xmath404 because the unconditional convergence in implies that in probability the conditional distribution of @xmath405 given @xmath13 converges to zero . to prove
, we can without loss of generality focus on the uniform convergence of the first coordinate , that is , uniformly in @xmath406 , @xmath407 notice that uniformly in @xmath406 , @xmath408 where @xmath409 and @xmath410 have obvious definitions .
now we show @xmath411 , and @xmath412 , uniformly in @xmath406 . for @xmath409 ,
notice if @xmath413 , then @xmath285 is weakly stationary , so @xmath414 is @xmath415 ( see berk , 1974 ) , and it follows straightforwardly that @xmath411 uniformly in @xmath406 .
on the other hand , if @xmath416 , then by theorem [ aug real ] , @xmath417 .
let @xmath418 it suffices to show that @xmath419 .
by continuous mapping theorem , it suffices to prove @xmath420 , where @xmath421 .
it is straightforward to show the weak convergence of the finite dimensional distributions of @xmath422 .
furthermore , @xmath422 is tight , since ( see billingsley , 1999 , pp .
146 - 147 ) @xmath423 , @xmath424\\ = & e[var^{\circ}[\frac{q_{t}(r_{2})}{t}-\frac{q_{t}(r)}{t}]var^{\circ}[\frac{q_{t}(r)}{t}-\frac{q_{t}(r_{1})}{t}]]\rightarrow 0 .
\end{aligned } \label{tightness}\ ] ] hence @xmath420 , and @xmath411 uniformly in @xmath406 follows . for @xmath410 , in light of the derivation of theorem [ aug real ] , it can be shown that @xmath425 holds not only under alternative hypotheses but also under the null .
hence , it follows that uniformly in @xmath406 , @xmath412 .
therefore , recalling , we have @xmath426 further , it is straightforward to show @xmath427\stackrel{p}\rightarrow 0 $ ] , and @xmath428\stackrel{p}\rightarrow 0 $ ] . using the same decomposition as in , @xmath429 .
hence we have proven .
secondly we prove , notice that the standard deviations in the definition of @xmath430 are bounded in probability .
for example , @xmath431 further , given @xmath13 , for fixed @xmath27 , @xmath432 are conditionally iid random variables .
finally , for all @xmath433 , @xmath434\stackrel{p}\rightarrow u,\ ] ] @xmath435 the convergence @xmath400 of to @xmath436 follows by generalizing ( see kreiss and paparoditis , 2015 ) the real world result of helland ( 1982 , theorem 3.3 ) to the bootstrap world . without loss of generality ,
assume block size @xmath230 is a multiple of four .
let @xmath437 .
then the @xmath438th block of @xmath439 starts from @xmath440 .
let @xmath441 be the rescaled aggregation of the @xmath438th block , defined by @xmath442 let @xmath443 be the partial sum of the block aggregations above .
formally , @xmath444 to prove theorem [ sbb fclt ] , it suffices to show @xmath445 @xmath446 where @xmath320 denotes the @xmath321 norm . to show , without loss of generality we focus on the uniform convergence of the first coordinate , that is , uniformly in @xmath406 , @xmath447 .
notice that , @xmath448 where @xmath449 denotes the total number of the blocks , and @xmath450 is the length of the @xmath438th block .
it suffices to only consider the first term in , since @xmath451 by the definition of @xmath452 , @xmath453 now we show the second term on the right hand side of the equation above converges uniformly in @xmath406 to 0 in probability .
here we only present the result for j=1 , s=0 .
notice if @xmath454 for some @xmath12 , then @xmath455 .
hence , the result follows the weakly stationarity of the vector sequence @xmath456 . on the other hand , if @xmath457 for all @xmath12 , then @xmath458 .
hence , we only need to show that @xmath459 has @xmath460 , where @xmath461 . the convergence of finite dimensional distribution of @xmath462 can be proven by the line of politis and paparaditis ( 2003 , p. 841 ) .
furthermore , it can be shown that @xmath462 is tight using .
hence @xmath460 . therefore , @xmath463 uniformly in @xmath406 .
since it is straightforward to show @xmath464)| \stackrel { p } \rightarrow 0\ ] ] uniformly in @xmath406 , and @xmath465 ) -\frac{1}{\sqrt{4t}}\sum_{m=1}^{\lfloor lu_{1 } \rfloor}\sum_{h=1}^{b}(v_{i_{m}+h-1}-e^{\circ}[v_{i_{m}+h-1}])| \stackrel { p } \rightarrow 0\ ] ] uniformly in @xmath406 , we have obtained that @xmath466)| \stackrel { p } \rightarrow 0 \label{sbb fclt 2}\ ] ] uniformly in @xmath406 .
now we show that @xmath467 -var^{\circ}[\frac{1}{\sqrt{b}}\sum_{h=1}^{b}v_{i_{m}+h-1}]\stackrel { p}\rightarrow 0 $ ] .
notice , @xmath468 ) + \frac{1}{\sqrt{4t}}\sum_{m=1}^{l}\sum_{s=-3}^{0}\sum_{t=1}^{b/4}(e^{\circ}v_{i_{m}+4t+s-1}-\frac{1}{t}\sum_{t=1}^{t}v_{4t+s})\\ & \mathrel{\phantom{=}}-\sum_{j=1}^{4}\sum_{s=-3}^{0}(\hat{\pi}_{j , s}-\pi_{j , s})\frac{1}{\sqrt{4t}}\sum_{m=1}^{l}\sum_{t=1}^{b/4}(j , y_{i_{m}+4t+s-1}-\frac{1}{t}\sum_{t=1}^{t}y_{4t+s})\\ & = a_{t}+b_{t}-\sum_{j=1}^{4}c_{t , j}\end{aligned}\ ] ] where @xmath469 and @xmath470 have obvious definitions .
it is straightforward to show @xmath471\stackrel { p } \rightarrow 0 $ ] , @xmath472\stackrel { p } \rightarrow 0 $ ] for @xmath27 , and @xmath473=var^{\circ}[\frac{1}{\sqrt{b}}\sum_{h=1}^{b}v_{i_{m}+h-1}]$ ] .
hence , we have @xmath474 -var^{\circ}[\frac{1}{\sqrt{b}}\sum_{h=1}^{b}v_{i_{m}+h-1}]\stackrel { p}\rightarrow 0 .
\label{sbb fclt 3}\ ] ] by and , we have shown @xmath475 uniformly in @xmath406 , and thus @xmath476 uniformly in @xmath406 , @xmath477 , @xmath478 and @xmath479 . secondly we prove . given assumption b.1 , it is sufficient to show that the following three properties hold : @xmath480 \stackrel { p } \rightarrow u , \ : \forall u\geq 0 , \ : and \ : \forall \ : i=1, ... ,4 , \label{sbb fclt3 } \\ &
\sum\limits_{m=1}^{\lfloor lu \rfloor}e^{\circ}[{\upsilon}_{l , m}^{(i)^{2}}1(|{\upsilon}_{l , m}|>\epsilon ) ] \stackrel { p } \rightarrow 0 , \ : \forall u\geq 0 , \ : \ : \forall \ : i=1, ... ,4 , \label{sbb fclt4 } \\ & \sum\limits_{m=1}^{\lfloor lu \rfloor}e^{\circ}[{\upsilon}_{l , m}^{(i)}{\upsilon}_{l , m}^{(j ) } ] \stackrel { p } \rightarrow 0 , \ : \forall u\geq 0 , \ \forall
\ : i , j \in \{1,2,3,4\ } , \
: i\neq j. \label{sbb fclt5}\end{aligned}\ ] ] helland ( 1982 ) shows that if @xmath481 is a martingale difference array and the above three properties hold in real world , then @xmath482 . by beveridge - neilson decomposition ( hamilton , 1994 , proposition 17.2 , p. 504 )
, helland s result can be generalized to the case when @xmath481 is a convolution of a constant array and a martingale difference array .
further , helland s result can be generalized to the bootstrap world ( see kreiss and paparoditis , 2015 ) . hence the sufficiency of the three properties above . to verify and ,
notice that @xmath483 , @xmath484 , @xmath485=\lfloor lt
\rfloor / l \rightarrow t,\ ] ] and , by the dominated convergence theorem , @xmath486 \stackrel { p } \rightarrow 0.\ ] ] hence , it remains to verify the , which indicates asymptotic independence between coordinates of @xmath443 .
note that the third property need to be proved for all @xmath487 . here
we cite as an example the case @xmath339 and @xmath488 .
the rest of cases can be shown by similar calculations .
notice , @xmath489 = & \frac{e^{\circ}[\frac{1}{\sqrt{b}}\sum_{h=1}^{b}v_{i_{1}+h-1}\frac{1}{\sqrt{b}}\sum_{r=1}^{b}\sqrt{2}\sin(\pi r/2)v_{i_{1}+r-1 } ] } { std^{\circ}[\frac{1}{\sqrt{b}}\sum_{h=1}^{b}v_{i_{1}+h-1}]std^{\circ}[\frac{1}{\sqrt{b}}\sum_{r=1}^{b}\sqrt{2}\sin(\pi r/2)v_{i_{1}+r-1}]}\\ & -\frac{e^{\circ}[\frac{1}{\sqrt{b}}\sum_{h=1}^{b}v_{i_{1}+h-1}]e^{\circ}[\frac{1}{\sqrt{b}}\sum_{r=1}^{b}\sqrt{2}\sin(\pi r/2)v_{i_{1}+r-1}]}{std^{\circ}[\frac{1}{\sqrt{b}}\sum_{h=1}^{b}v_{i_{1}+h-1}]std^{\circ}[\frac{1}{\sqrt{b}}\sum_{r=1}^{b}\sqrt{2}\sin(\pi r/2)v_{i_{1}+r-1}]}.\end{aligned}\ ] ] since @xmath490\stackrel { p } \rightarrow 0,\ : e^{\circ}[\frac{1}{\sqrt{b}}\sum_{r=1}^{b}\sqrt{2}\sin(\pi r/2)v_{i_{1}+r-1}]\stackrel { p } \rightarrow 0,\ ] ] and both @xmath491 $ ] and @xmath492 $ ] converge in probability to constants ( dudek et al . , 2014 ) , we only need to show that @xmath493 \stackrel { p } \rightarrow 0.\ ] ] notice , @xmath494\\ & = \frac{\sqrt{2}}{b(t - b/4)}\sum_{i=1}^{t - b/4}\sum_{h=1}^{b}\sum_{r=1}^{b}\sin(\pi r/2)v_{4i+h-4}v_{4i+r-4}\\ & = -a+b+o_{p}(1),\end{aligned}\ ] ] where @xmath495 the proof under assumption [ assump 1b ] is complete after showing @xmath496 by lemma [ martingale difference ] below .
now consider assumption [ assump 2b ] .
let @xmath497 .
let ( @xmath498,@xmath499,@xmath500,@xmath501 be the eigenvalues of @xmath502 .
it is sufficient ( wooldridge and white , 1988 , corollary 4.2 ) to show that the following two properties hold : @xmath503 notice , to show , it suffices to show , which is ensured by lemma [ boundedness ] and lemma [ martingale difference ] .
equation follows from the continuity of the eigenvalue function .
hence we have completed the proof when block size @xmath230 is a multiple of four .
when @xmath230 is not a multiple of four , it is straightforward to show . for
, let @xmath504 since @xmath505 , @xmath6 are mutually independent with respect to @xmath182 , and @xmath506 in probability for all @xmath11 , we have @xmath507 in probability .
[ martingale difference ] suppose ( i ) @xmath342 is a fourth - order stationary time series with finite @xmath62 moment for some @xmath63 .
( ii ) @xmath64 , @xmath65 @xmath2 , @xmath47 , @xmath48 , and @xmath66 , @xmath508
. suppose @xmath509 and @xmath510 .
then , @xmath511\rightarrow 0.\ ] ] @xmath512\\ & = \frac{1}{b^{2}n^{2}}\sum_{t_1=1}^{n}\sum_{t_2=1}^{n}\sum_{j_1=1}^{b}\sum_{j_2=1}^{b}cov[z_{0}z_{-j_1},z_{t_2-t_1}z_{t_2-t_1-j_1}]\\ & = \frac{1}{b^{2}n^{2}}\sum_{h=1-n}^{n-1}(n-|h|)\sum_{j_1=1}^{b}\sum_{j_2=1}^{b}cov[z_{0}z_{-j_1},z_{h}z_{h - j_1}]\\ & < \frac{k}{n}\rightarrow 0 . & \mbox{\qedhere}\end{aligned}\ ] ] 99 berk , kenneth n. `` consistent autoregressive spectral estimates . ''
the annals of statistics ( 1974 ) : 489 - 502 .
billingsley , patrick .
convergence of probability measures .
new jersey : john wiley & sons , 1999 .
breitung , jorg , and philip hans franses .
`` on phillips - perron - type tests for seasonal unit roots . '' econometric theory 14.02 ( 1998 ) : 200 - 221 .
boswijk , h. peter , philip hans franses , and niels haldrup .
`` multiple unit roots in periodic autoregression . ''
journal of econometrics 80.1 ( 1997 ) : 167 - 193 .
burridge , peter , and am robert taylor .
`` on regression - based tests for seasonal unit roots in the presence of periodic heteroscedasticity . '' journal of econometrics 104.1 ( 2001a ) : 91 - 117 .
burridge , peter , and am robert taylor .
`` on the properties of regression - based tests for seasonal unit roots in the presence of higher - order serial correlation . '' journal of business and economic statistics 19 , no .
3 ( 2001b ) : 374 - 379 .
burridge , peter , and am robert taylor .
`` bootstrapping the hegy seasonal unit root tests . ''
journal of econometrics 123.1 ( 2004 ) : 67 - 87 .
chan , ngai hang , and c. z. wei .
`` limiting distributions of least squares estimates of unstable autoregressive processes . '' the annals of statistics ( 1988 ) : 367 - 401 . de jong , robert m. , and james davidson . `` the functional central limit theorem and weak convergence to stochastic integrals i. '' econometric theory 16.05 ( 2000 ) : 621 - 642 . del barrio castro , toms , and denise r. osborn .
`` testing for seasonal unit roots in periodic integrated autoregressive processes . ''
econometric theory 24.04 ( 2008 ) : 1093 - 1129 .
del barrio castro , toms , and denise r. osborn .
`` hegy tests in the presence of moving averages*. '' oxford bulletin of economics and statistics 73.5 ( 2011 ) : 691 - 704
. del barrio castro , toms , denise r. osborn , and am robert taylor . `` on augmented hegy tests for seasonal unit roots . ''
econometric theory 28.05 ( 2012 ) : 1121 - 1143 .
del barrio castro , toms , denise r. osborn , and am robert taylor . `` the performance of lag selection and detrending methods for hegy seasonal unit root tests .
'' econometric reviews 35.1 ( 2016 ) : 122 - 168 .
dickey , david a. , and wayne a. fuller .
`` distribution of the estimators for autoregressive time series with a unit root . '' journal of the american statistical association 74.366a ( 1979 ) : 427 - 431 .
dudek , anna e. , jacek lekow , efstathios paparoditis , and dimitris n. politis .
`` a generalized block bootstrap for seasonal time series . ''
journal of time series analysis 35.2 ( 2014 ) : 89 - 114 .
dudek , anna e. , efstathios paparoditis , and dimitris n. politis .
`` generalized seasonal tapered block bootstrap . '' statistics and probability letters ( 2016 ) .
franses , philip hans .
`` a multivariate approach to modeling univariate seasonal time series . ''
journal of econometrics 63.1 ( 1994 ) : 133 - 151 .
franses , philip hans , and richard paap . periodic time series models . oxford university press , oxford .
galbraith , john w , and victoria zinde - walsh .
`` on the distributions of augmented dickey - fuller statistics in processes with moving average components . '' journal of econometrics 93.1 ( 1999 ) : 25 - 47 .
ghysels , eric , alastair hall , and hahn shik lee .
`` on periodic structures and testing for seasonal unit roots .
'' journal of the american statistical association 91.436 ( 1996 ) : 1551 - 1559 .
ghysels , eric , and denise r. osborn . the econometric analysis of seasonal time series .
cambridge university press , 2001 .
hamilton , james douglas .
time series analysis .
princeton : princeton university press , 1994 .
hylleberg , svend , robert f. engle , clive wj granger , and byung sam yoo .
`` seasonal integration and cointegration . ''
journal of econometrics 44.1 ( 1990 ) : 215 - 238 .
helland , inge s. `` central limit theorems for martingales with discrete or continuous time . ''
scandinavian journal of statistics ( 1982 ) : 79 - 94 .
johansen , sren .
`` statistical analysis of cointegration vectors . ''
journal of economic dynamics and control 12.2 ( 1988 ) : 231 - 254 .
kreiss , j .-
p . and paparoditis ,
e. bootstrap methods for time series , manuscript ( in progress ) , 2015 .
osborn , denise r. `` the implications of periodically varying coefficients for seasonal time - series processes . ''
journal of econometrics 48.3 ( 1991 ) : 373 - 384 .
paparoditis , efstathios , and dimitris n. politis .
`` residualbased block bootstrap for unit root testing . ''
econometrica 71.3 ( 2003 ) : 813 - 855 .
paparoditis , efstathios , and dimitris n. politis .
`` the asymptotic size and power of the augmented dickey - fuller test for a unit root . '' econometric reviews just - accepted ( 2016 ) .
park , joon y. `` bootstrap unit root tests . ''
econometrica 71.6 ( 2003 ) : 1845 - 1895 .
phillips , peter cb , and pierre perron .
`` testing for a unit root in time series regression . ''
biometrika 75.2 ( 1988 ) : 335 - 346 .
politis , dimitris n. , joseph p. romano , and michael wolf .
`` subsampling . ''
springer , new york .
said , said e. , and david a. dickey .
`` testing for unit roots in autoregressive - moving average models of unknown order . ''
biometrika 71.3 ( 1984 ) : 599 - 607 .
wooldridge , jeffrey m. , and halbert white .
`` some invariance principles and central limit theorems for dependent heterogeneous processes . '' econometric theory 4.02 ( 1988 ) : 210 - 230 . | both seasonal unit roots and seasonal heterogeneity are common in seasonal data . when testing seasonal unit roots under seasonal heterogeneity , it is unclear if we can apply tests designed for seasonal homogeneous settings , i.e. the hegy test ( hylleberg , engle , granger , and yoo , 1990 ) . in this paper ,
the validity of both augmented hegy test and unaugmented hegy test is analyzed .
the asymptotic null distributions of the statistics testing the single roots at @xmath0 or @xmath1 turn out to be standard and pivotal , but the asymptotic null distributions of the statistics testing any coexistence of roots at @xmath0 , @xmath1 , @xmath2 , or @xmath3 are non - standard , non - pivotal , and not directly pivotable .
therefore , the hegy tests are not directly applicable to the joint tests for the concurrence of the roots . as a remedy ,
we bootstrap augmented hegy with seasonal independent and identically distributed ( iid ) bootstrap , and unaugmented hegy with seasonal block bootstrap .
the consistency of both bootstrap procedures is established .
simulations indicate that for roots at @xmath0 and @xmath1 seasonal iid bootstrap augmented hegy test prevails , but for roots at @xmath4 seasonal block bootstrap unaugmented hegy test enjoys better performance .
* keywords : * seasonality , unit root , ar sieve bootstrap , block bootstrap , functional central limit theorem . |
one of the challenges in the modern physics is the description of the internal structure of the baryons and mesons .
the electromagnetic structure of the nucleon @xmath1 and the nucleon resonances @xmath2 can be accessed through the @xmath0 reactions , which depend of the ( photon ) momentum transfer squared @xmath13 @xcite .
the data associated with those transitions are represented in terms of helicity amplitudes and have been collected in the recent years at jefferson lab , with increasing @xmath13 @xcite .
the new data demands the development of theoretical models based on the underlying structure of quarks and quark - antiquark states ( mesons ) @xcite .
those models may be used to guide future experiments as the ones planned for the jlab12 gev upgrade , particularly for resonances in the second and third resonance region [ energy @xmath14@xmath15 gev ] ( see figure [ figsigmaw ] ) @xcite . in that region
there are several resonances @xmath2 from the multiplet @xmath4 $ ] of @xmath16 , characterized by a negative parity @xcite . according with the single quark transition model ( sqtm ) , when the electromagnetic interaction is the result of the photon coupling with just one quark , the helicity amplitudes of the @xmath4 $ ] members depend only on three independent functions of @xmath13 : @xmath17 and @xmath18 @xcite . in this work we use the covariant spectator quark model @xcite developed for the @xmath19 and @xmath20 transitions , also members of @xmath4 $ ] , to calculate those functions @xcite . since the covariant spectator quark model breaks the @xmath21-flavor symmetry , we restrict our study to reactions with proton targets ( average on the sqtm coefficients ) @xcite . later on , with the knowledge of the functions @xmath17 , and @xmath18 we predict the helicity amplitudes for transitions associated with the remaining members of the multiplet @xmath4 $ ] @xcite .
representation of the @xmath22 cross section .
the graph define the 3 resonance regions .
the vertical lines represent resonant states described by the covariant spectator quark model . at red
we indicate the states studded in this work . at blue are the states used as input.,width=362 ]
the covariant spectator quark model is based on the formalism of the covariant spectator theory @xcite . in the covariant spectator quark model ,
baryons are treated as three - quark systems .
the baryon wave functions are derived from the quark states according with the @xmath23 symmetry group .
a quark is off - mass - shell , and free to interact with the photon fields , and other two quarks are on - mass - shell @xcite . integrating over the quark - pair degrees of freedom we reduce the baryon to a quark - diquark system , where the diquark can be represented as an on - mass - shell spectator particle with an effective mass @xmath24 @xcite .
the electromagnetic interaction with the baryons is described by the photon coupling with the constituent quarks in the relativistic impulse approximation .
the quark electromagnetic structure is represented in terms of the quark form factors parametrized by a vector meson dominance mechanism @xcite .
the parametrization of the quark current was calibrated in the studies of the nucleon form factors data @xcite , by the lattice qcd data for the decuplet baryon @xcite , and encodes effectively the gluon and quark - antiquark substructure of the constituent quarks .
the quark current has the general form @xcite j_q^(q^2 ) = j_1(q^2 ) ^+ j_2(q^2 ) , [ eqjq ] where @xmath25 is the nucleon mass and @xmath26 @xmath27 are the dirac and pauli quark form factors . in the @xmath21-flavor sector
the functions @xmath26 can also be decomposed into the isoscalar ( @xmath28 ) and the isovector ( @xmath29 ) components : @xmath30 , where @xmath31 acts on the isospin states of baryons ( nucleon or resonance ) .
the details can be found in @xcite .
when the nucleon wave function ( @xmath32 ) and the resonance wave function ( @xmath33 ) are both expressed in terms of the single quark and quark - pair states , the transition current in impulse approximation as can be written @xcite j^= 3 _ _
k |_r ( p_+,k ) j_q^_n(p_-,k ) , [ eqjmu ] where @xmath34 , and @xmath35 are the nucleon , the resonance , and the diquark momenta respectively . in the previous equation
the index @xmath36 labels the possible states of the intermediate diquark polarizations , the factor 3 takes account of the contributions from the other quark pairs by the symmetry , and the integration symbol represents the covariant integration over the diquark on - mass - shell momentum . in the study of inelastic reactions
we replace @xmath37 in equation ( [ eqjq ] ) .
this procedure ensures the conservation of the transition current and it is equivalent to the use of the landau prescription @xcite .
using equation ( [ eqjmu ] ) , we can express the transition current in terms of the quark electromagnetic form factor @xmath38 ( @xmath39 ) and the radial wave functions @xmath40 and @xmath41 @xcite .
the radial wave functions are scalar functions that depend on the baryon ( @xmath42 ) and diquark ( @xmath35 ) momenta and parametrize the momentum distributions of the quark - diquark systems . from the transition
current we can extract the form factors and the helicity transition amplitudes , defined in the rest frame of the resonance ( final state ) , for the reaction under study @xcite .
there are however some processes such as the meson exchanged between the different quarks inside the baryon , which can not be reduced to simple diagrams with quark dressing .
those processes are regarded as arising from a meson exchanged between the different quarks inside the baryon and can be classified as meson cloud corrections to the hadronic reactions @xcite .
the covariant spectator quark model was used already in the study of several nucleon excitations including isospin 1/2 systems @xmath43 @xcite and the isospin 3/2 systems @xcite .
the model generalized to the @xmath44-flavor sector was also used to study the octet and decuplet baryons as well as transitions between baryons with strange quarks @xcite . in figure [ figsigmaw ]
the position of the nucleon excitations are represented and compared with the bumps of the cross sections .
based on the parametrization of the quark current ( [ eqjq ] ) in term of the vector meson dominance mechanism , the model was extended to the lattice qcd regime ( heavy pions and no meson cloud ) @xcite , to the nuclear medium @xcite and to the timelike regime @xcite .
the model was also used to study the nucleon deep inelastic scattering @xcite and the axial structure of the octet baryon @xcite .
for the study of the states @xmath45 ( @xmath47 ) and @xmath46 ( @xmath48 ) it is necessary to specify the shape of the radial wave function of the nucleon and resonant states .
the radial wave function can be represented using the dimensionless variable @xmath49 , where @xmath50 is the mass of the baryon @xmath51 .
we choose in particular _ n(p , k)= , _
r(p , k)= , [ eqpsiradial ] where @xmath52 are normalization constants , @xmath53 and @xmath54 are momentum range parameters in units @xmath55 .
we use @xmath56 for @xmath45 and @xmath57 for @xmath46 .
those parameters are fixed by a fit to the the large @xmath13 data @xcite .
the coefficient @xmath58 is determined by an orthogonality condition between the nucleon and the state @xmath59 . in the following we will use also the spectroscopic notation to represent the states @xmath45 ( or @xmath60 ) and @xmath45 ( or @xmath61 ) . in addition we use @xmath62 and @xmath63 to represent the @xmath60 and @xmath61 masses , respectively . from the study of @xcite
we conclude that we can write the amplitude @xmath7 for the @xmath45 state in terms of the dirac transition form factor ( @xmath64 ) .
the final result is then a_1/2 = - f_s ( f_1 + + 2 f_1- _ 3 ) i_s11 _ s , _
s11 ( q^2)= _ k _ s11(p_+,k ) _
n(p_-,k ) , [ eqa12-s11 ] where @xmath65 and @xmath66 is a covariant integral calculated on the @xmath60 rest frame .
the result ( [ eqa12-s11 ] ) is valid for large @xmath13 since only valence quark effects are considered and it is the consequence of the observation that the pauli transition form factor ( @xmath67 ) vanishes for @xmath68 gev@xmath69 , which is interpreted as the consequence of the cancellation between valence quark and meson cloud effects @xcite . in this work we updated the model from @xcite using the radial wave function @xmath41 given by equation ( [ eqpsiradial ] ) , in order to ensure the exact orthogonality between nucleon and @xmath45 states . in the process
we introduce a new parameter ( @xmath70 ) that is adjusted by the large @xmath13 data ( no meson cloud ) @xcite . the model for from @xcite can be used to calculate the electromagnetic transition form factors for the @xmath71 transition , including the magnetic dipole @xmath72 and the electric quadrupole @xmath73 form factors , based in the effects of the valence quarks .
one obtain then @xmath74 , where g_m= r(f_1 + + 2 f_1- _ 3 ) i_d13 + ( f_2 + + 2 f_2- _ 3 ) i_d13 . in the equations
@xmath75 , and @xmath76 is the new invariant integral defined at the resonance rest frame .
the result @xmath77 is interesting , since it is consistent with the expected qcd behavior for large @xmath13 , but it is inconsistent with the data at low @xmath13 , that shows a significant magnitude for the amplitude @xmath78 , near @xmath79 @xcite .
in general , quark models predict small contributions for @xmath8 ( 2040% ) @xcite .
there are however indications that the effects of the meson cloud contribution dominate the amplitude , as supported by the calculation from ebac at jefferson lab @xcite . based on that information
we represent the helicity amplitudes as a_1/2= f_d g_m + f_d g_4^ , a_3/2= f_d g_4^ , [ eqamps - d13 ] where @xmath80 . in the equations ( [ eqamps - d13 ] ) , @xmath81 is a function that is not determined by the covariant spectator quark model ( that predicts @xmath82 ) and parametrize the amplitude @xmath8 assuming the dominance of the pion / meson cloud effects . the function @xmath81 is fitted to the data with a model inspired on the pion and meson cloud contributions for the @xmath83 transition @xcite . the results from the amplitudes @xmath7 ( valence quark ) and @xmath8 ( meson cloud )
are presented on figure [ fign1520 ] .
results for the resonance @xmath46 .
the amplitude @xmath8 the result of meson cloud effects . was no contributions from valence quarks .
data from clas @xcite.,title="fig:",width=192 ] results for the resonance @xmath46 .
the amplitude @xmath8 the result of meson cloud effects . was no contributions from valence quarks .
data from clas @xcite.,title="fig:",width=192 ]
the combination of the wave functions of a baryon ( three - quark system ) given by @xmath16 group and the description of electromagnetic interaction in impulse approximation leads to the so - called single quark transition model @xcite . in this context _ single _ means that only one quark couples with the photon . in these conditions
the sqtm can be used to parametrize the transition current between two multiplets , in an operational form that includes only four independent terms , with coefficients exclusively dependent of @xmath13 . in particular
, the sqtm can be used to parametrize the @xmath0 transitions , where @xmath2 is a @xmath1 ( isospin 1/2 ) or a @xmath84 ( isospin 3/2 ) state from the @xmath4 $ ] multiplet , in terms on three independent functions of @xmath13 : @xmath17 , and @xmath18 @xcite .
the relations between the functions @xmath17 , and @xmath18 and the amplitudes are presented in the table [ tableamp ] . using the results for the @xmath85 and @xmath71 amplitudes ,
respectively @xmath86 , @xmath87 , and @xmath88 in the spectroscopic notation , we can write & & a= 2 + a_1/2^d13 + a_3/2^d13 , b= 2 - 2 a_1/2^d13 , + & & c= - 2 - a_1/2^d13 + a_3/2^d13 .
[ eqabc ] .[tableamp ] amplitudes @xmath7 and @xmath8 estimated by sqtm for the proton targets ( @xmath89 ) @xcite .
the angle @xmath90 is the mixing angle associated with the @xmath91 states ( @xmath92 ) .
the angle @xmath93 is the mixing angle associated with the @xmath94 states ( @xmath95 ) . [
cols="^,^,^",options="header " , ] another interesting prediction relatively to the helicity amplitudes of baryons with negative parity is the correlation between the amplitudes @xmath7 and @xmath96 associated with the @xmath97 transition . the consequence of the experimental result , @xmath98 , observed for @xmath68 gev@xmath69 is that we can write in that regime @xmath99 , where @xmath100 @xcite .
the correlation between the 2 amplitudes is shown on figure [ figscaling ] .
we combine the frameworks of the covariant spectator quark model and the single quark transition model in order to make predictions for the helicity amplitudes associated with negative parity resonances in the region of masses @xmath101 1.51.8 gev .
the predictions are expected to be valid for @xmath102 gev@xmath69 .
simple parametrizations for the amplitudes are calculated to facilitate the comparison with future experiments for @xmath103 gev@xmath69 .
contrarily to what it was expected , in some transitions , like for the resonances @xmath45 and @xmath104 , the valence quark effects are not dominant in the region @xmath105 2 gev@xmath69 .
the author thanks kumar gupta for comments and suggestions .
the author was supported by the brazilian ministry of science , technology and innovation ( mcti - brazil ) .
35 ramalho g , pea m t and gross f 2008 _ eur .
phys . j. a _ * 36 * , 329 ; _ phys . rev . d _ * 78 * , 114017 ; ramalho g , pea m t and gross f 2010 _ phys . rev .
d _ * 81 * , 113011 ; ramalho g , pea m t and stadler a 2012 _ phys
. rev .
d _ * 86 * , 093022 . | we present a model for the @xmath0 helicity amplitudes , where @xmath1 is the nucleon and @xmath2 is a negative parity nucleon excitation , member of the @xmath3-multiplet @xmath4 $ ] .
the model combines the results from the single quark transition model for the helicity amplitudes with the results of the covariant spectator quark model for the @xmath5 and @xmath6 transitions . with the knowledge of the amplitudes @xmath7 and @xmath8 for those transitions we calculate three independent coefficients defined by the single quark transition model and make predictions for the helicity amplitudes associated with the @xmath9 , @xmath10 , @xmath11 , and @xmath12 transitions . in order to facilitate the comparison with future experimental data at high @xmath13
, we provide also simple parametrizations for the amplitudes , compatible with the expected falloff at high @xmath13 . |
the problem of index coding over noiseless broadcast channels was introduced in @xcite and has been well studied @xcite - @xcite .
it involves a single source and a set of caching receivers .
each of the receivers wants a subset of the set of messages transmitted by the source and knows another non - intersecting subset of messages a priori as side information .
the problem is to minimize the number of binary transmissions required to satisfy the demands of all the receivers , which amounts to minimizing the bandwidth required .
an index coding problem @xmath1 , involves a single source , @xmath2 that wishes to send a set of @xmath3 messages @xmath4 to a set of @xmath5 receivers , @xmath6 .
the messages , @xmath7 , @xmath8 take values from some finite field @xmath9 .
a receiver @xmath10 , @xmath11 , is defined as @xmath12 .
@xmath13 is the set of messages demanded by @xmath10 and @xmath14 is the set of messages known to @xmath10 , a priori , known as the side information that @xmath10 has .
an index code for the index coding problem with @xmath15 consists of 1 .
an encoding map , @xmath16 , where @xmath17 is called the length of the index code , and 2 . a set of decoding functions @xmath18 such that , for a given input @xmath19 , @xmath20 .
an optimal index code for binary transmissions minimizes @xmath17 , the number of binary transmissions required to satisfy the demands of all receivers .
a linear index code is one whose encoding function is linear and it is linearly decodable if all the decoding functions are linear .
it was shown in @xcite that for the class of index coding problems over @xmath21 which can be represented using side information graphs , which were labeled later in @xcite as single unicast index coding problems , the length of optimal linear index code is equal to the minrank over @xmath21 of the corresponding side information graph .
this was extended in @xcite to general index coding problems , over @xmath22 , using minrank over @xmath22 of their corresponding side information hypergraphs . in this paper , we consider noisy index coding problems with @xmath15 over awgn channels . in the noisy version of index coding ,
the messages are sent by the source over a noisy broadcast channel . instead of binary transmissions
if multilevel ( @xmath23-ary , @xmath24 ) modulation schemes are used further bandwidth reduction can be achieved .
this has been introduced in @xcite for gaussian broadcast channels .
it was also found in @xcite that using @xmath23-ary modulation has the added advantage of giving coding gain to receivers with side information , termed the `` side information gain '' .
the idea of side information gain was characterized for the case where the source use @xmath23-psk for transmitting the index coded bits in @xcite , where @xmath25 , where @xmath17 is the length of the index code used , with the average energy of the @xmath23-qam signal being equal to the total energy of @xmath17 binary transmissions .
this paper discusses the case of noisy index coding over awgn channels where the source uses @xmath23-qam to transmit the index coded bits . as in @xcite , here also , m = @xmath26 , where @xmath17 is the length of the index code used . the contributions and organization in this paper may be summarized as follows : 1 .
an algorithm to map binary symbols to appropriate sized qam constellation is presented which uses the well known ungerboeck labelling as an ingredient .
( section [ sec3 ] 2 . a necessary and sufficient condition for a receiver to get side information coding gain is presented .
( theorem [ thm1 ] in section [ sec4 ] ) 3 .
it is shown that the difference in probability of error performance between the best and worst performing receivers increases monotonically as the length of the index code used increases .
( theorem [ thm2 ] in section [ sec4 ] ) in section [ sec2 ] the notions of bandwidth gain and qam side information coding gain are explained and simulation results are presented in section [ sec5 ] . concluding remarks constitute section [ sec6 ]
consider a general index coding problem @xmath1 with @xmath3 messages , @xmath4 and @xmath5 receivers , @xmath6 . let the length of a linear index code ( not necessarily optimal ) for the index coding problem at hand be @xmath17 .
we have @xmath27 , where @xmath28 is the length of the optimal index code which is equal to the minrank over @xmath29 of the corresponding side information hypergraph .
let the encoding matrix corresponding to the linear index code chosen be @xmath30 , where @xmath30 is an @xmath31 matrix over @xmath29 .
the index coded bits are given by @xmath32 = \textbf{x}l,$ ] where @xmath33.$ ] the noiseless index coding involves @xmath17 binary transmissions .
it was shown in @xcite that for noisy index coding problems , if we transmit the index coded bits as a point from @xmath26-psk signal set instead of @xmath17 binary transmissions , the receivers satisfying certain conditions will get coding gain in addition to bandwidth gain whereas other receivers trade off coding gain for bandwidth gain .
this gain which was termed as the `` psk side information coding gain '' was obtained by proper mapping of index coded bits to psk symbols an algorithm for which was presented . in this paper , we extend the results in @xcite for the case where we use @xmath26-qam to transmit the @xmath17 index coded bits .
the term * qam bandwidth gain * is defined as the bandwidth gain obtained by each receiver by going from @xmath17 binary transmissions to a single @xmath26-qam symbol transmission .
when we transmit a single @xmath34 signal point instead of transmitting @xmath17 binary transmissions we are going from an @xmath17- real dimensional or equivalently @xmath35 - complex dimensional signal set to 1 complex dimensional signal set .
hence all the receivers get a @xmath35 - fold qam bandwidth gain .
we state this simple fact as [ lem : qam bg ] each receiver gets an @xmath35 - fold qam bandwidth gain .
the term * qam side information coding gain * ( qam - sicg ) is defined as the coding gain a receiver with a non - empty side information set gets w.r.t a receiver with no side information while using @xmath26-qam to transmit the index coded bits .
let the set @xmath36 , @xmath37 , be defined as the set of all binary transmissions which a receiver @xmath38 knows a priori due to its available side information , i.e. , @xmath39 also , let @xmath40 , @xmath41 be defined as follows .
in this section , we describe an algorithm to map the index coded bits to signal points of an appropriate sized qam constellation so that the receivers satisfying the conditions of theorem [ thm1 ] in the following section will get qam - sicg . for the given index coding problem ,
choose an index code of length @xmath17 .
this fixes the value of @xmath43 .
order the receivers in the non - decreasing order of @xmath44 .
wlog , let @xmath45 be such that @xmath46 . before starting to run the algorithm to map the index coded bits to @xmath26-qam symbols , we need to 1 .
choose an appropriate @xmath26-qam signal set .
2 . use ungerboeck set partitioning @xcite to partition the @xmath26-qam signal set chosen into subsets with increasing minimum subset distances . to choose the appropriate qam signal set , do the following : * *
if * @xmath17 is even , then choose the @xmath26-square qam with average symbol energy being equal to @xmath17 .
* * else * , take the @xmath47-square qam with average symbol energy equal to @xmath17 .
use ungerboeck set partitioning @xcite to partition the @xmath47 qam signal set into two @xmath26 signal sets .
choose any one of them as the @xmath26-qam signal set .
let @xmath48 denote the different levels of partitions of the @xmath26-qam with the minimum distance at layer @xmath49 , @xmath50 , being such that @xmath51 .
the algorithm to map the index coded bits to qam symbols is given in * algorithm 1*. , do an arbitrary order mapping and * exit*. @xmath52 , * exit*. fix @xmath53 such that the set of codewords , @xmath54 , obtained by running all possible combinations of @xmath55 with @xmath56 has maximum overlap with the codewords already mapped to psk signal points .
, * @xmath57=@xmath58 together with all combinations of @xmath55 will result in @xmath59 . * @xmath60 * * if * @xmath61 * then * , * * @xmath52 . * * goto * step 3 * * * else * , goto * step 3 * * of the codewords in @xmath62 which are yet to be mapped , pick any one and map it to a qam signal point in that @xmath63 sized subset at level @xmath64 which has maximum number of signal points mapped by codewords in @xmath62 without changing the already labeled signal points in that subset .
+ if all the signal points in such a subset have been already labeled , then map it to a signal point in another @xmath63 sized subset at the same level @xmath64 that this point together with the signal points corresponding to already mapped codewords in @xmath62 , has the largest minimum distance possible .
clearly this minimum distance , @xmath65 is such that @xmath66 . *
@xmath52 * goto * step 3 * note that algorithm [ algo1 ] above does not result in a unique mapping of index coded bits to @xmath26-qam symbols .
the mapping will change depending on the choice of @xmath67 in each step .
however , the performance of all the receivers obtained using any such mapping scheme resulting from the algorithm will be the same .
if @xmath68 for some @xmath69 , depending on the ordering of @xmath40 done before starting the algorithm , @xmath10 and @xmath70 may give different performances in terms of probability of error .
@xmath10 and @xmath70 with @xmath68 will give the same performance if and only if @xmath71 or vice - versa .
in this section we present the main results apart from the algorithm given in the previous section . [ thm1 ] a receiver @xmath72 gets qam side information coding gain , with the scheme proposed , if and only if @xmath73 , where @xmath17 is the length of the index code used .
consider a receiver @xmath12 .
let @xmath74 and @xmath57 @xmath75 @xmath76 , @xmath77 . for any given realization of @xmath67 , the effective signal set seen by the receiver @xmath10
consists of @xmath63 points .
let @xmath78 the minimum distance of the signal set seen by the receiver @xmath10 , @xmath79 _ proof of the ` if part ' : _ if @xmath73 , then the effective signal set seen by the receiver @xmath10 will have @xmath80 points .
hence by appropriate mapping of index coded bits to qam symbols , we can increase @xmath65 .
thus @xmath10 will get coding gain over a receiver that has no side information because the minimum distance seen by a receiver with no side information will be the minimum distance of @xmath26-qam signal set .
_ proof of the ` only if part ' : _ let us a consider a receiver @xmath10 such that @xmath81 .
then @xmath65 , will not increase .
@xmath65 will remain equal to the minimum distance of the corresponding @xmath82 - qam , same as that of a receiver with no side information .
thus a receiver @xmath10 with @xmath81 will not get qam - sicg .
it is to be noted that the value of @xmath40 not only depends on @xmath83 but also on @xmath84 , which , in turn , depends on the index code chosen .
hence for the same index coding problem , a particular receiver may satisfy theorem [ thm1 ] and get qam - sicg for some index codes and may not get qam - sicg for other index codes .
[ thm2 ] the difference in probability of error performance between the best performing receiver and the worst performing receiver for a given index coding problem , while using @xmath26- qam signal point to transmit the index coded bits , will increase monotonically while @xmath17 increases from @xmath28 to @xmath3 if the following conditions are satisfied . +
( 1 ) the best performing receiver gets qam - sicg .
+ ( 2 ) the worst performing receiver has no side information .
if there is a receiver with no side information , say @xmath85 , whatever the length , @xmath17 , of the index code used is , the effective signal set seen by @xmath85 will be @xmath26-qam .
therefore the minimum distance seen by @xmath85 will be the minimum distance of @xmath26-qam signal set . for @xmath26-qam with average symbol energy equal to @xmath17 , the squared minimum pair - wise distance of @xmath26-qam , @xmath86-qam@xmath87 , obtained by the proposed mapping scheme in algorithm [ algo1 ]
is given by @xmath88 which is monotonically decreasing in @xmath17 .
therefore the performance of the receiver with no side information deteriorates as the length of the index code increases from @xmath28 to @xmath3 .
although condition ( 1 ) in theorem [ thm2 ] is necessary , the same can not be said about condition ( 2 ) . even when condition ( 2 ) above is not satisfied , i.e. , the worst performing receiver has at least 1 bit of side information but not the same amount of side information as the best performing receiver , the difference between their performances can still increase monotonically as we move from @xmath28 to @xmath3 .
this is because the error performance is determined by the effective minimum distances seen by the receivers , which , in turn , depend on the mapping used between index coded bits and qam symbols .
a general index coding problem can be converted into one where each receiver demands only one message since a receiver @xmath12 can be converted into @xmath89 receivers all with the same side information @xmath90 and each demanding a single message .
so it is enough to consider index coding problems where the receivers demand a single message each and hence both the examples considered in this section are such problems . even though the examples considered are what are called single unicast index coding problems in @xcite , the results hold for any general index coding problem . in this subsection , we give an example with simulation results to support our claims in section [ sec4 ] .
the mapping of index coded bits to qam symbols is done using our algorithm [ algo1 ] .
the receivers which satisfy theorem [ thm1 ] are shown to get qam - sicg .
we also compare the performance of different receivers while using qam and psk to transmit index coded bits . for a given index coding problem and a chosen index code , the mapping of index coded bits to qam symbols
is done using the algorithm described in section [ sec3 ] , whereas the mapping to psk symbols is done using the algorithm 1 in @xcite .
we also give the effective minimum distances which are seen by different receivers which explains the difference in their error performance .
[ ex_psk_qam ] let @xmath917 .
@xmath93 . + the minrank over @xmath21 of the side information graph corresponding to the above problem evaluates to @xmath28=4 .
an optimal linear index code is given by the encoding matrix , + @xmath94 $ ] .
the index coded bits are , @xmath95 ; @xmath96 ; @xmath97 ; @xmath98 .
the 16-qam mapping for the above example is given in fig .
[ map_ex - psk_qam ] .
the simulation result which compares the performance of different receivers when they use 16-qam and 16-psk for transmission of index coded bits is shown in fig .
[ sim_ex - psk_qam ] .
the probability of error plot corresponding to 4-fold binary transmission is also shown in fig .
[ sim_ex - psk_qam ] .
the reason for the difference in performance while using qam and psk can be explained using the minimum distance seen by the different receivers for the 2 cases .
this is summarized in table [ table - ex_psk_qam ] . .
[ cols="^,^,^,^,^,^,^,^",options="header " , ]
in this paper we considered noisy index coding over awgn channel .
the problem of finding an optimal index code , for a given index coding problem , is , in general , exponentially hard .
however , we have shown that finding the minimum number of binary transmissions required is not required for reducing transmission bandwidth over a noisy channel since , we can use an index code of any given length as a single qam point thus saving bandwidth .
the mapping scheme by the proposed algorithm and qam transmission are valid for any general index coding problem .
it was further shown that if the receivers have huge amount of side information , it is more advantageous to transmit using a longer index code as it will give a higher coding gain as compared to binary transmission scheme .
1 anjana a. mahesh and b. sundar rajan , `` index coded psk modulation , '' arxiv:1356200 , [ cs.it ] 19 september 2015 .
y. birk and t. kol , informed - source coding - on - demand ( iscod ) over broadcast channels , " in _ proc .
ieee conf .
_ , san francisco , ca , 1998 , pp .
1257 - 1264
. z. bar - yossef , z. birk , t. s. jayram and t. kol , index coding with side information , " in _ proc .
47th annu .
ieee symp . found .
_ , 2006 , pp .
197 - 206 .
l ong and c k ho , optimal index codes for a class of multicast networks with receiver side information , " in _ proc .
ieee icc _ , 2012 ,
2213 - 2218 .
l. natarajan , y. hong , and e. viterbo , index codes for the gaussian broadcast channel using quadrature amplitude modulation , " in _ ieee commun .
_ , aug . 2015 .
s h dau , v skachek , y m chee , error correction for index coding with side information , `` in _ ieee trans .
inf theory _ , vol .
59 , no.3 , march 2013 . g. ungerboeck , ' ' channel coding for multilevel / phase signals " , in _ ieee trans .
inf theory _ , vol .
it-28 , no .
1 , january 1982 . | this paper discusses noisy index coding problem over gaussian broadcast channel .
we propose a technique for mapping the index coded bits to m - qam symbols such that the receivers whose side information satisfies certain conditions get coding gain , which we call the * qam side information coding gain*. we compare this with the psk side information coding gain , which was discussed in @xcite .
index coding , awgn broadcast channel , m@xmath0qam , qam side information coding gain . |
recently it was demonstrated @xcite that two qubits subject to common thermal bosonic environment ( bath ) can develop considerable entanglement .
a similar result has also been obtained for qubits interacting via fermionic environment @xcite . here
we review our results on the derivation of the induced exchange interaction and quantum noise in a unified formulation @xcite , focusing the presentation on the onset and development of the cross - qubit correlations due to the bath .
we consider a 1d channel model for the bath , motivated by recent experiments @xcite , and allow bosons ( e.g. phonons , photons ) to propagate along a single direction with wave vector @xmath0 and dispersion @xmath1 .
more general results are available in @xcite .
two qubits immersed in this environment are separated by distance @xmath2 such that the interaction due to the wave function overlap is negligible . the qubits interaction with the bosonic bath
is introduced @xcite as @xmath3 where @xmath4 is the standard pauli matrix of qubit @xmath5 and 2 , and @xmath6 the total hamiltonian is @xmath7 , where @xmath8 , @xmath9 represents the hamiltonian of the qubit system , and we set @xmath10 . the reduced density matrix that describes the dynamics of the qubit system is , then , given as the trace of the total density matrix over the bath modes ,
@xmath11 where the initial density matrix is assumed factorized and consists of the system and bath parts .
the latter is @xmath12 . for large times , a more realistic model of the environment assumes rethermalization , and markovian schemes are appropriate for the description of the dynamics @xcite .
however , for short times the present formulation is adequate and provides a useful solvable model for the case of otherwise gapless qubits , @xmath13 , which we consider from now on .
with the assumptions outlined above , we utilized bosonic operator techniques@xcite to derive an exact expression @xmath14 here the projection operator is defined as @xmath15 , with @xmath16 the eigenvectors of @xmath17 .
the real part of the exponent in ( [ eq : s3:adiabaticsolution ] ) leads to decay of off - diagonal density - matrix elements resulting in decoherence , @xmath18.\end{aligned}\ ] ] the imaginary part , yielding the induced interaction , is @xmath19 we defined the standard `` spectral '' functions @xcite @xmath20 @xmath21 to evaluate ( [ eq : s3:decoherencefunction - rel ] ) and ( [ eq : s3:decoherencefunction - iml ] ) , we consider the model in which the density of modes together with the coupling constants are approximated by the power - law function of the frequency with superimposed exponential cutoff @xcite , i.e. , @xmath22 for @xmath23 this corresponds to the well known ohmic model @xcite .
one can show that if the real part of @xmath24 were absent , the exponential involving the imaginary part would yield coherent dynamics with the unitary evolution operator @xmath25 $ ] .
the constant hamiltonian @xmath26 represents the induced interaction , @xmath27\!\sigma _
x^1 \sigma _ x^2.\ ] ] the time dependent term is given by @xmath28 @xmath29 commutes with @xmath26 and therefore could be viewed as the initial time - dependent modification of the interaction during its onset : @xmath29 vanishes for large times as @xmath30 , but note that @xmath31 . the interaction hamiltonian ( [ eq : s3:h - int ] ) is consistent with the results obtained @xcite within a perturbative markovian approach , for more general cases . in figure 1
, we plot the magnitude of the interaction hamiltonian @xmath32 , defined via @xmath33 , as a function of the qubit - qubit separation for various @xmath34 . at large distances
the interaction decreases as @xmath35 , for even @xmath34 , and @xmath36 , for odd @xmath34 .
this means , for instance , that for spins ( as qubits ) with @xmath37 , the induced interaction decreases slower as compared to the dipole - dipole magnetic interaction ; see estimates for semiconductor impurity electron spins in @xcite . ) and super - ohmic ( @xmath38 ) bath models as a function of qubits separation.,width=283 ] the decoherence terms , ( [ eq : s3:decoherencefunction - rel ] ) , describe quantum noise that ultimately destroys the coherent dynamics given by @xmath26 ( and @xmath29 ) . to study the effect of these terms , we evaluate the concurrence @xcite which measures the entanglement of the spin system and is monotonically related to the entanglement of formation @xcite . for a mixed state of two qubits
we first define the spin - flipped state , @xmath39 , and then the hermitian operator @xmath40 , with eigenvalues @xmath41 .
the concurrence is then given @xcite by @xmath42 , @xmath43 , @xmath23 .
the top inset shows dynamics for different temperatures : @xmath44.,width=283 ] in figure 2 , we plot the concurrence as a function of time and the qubit - qubit separation , for the ( initially unentangled ) state @xmath45 , and @xmath23 .
the bath - mediated interaction between the qubits creates entanglement , which oscillates according to the magnitude of @xmath26 .
the same bath also damps the oscillations destroying the entanglement for larger times .
the decay rate of the envelope is proportional to the temperature , as shown in the inset of figure 2 . for the corresponding dynamics of the density matrix elements
see section 5 .
let us now investigate in greater detail the onset of the induced interaction the time - dependence of which is given by @xmath29 . in figure 3 , we plot the magnitude defined via @xmath46 , as a function of time for various qubit - qubit separations and @xmath23 .
the correction is initially non - monotonic , but decreases for larger times as mentioned above .
the behavior for other non - ohmic regimes is initially more complicated , however the large time behavior is similar
. it may be instructive to consider the time dependent correction , @xmath47 , to the interaction hamiltonian during the initial evolution , corresponding to @xmath29 .
since @xmath29 commutes with itself at different times , it generates unitary evolution according to @xmath48 $ ] , with @xmath49/dt$ ] , @xmath50,\end{aligned}\ ] ] where @xmath51/[1+\xi^2]^{n/2}$ ] .
the above expression is a superposition of two waves propagating in opposite directions . in the ohmic case ,
@xmath23 , the shape of the wave is simply @xmath52 . in figure 4
, we present the amplitude of @xmath47 , defined via @xmath53 , as well as the sum of @xmath26 and @xmath47 , for @xmath23
. one can observe that the `` onset wave '' of considerable amplitude and shape @xmath54 propagates once between the qubits , `` switching on '' the interaction .
it does not affect the qubits once the interaction has set in . .
the ohmic ( @xmath23 ) case is shown . ,
width=283 ] ) case is shown .
the inset demonstrates the onset of the cross - qubit interaction on the same time scale.,width=283 ]
to understand the dynamics of the qubit system and its entanglement , let us again begin with the analysis of the coherent part in ( [ eq : s3:adiabaticsolution ] ) .
after the interaction , @xmath26 , has set in , it will split the system energies into two degenerate pairs , @xmath55 and @xmath56 .
the wave function is then @xmath57|\psi(0)\rangle$ ] . for the initial `` up - up '' state , @xmath58
, it develops as @xmath59 .
one can easily notice that at times @xmath60 , maximally entangled states are obtained , while at times @xmath61 , the entanglement vanishes ; these special times can also be seen in figure 2 .
the bath also induces decoherence that enters via ( [ eq : s3:decoherencefunction - rel ] ) .
the result for the entanglement is that the decaying envelope function is superimposed onto the coherent dynamics described above .
the magnitudes of the first and subsequent peaks of the concurrence are determined only by this function . as temperature increases
, the envelope decays faster resulting in lower values of the concurrence , see the inset in figure 2 .
note also that generation of entanglement is possible only provided that the initial state is a superposition of the eigenvectors of the induced interaction with more than one eigenvalue ( for pure initial states ) .
for example , the initial state @xmath62 in our case would only lead to the destruction of entanglement , i.e. , monotonically decreasing concurrence , similar to the results in @xcite . since @xmath13
, there is no relaxation by energy transfer in the system , and the exponentials in ( [ eq : s3:adiabaticsolution ] ) , with ( [ eq : s3:decoherencefunction - rel ] ) , suppress only the off - diagonal matrix elements , i.e. , those with @xmath63 .
it happens , however that at large times the @xmath64-dependence is not important in ( [ eq : s3:decoherencefunction - rel ] ) and @xmath65 vanishes for certain values of @xmath63 .
in the basis of the qubit - bath interaction , @xmath66 , the limiting @xmath67 density matrix for our initial state ( @xmath45 ) is @xmath68 , which takes the form @xmath69 in the basis of states @xmath45 , @xmath70 , @xmath71 , and @xmath72 . the significance of such results , see also @xcite , is that in the present model not all the off - diagonal matrix elements are suppressed by decoherence , even though the concurrence of this mixed state is zero .
the probabilities for the qubits to occupy the states @xmath45 , @xmath70 , @xmath71 , and @xmath72 are presented in figure 5 . for the initial state @xmath45 , only the diagonal and inverse - diagonal matrix elements are affected , and the system oscillates between the two states @xmath45 and @xmath72 , as mentioned earlier in the description of the coherent dynamics , while decoherence dampens these oscillations down .
in addition , decoherence actually raises the other two diagonal elements to a certain level , see ( [ eq : rolimit ] ) .
the dynamics of the inverse - diagonal density matrix elements is shown in figure 6 . , @xmath70 , @xmath71 , and @xmath72 .
the parameters are the same as in figure 1 .
the inset shows the structure of the reduced density matrix ( the non - shaded entries are zeros).,width=188 ]
to summarize , we studied the initial stages of the cross - qubit interaction induced by a thermal bosonic bath .
it was shown that thermal environment can create a sufficiently large entanglement for quantum control , though it is erased for larger times .
the dynamics of the entanglement and the density matrix elements have been investigated .
m. a. ruderman and c. kittel , phys .
rev . * 96 * , 99 ( 1954 ) ; t. kasuya , prog .
* 16 * , 45 ( 1956 ) ; k. yosida , phys . rev . * 106 * , 893 ( 1957 ) ; v. privman , i. d. vagner , and g. kventsel , phys . lett .
a * 239 * , 141 ( 1998 ) ; c. piermarocchi , p. chen , l. j. sham , and d. g. steel , phys .
lett . * 89 * , 167402 ( 2002 ) ; d. mozyrsky , a. dementsov , and v. privman , phys . rev .
b * 72 * , 233103 ( 2005 ) ; y. rikitake and h. imamura , phys . rev .
b * 72 * , 033308 ( 2005 ) .
n. j. craig , j. m. taylor , e. a. lester , c. m. marcus , m. p. hanson , and a. c. gossard , science * 304 * , 565 ( 2004 ) ; j. m. elzerman , r. hanson , l. h. willems van beveren , b. witkamp , l. m. k. vandersypen , and l. p. kouwenhoven , nature * 430 * , 431 ( 2004 ) ; m. r. sakra , h. w. jiang , e. yablonovitch , and e. t. croke , appl .
lett . * 87 * , 223104 ( 2005 ) . | we summarize our recent results @xcite for the induced exchange interaction due to thermal bosonic environment ( bath ) which also generates quantum noise .
our focus here is on the onset of the interaction .
we demonstrate that the induced interaction can be used to manipulate and create entanglement over time scales sufficiently large for controlling the two - qubit system for quantum computing applications , though ultimately the noise effects will dominate . |
low mass x - ray binary ( lmxb ) is composed of a compact object ( neutron star or black hole ) accreting matter from its low mass companion star ( @xmath3 ) . according to the x - ray spectral proporties and rapid timing variabilities , the neutron star ( ns ) lmxbs are usually classified as the z sources and the atoll sources .
they are named after the shapes which display in x - ray color - color diagram ( cd ) or hardness - intensity diagrams ( hid ) .
the z sources produce approximate eddington luminosities with soft x - ray spectra , whereas atoll sources produce a lower luminosities in the range @xmath4 * ? ? ?
a typical z source track shows three branches , from top to bottom , which are usually called the horizontal branch ( hb ) , the normal branch ( nb ) , and the flaring branch ( fb ; @xcite ) , respectively . for atoll sources , the three branches are called extreme island , island , and banana state .
based on the orientation of branches , six typical z sources are further divided into the cyg- like z sources ( , and ) with a horizontal hb ( z "- shaped tracks ) and the sco - like z sources ( , , and ) with a vertical hb ( @xmath5"-shaped tracks ) . the black hole ( bh ) and
ns lmxbs show many similarities in their timing behaviors .
low - frequency quasi - periodic oscillations ( lf - qpos ) are observed in z sources , atoll sources and bh lmxbs . in general , the centroid frequencies of lf - qpos are @xmath61 - 70 hz .
the type c , b and a lf - qpos in bh lmxbs were considered as corresponding to hbos , nbos , fbos of z sources @xcite , respectively .
moreover , the w - k relation , the strong correlation between the centroid frequency of lf - qpo and the break frequency in power density spectral , was identified in bh lmxbs , accreting millisecond pulsars and atoll sources @xcite .
z sources show a similar but slightly shifted relation .
these similar characteristics suggest that lf - qpos are likely produced from the same physical mechanism in lmxbs .
lense - thirring precession was introduced to interpret hbos in ns lmxbs as well as type c lf - qpos in bh lmxbs @xcite . in this model
, the lf - qpos were arisen from the misalignment between the compact star s spin axis and the rotational axis of the inner accretion disk @xcite .
@xcite discussed the possible origin of hbo from hot inner flow precession .
the evolution of accretion rate @xmath7 is studied from the hid of lmxb because it contains the variation of x - ray spectrum and radiation intensity . in bh lmxbs
, @xmath7 increases in the following direction : the low hard state ( lhs ) the hard intermediate state ( hims ) the soft intermediate state ( sims ) the high - soft state ( hss ) @xcite
. meanwhile , the type c lf - qpo was only observed in the hims in bh lmxbs .
the frequency of type c qpo positively correlates with accretion rate and energy flux @xcite .
however , the evolution of accretion rate @xmath7 in the ns lmxbs is still controversial in the cyg - like and sco - like z sources .
the ascending trend of accretion rate is not always consistent with the source intensity increasing direction in the hids . according to multi - wavelength campaigns of the classic z source cyg x-2 , accretion rate monotonically increases as hb - nb - fb @xcite .
based on the boundary layer emission model , @xcite also found @xmath7 increasing from the hb to the fb in gx 340 + 0 .
however , church and co - workers @xcite applied the extend adc model for the cyg - like z sources and suggested an opposite direction , i.e. , @xmath7 increases from the fb / nb vertex to the hb / nb vertex .
@xcite considered that @xmath7 maintains constant along z tracks .
for classical z sources , the frequency of hbo increased from hb to hb / nb vertex .
so , three distinct correlations between the accretion rate and the frequency of hbo were proposed , that is , positive correlation , anticorrelation and non - correlation .
the referred sources in the above works were analyzed either in the cyg - like source or in the sco - like source .
the unique z source , , switched from a cyg - like z source to a sco - like z source at high luminosity and from z source to atoll source at low luminosity , which was observed by the rossi x - ray timing explorer ( rxte ) during its 2006 - 2007 outburst .
the secular change of was driven by the accretion rate variation .
@xcite studied the spectra evolutions of prudently , and suggested that the accretion rate @xmath7 maintains constant on the nb and fb of z sources .
while on the hb , the @xmath8 correlation biased from the constant @xmath7 line because the disk component encountered a comptonization upscattering
. the constant @xmath7 should be satisfied after the comptonization component accounted .
in , the constant @xmath7 was also established @xcite .
@xcite indicated that the accretion rate was invariant in the z sources and the oscillation of accretion rate @xmath7 produced the z tracks .
however , @xcite concluded that the accretion rate of the disk follow @xmath9 after considering the magnetic field effect during the accretion . in previous works , the relation between the spectra parameters and the characteristics of timing variability was not utilized to study the accretion rate variation . provides us a great opportunity to understand the temporal variabilities varying with the accretion rate evolution while the ns lmxb source transited from the cyg - like , via the sco - like z source , to an atoll source .
@xcite indicated two exactly opposite disk radius - hbo frequency relations in when the cutoff power law was replaced by nthcomp . in this paper
, we will provide a model independent method to study the hbos behaviors with decreasing accretion rate . in sec .
2 , we analyze the public archive data of . in sec . 3 , we study the x - ray spectra and timing variabilities of the hb / nb vertices and the nb / fb vertices , and then investigate the energy dependence of the hbo .
the discussions and conclusions are displayed in sec .
4 & sec . 5 , respectively .
during its 2006 - 2007 outburst , experienced from a z source to an atoll source in the x - ray spectral and timing behaviours collected by the the proportional counter array ( pca ) instrument on board of _ rxte _ @xcite .
we used the heasoft 6.12 package to extract background - subtracted light curves with a 16 s time resolution from the standard 2 " mode data in figure [ fig1 ] .
only data collected by pcu 2 were used .
we applied the following standard data selection criteria , a source elevation of @xmath10 , a pointing offset of @xmath11 and a south atlantic anomaly exclusion time of 30 minutes . in order to obtain the ccds / hids from the light curves , we define the soft color as 4.5 - 7.4 kev/2.9 - 4.1 kev ( channels 10 - 17/6 - 9 ) count rate ratios , and the hard color as 10.2 - 18.1 kev/7.8 - 9.8 kev ( channels 24 - 43/18 - 23 ) count rate ratios ,
then define the intensity as count rate covering the energy range 2.9 - 18.1 kev ( channels 6 - 43 ) .
the selected time intervals and observations are list in table [ tbl-1 ] .
we constructed two cds / hids from the observation of in figure [ fig2 ] and figure [ fig3 ] , interval a for the cyg - like phase and interval b for the sco - like phase .
suggested by @xcite , we also reproduced their hb / nb and nb / fb vertex in stages i - iii as the reference region for estimating accretion rate .
each point represents 960 s duration in figure [ fig4 ] . to investigate the hbos for the intervals a and b and the hb / nb vertex , we extracted the light curves of channels 0 - 13 , 14 - 35 and 36 - 149 from sb_125us_0_13_1s , sb_125us_14_35_1s and e_125us_64m_36_1s , respectively .
we merged these three energy bands light curves to obtain channel 0 - 149 light curves with 1/512 s time resolution .
when the source counts became lower , we extracted the channel 0 - 149 light curves from e_125us_64m_0_1s mode directly .
no dead time correction or background substraction were applied . in order to study the energy dependence of hbos
, we extracted the light curves pca channels 0 - 10 , 11 - 13 , 14 - 16 , 17 - 22 , 23 - 35 with 1/512 s resolution from b_2ms_8b_0_35_q .
the choice of energy bands and their centroid energies are listed in table [ tbl-2 ] .
we created power density spectra ( pdss ) by using 16 s data segments for the full channel 0 - 149 and all sub - channel light curves .
therefore , the fourier frequency is in the range 1/16 - 256 hz .
a multi - lorentzian model is used to fit the pds . in figure
[ fig5 ] , we show the pds with frequency - times - power representation ( @xmath12 ) of interval a. in order to obtain the best fitting result , four lorentzian components are needed @xcite .
we obtain three characteristic frequencies of pds , which are the centroid frequency of hbo ( @xmath13 ) , the break frequency of low frequency noise ( @xmath14 ) and @xmath15 ( the frequency where the maximum of @xmath12 occurs , where @xmath16 is the full - width - half - maximum ( fwhm ) of the lorentzian component ) , respectively .
we obtained the stages i - iii vertices and intervals a and b spectra with 16 s resolution from standard 2 model data and added a @xmath17 systematic error to each channel .
we also extracted the hexte data from cluster b to study the high energy emission mechanism simultaneously .
for the hexte data , we rebinned the spectra to obtain high s / n ratio . in the z stage of , @xcite fitted the spectral with the model * wabs(bbodyrad+diskbb+gaussian+cbpl ) * ( in xspec , hereafter model i ) , which accounted for the absorbed black body from the ns surface , the multi - color disk , the iron line , the comptonized component , respectively .
suggested by @xcite , we fixed the hydrogen column density and the iron line width at @xmath18 and 0.3 kev respectively .
for the constrained broken power law ( cbpl ) model , the break energy is fixed at 20 kev and the low energy photon index is set as 2.5 .
we assume a 70 degree inclination and the source distance is 8.8 kpc with @xmath19 uncertainty@xcite .
lin et al .
( 2012 ) provided opposite @xmath20 relations when the cbpl was replaced by the nthcomp in , i.e. , * wabs(bbodyrad+diskbb+gaussian+nthcomp ) * ( in xspec , hereafter model ii ) .
therefore , we attempted to study the @xmath20 relation when these two different models were accounted for the comptonized component in .
we utilized the xspec procedure to fit the hb of intervals a and b by using model i and model ii simultaneously . during the energy spectral fitting
, we fixed @xmath21 at 6.0 kev and @xmath22 at 2.5 .
nevertheless , the vertices of stages i - iii could be well fitted by the model * wabs(bbodyrad+diskbb+gaussian ) * , where the comptonized component could be ignored . from the above model parameters ,
the apparent ns radius was derived from @xmath23 , where the @xmath24 is the normalization of blackbody component .
it is noticeable that the inner radius from model ii in @xcite ( i.e. , marked as @xmath25 in their paper ) accounted the mcd component as well as the nthcomp component .
we also attempted to account the high energy photon from nthcomp to obtain the refined inner disk radius .
however , the inner disk radius did not vary significiantly after adding the nthcomp to mcd .
hereafter , for both model i and model ii , the inner disk radius @xmath26 was deduced from @xmath27 , where the @xmath28 and @xmath29 are the normalization of mcd component and inclination angle , respectively .
the individual component flux was integrated from the fitting model in the energy range @xmath63 - 60 kev .
the luminosity was obtained from flux by multiplying @xmath30 .
figure [ fig6 - 1]-[fig6 - 4 ] showed the unfolded spectra of the hb in intervals a and b. we listed the spectral and timing parameters in tabel [ tbl-3 ] - [ tbl-6 ] . all quoted errors are 90% confidence level . the fitting results were accounted 15% uncertainty of the distance to the source .
in figure [ fig4 ] , the hb / nb and nb / fb vertex of the z source followed two straight lines in hid . in figure
[ fig7 ] , we showed the black body component ( @xmath31 ) and the multi - color disk component ( @xmath32 ) as a function of total luminosity ( @xmath33 ) .
we obtained similar results as figure 16 in @xcite . for the hb / nb vertices ,
both @xmath31 and @xmath32 increased with @xmath33 enhancement . for the nb / fb vertices
, @xmath32 performed a similar behavior .
however , @xmath31 descends at the beginning of stage ii .
we presented the disk accretion rate versus the inner disk radius in figure [ fig8 ] . for the hb / nb vertices ,
the relation @xmath34 was obtained . however ,
a more gradual relation was shown for the nb / fb vertices , @xmath35 .
the obviously discrepancy of slop indicates that the accretion disk transfers from a slim disk in hb / nb vertex to a standard disk in nb / fb vertex @xcite .
@xcite suggested that the disk accretion rate of hb / nb vertex could be traced by using @xmath36 in place of @xmath37 to take into account the comptonization emission in the hb .
as we mentioned above , the @xmath38 component was ignored because of its faintness .
hereafter , we adopted the disk accretion rate as @xmath39 , where we assumed the 1.4@xmath40 for the mass of neutron star and @xmath41 for the eddington accretion rate .
the maximum @xmath42 is @xmath43 which located at the cyg - like stage .
the total luminosity / disk accretion rate-@xmath42 relation of hb / nb vertex were displayed in figure [ fig9 ] . with decreasing accretion rate ,
the correlation coefficient between the total luminosity and the frequency of hbo is merely 0.36 .
we used @xmath44 to fit the data from stage 1 and 2 displayed in figure [ fig9 ] , and the least squared method was applied .
then we obtained @xmath45 .
we also discuss whether a constant model can fit better than the above model .
we denote the above model as model 1 , and @xmath46 as model 2 . an approach to
this question is using f - test .
there are @xmath47 data points ( n=23 in our works ) to estimate parameters of model 1 and 2 .
we can calculate @xmath48 statistic , which is @xmath49 , where , @xmath50 is the regression sum of squares of model @xmath29 , and @xmath51 is the residual sum of squares of model 2 . assuming the null hypothesis that model 1 does not provide a significantly better fitting than model 2 , the statistic @xmath48 will have f - distribution with degree of freedom of @xmath52 . for a given false - rejection probability ( eg . , @xmath53 ) , the null hypothesis is accepted if the statistic @xmath48 calculated from our data is less than the critical value of the @xmath48 distribution , @xmath54 . from our data
, @xmath48 equals to 2.54 , which is less than @xmath55 .
it means that model 1 does not provide a significantly better fit than model 2 .
so , a constant model fits the data well enough .
it means that the total luminosity as well as the disk accretion rate are independent with the hbo . in figure
[ fig10 ] , we also show that the inner disk radius has no correlation to the @xmath42 at the hb / nb vertex .
we fitted the hb of intervals a and b with model i and ii simultaneously .
two models provided nearly the same results , which show positive @xmath56 relation in figure [ fig8 ] . within these two models , @xcite offered two distinct opposite @xmath56 relations in .
however , it was not appeared in our results .
the presumable reason is that the comptonized component is relatively weak on hb in .
since the maximum of the comptonized component occupies only 20% of the multi - color disk component in table [ tbl-3 ] of our fitting results as well as in @xcite and @xcite , it could not strongly effect the @xmath56 relation . from the energy spectral fitting by model
i , the box 1 of interval a had minimum inner radius , which was roughly @xmath57 with 15% distance uncertainty accounted .
causality , the radius of ns should be smaller than the minimum inner radius in accretion disk system .
therefor , we utilized the minimum inner radius of accretion disk to constrain the upper radius of ns .
the deduced inner radius depends on the inclination angle , which could not be directly measured in so far .
however , the lack of dips / eclipses in the light curve of provided an upper limit of the inclination angle to a value close to 75 degrees @xcite . here , we plotted the upper limit of ns radius for a wide range of inclination angle in figure [ fig11 ] .
we computed the characteristic frequencies for the hb in intervals a and b and the hb / nb vertex in stages i - ii .
the results accompany with data from @xcite displayed in figure [ fig14 ] . at low frequency ( @xmath58 hz ) , the w - k relation of locates in the atoll sources region
. however , at high frequency ( @xmath59 hz ) , the w - k relation of happens to locating between the z sources and the atoll sources .
that is , follows a linear correlation between the break frequency and the hbo , without offset appearing .
we studied the hbo - total flux dependence in both cyg - like phase and sco - like phase .
the total flux was integrated in the energy range @xmath63 - 60 kev . in figure
[ fig13 ] , the positive correlations between the frequency of hbo and the total flux were displayed .
compared with the sco - like phase , the cyg - like phase showed more complicated correlation , that is , the frequency of hbo increases gradually at lower flux and smoothly at higher flux . in order to investigate the energy dependence of hbo in interval a , we extracted six sub - bands and one full band lighcurves , and then fitted their pdss .
when the @xmath42 is larger than 30 hz , we could not obtain high signal to noise ratio qpos in channel 0 - 10 , 11 - 13 and 36 - 149 . in our analysis , only upper hb regions ( i.e. , @xmath60 )
were returned .
the centroid frequency of full band and six subbands were derived . in left panels of figure [ fig15 ] ,
the qpo decreases with increasing photon energy . by using the bootstrap procedure , the correlation coefficients between the hbo frequency and energy
are @xmath61 , @xmath62 , @xmath63 , @xmath64 , @xmath65 for boxes 1 - 5 in interval a , respectively .
all boxes show negative hbo - enenrgy correlations , but the boxes 1 and 5 have relatively large errors .
the hbo - energy relation in is distinct from the black hole transit xte j1550 - 564 , in which displayed more complicated correlations between type c qpo frequency and photon energy @xcite . for each subband ,
the rms of hbo are also computed , which positively correlate with photon energy in five lower energy bands and slightly drop at highest energy band in right panels of figure [ fig15 ] .
the unique transient ns lmxb emitted from near eddington luminosity to quiescence .
the accretion rate @xmath7 played a leading role in the whole process . for
, the nb / fb vertex was considered as reference position to trace accretion rate .
however , we could not confirm the strength of accretion rate only from the source intensity in a single z track .
the spectral state should be analyzed simultaneously .
homan and co - workers @xcite argued that the accretion rate maintains constant along tracks in z stage of . @xcite considered the consequence of magnetic field during accretion and derived that @xmath66 . in this paper , we simultaneously studied the timing properties and spectral states on the relative regions ( the hb / nb vertex and the nb / fb vertex ) when the accretion rate decreased .
we implemented a detailed temporal and spectral analysis of the hb / nb vertex and the nb / fb vertex in figure [ fig4 ] .
the hb / nb vertex and the nb / fb vertex trace two distinct lines in hid .
their energy spectra can be well fitted by * wabs(diskbb+bbodyrad+gauss ) * without comptonization component . the inner radius and the apparent radius of ns can be deduced from normalizations of * diskbb * and * bbodyrad * respectively .
the hb / nb vertices display @xmath67 relation , where the nb / fb vertices show a flatter trend , that is , @xmath68 .
the hb / nb vertices have a slightly steeper @xmath69 relation compared with the nb / fb vertices , which indicate the accretion disk evolves from a slim disk to a standard disk @xcite because of decreasing accretion rate .
since the disk accretion rate is proportional to @xmath70 , the disk accretion rate decreases with the luminosity decaying for the hb / nb vertices .
the hb / nb vertices of stages i and ii present near 50 hz hbos with luminosity variation .
we firstly conclude that the frequency of hbo is independent of the accretion rate or the source intensity , which might be applied in other z sources .
this conclusion is obtained without prior model restriction .
the radius - mass relation of ns can be constrained from type i x - ray burst@xcite , khz qpo@xcite , x - ray emission of quiescent lmxbs in globular clusters@xcite , emission lines@xcite and so on . in , the inner disk radius depends on the inclination angle of the binary system .
although , the inclination angle of has not measured directly until now , we still plot the inner radius versus inclination angle in figure [ fig11 ] .
when the accretion disk moves closest to the ns , the deduced inner radius must larger than the ns radius .
this figure shows the upper limit of ns radius . for 70 degree of inclination angle ,
the upper limit of ns radius is @xmath71 km , which covers the radius confidence intervals from type i x - ray burst @xcite and quiescent lmxbs@xcite .
the w - k relation between break frequency and lf - qpo was initially found in the bhs and atoll sources , as well as in z sources but with a slightly shift @xcite .
it spanned nearly three orders of magnitude in frequency .
recently , the accreting millisecond pulsars were also found following this correlation @xcite .
the w - k relation implies that the similar physical mechanism likely happens in both bh - lmxbs and ns - lmxbs . the w - k relation in bh - lmxbs ,
ns - lmxbs and accreting millisecond pulsars demonstrate that the intrinsic magnetic field and the solid surface are probably not the reason for it .
the w - k relation in may constrain the potential physics process .
, in the cyg - like phase , sco - like phase and hb / nb vertex , follows the w - k relation without shifted like other typical z sources or dropped at high break frequency like atoll sources .
more interesting , the low characteristic frequencies locate in the atoll sources region . as mentioned above , the accretion rate in hb / nb vertex decreases towards lower source intensity in figure [ fig4 ] . the well correlated w - k relation of hb / nb vertex in indicates that the characteristic frequencies on the hb are also not determined by the accretion rate .
some other mechanisms , e.g. , the propagation of accretion flow instabilities @xcite , are presumably responsible for it . the origin of lf - qpo is still puzzling .
the lense - thirring precession has been suggested as the explanation of the hbo in z sources , the low frequency qpo in atoll sources and the typ - c qpo in bh lmxbs .
the @xmath61 hz qpo in the dipping / eclipsing ns - lmxb was considered as a likely signature of lense - thirring @xcite .
the lense - thirring precession predicted a @xmath72 relation @xcite arose in when the comptonized component was fitted by nthcompt @xcite .
we investigated the @xmath56 relation in z stage of . whether the comptonized component fitted by the cutoff power law or nthcompt , the nearly identical @xmath56 relations were obtained in figure [ fig12 ] .
that is the @xmath42 positively correlate with the inner disk radius @xmath26 .
the possible reason of dissimilar with is that the relative weak comptonized component in unlikely affected the fitting results of inner disk radius . in bh lmxbs
, the energy dependence of type c qpo shows complicated correlations , i.e. , @xcite , @xcite and @xcite . in figure
[ fig15 ] , we studied the energy dependence of hbo in . on the hb ,
the frequency of hpo is anticorrelated with photon energy , which is similar to the bh lmxb grs 1915 + 105 with lf - qpos @xmath73 hz @xcite . in
, the hbo frequency is @xmath610 - 55 hz .
we only obtained the hbo frequency - energy relation with hbo @xmath74 hz . because at higher hbo frequency
, we could not have high s / n pds to constrain the hbo properly in some sub - bands .
if we have high s / n pdss above 30 hz in some other ns lmxbs , more complicated hbo frequency - energy relations could probably appear .
although the mechanism of type c qpo frequency - energy relation is still not known yet , the difference between the bh lmxbs and ns lmxbs may due to two reasons .
the first is the affection of emission from ns surface which can cool down the corona comparable to the disk emission .
the second is the central mass dependence of characteristic frequencies in the accretion flow as suggested by @xcite . in the right panels of figure [ fig15 ] , the rms of hbo increases with photon energy , from @xmath75 in 2 - 4.5 kev to @xmath76 in [email protected] kev , which indicates that the higher energy photon provide more contribution of hbo . from our fitting results ,
the temperature of black body from ns surface is about @xmath78 and the temperature of multi - color disk is about @xmath79 , which are much less than @xmath80 of the maximum rms of hbo occurring .
the amplitude of hbos as well as hard color dropped from upturn to hb @xcite , meanwhile , the strengh of the comptonization emission also became fainter from our spectral fitting results . on the nb , the inverse compton process turned into undetectable , while hbo disappeared simultaneously .
the hb / nb vertex could be considered as a spectral state transition zone , where the spectral switched from a non - thermal emission contained state to a thermal emission completely dominated state .
this feature of spectral transition also exhibited in gx 17 + 2 .
the spectral state change of gx 17 + 2 occurred on the nb / fb vertex , moreover , the hbo exhibit on both hb and nb . on fb , the comptonization component became detectable marginally and hbos disappeared similar as in the case of xte j1701 - 462 @xcite .
the non - thermal emission connection of hbo were researched in cyg x-2 and 4u 1728 - 34 , which the photon indices of cyg x-2 and 4u 1728 - 34 increased with hbo @xcite .
if the normalization of comptoniztion component did not change significantly , the hbo in cyg x-2 and 4u 1728 - 34 were also anti - correlated with comptonization emission strength . based on the above observational evidences , we interpret that the hbo possibly arise in the corona , where the hbo generate in the process of comptonization emission @xcite . the frequency and rms of hbo maybe associate with the scale as well as the optical depth of corona region , respectively .
this kind of dependence needs further observation constraining .
we study the hbos of with decreasing accretion rate in the hb / nb vertex .
we conclude that the hbo in , unlike the type c lf - qpos in bh lmxbs , is independent with accretion rate . in other word
, the ascending of hbo is not representing the accretion rate increasing .
we also find that the anti - correlation relation between the frequency of hbo and its centroid energy .
the energy dependence of hbo implies that the higher qpo produced a disk moves away from the ns .
both the cyg - like phase and the sco - like phase follow the w - k relation which are presumably caused by the same mechanism .
the derived @xmath81 relations contradict the prediction of lense - thirring precession .
we conclude that the hbo may origin from the corona .
we appreciate the referee s insightful suggestions and comments which improve our works distinctly . z.s .
li thank s.n .
zhang for critical remarks , z.b .
li for helpful discussions .
this work is supported by the national natural science foundation of china ( 10778716 ) , the national basic research program of china 973 program 2009cb824800 , the national natural science foundation of china ( 11173024 ) and the fundamental research funds for the central universities .
this research has made use of data obtained from the high energy astrophysics science archive research center ( heasarc ) , provided by nasa s goddard space flight center .
cllllllllc interval a & & & & & & & & & + 1 & @xmath82 & @xmath83 & @xmath84 & @xmath85 & @xmath86 & @xmath87 & @xmath88 & @xmath89 & 51.82/52 + 2 & @xmath90 & @xmath91 & @xmath92 & @xmath93 & @xmath94 & @xmath95 & @xmath96 & @xmath97 & 54.04/52 + 3 & @xmath98 & @xmath99 & @xmath100 & @xmath101 & @xmath102 & @xmath103 & @xmath104 & @xmath105 & 41.82/53 + 4 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & @xmath110 & @xmath111 & @xmath112 & @xmath113 & 35.38/52 + 5 & @xmath114 & @xmath115 & @xmath116 & @xmath101 & @xmath117 & @xmath118 & @xmath119 & @xmath120 & 23.00/52 + 6 & @xmath121 & @xmath122 & @xmath123 & @xmath124 & @xmath125 & @xmath126 & @xmath127 & @xmath128 & 43.19/52 + 7 & @xmath129 & @xmath130 & @xmath131 & @xmath132 & @xmath133 & @xmath134 & @xmath135 & @xmath136 & 36.05/52 + 8 & @xmath137 & @xmath138 & @xmath139 & @xmath140 & @xmath141 & @xmath142 & @xmath143 & @xmath144 & 41.71/52 + 9 & @xmath145 & @xmath146 & @xmath147 & @xmath148 & @xmath149 & @xmath150 & @xmath151 & @xmath152 & 27.20/52 + 10 & @xmath153 & @xmath154 & @xmath155 & @xmath156 & @xmath157 & @xmath142 & @xmath158 & @xmath159 & 32.34/52 + 11 & @xmath160 & @xmath161 & @xmath162 & @xmath163 & @xmath164 & @xmath165 & @xmath166 & @xmath167 & 58.19/52 + 12 & @xmath168 & @xmath169 & @xmath170 & @xmath171 & @xmath172 & @xmath173 & @xmath174 & @xmath175 & 50.64/52 + 13 & @xmath176 & @xmath177 & @xmath178 & @xmath179 & @xmath180 & @xmath181 & @xmath182 & @xmath183 & 45.24/48 + interval b & & & & & & & & & + 1 & @xmath184 & @xmath185 & @xmath186 & @xmath187 & @xmath188 & @xmath189 & @xmath190 & @xmath191 & 35.92/52 + 2 & @xmath192 & @xmath193 & @xmath194 & @xmath195 & @xmath196 & @xmath197 & @xmath198 & @xmath199 & 44.40/52 + 3 & @xmath200 & @xmath201 & @xmath202 & @xmath203 & @xmath204 & @xmath205 & @xmath206 & @xmath207 & 37.11/52 + 4 & @xmath208 & @xmath209 & @xmath210 & @xmath211 & @xmath212 & @xmath213 & @xmath214 & @xmath215 & 30.84/52 + 5 & @xmath216 & @xmath217 & @xmath218 & @xmath203 & @xmath219 & @xmath220 & @xmath221 & @xmath222 & 41.46/52 + 6 & @xmath223 & @xmath224 & @xmath225 & @xmath226 & @xmath227 & @xmath228 & @xmath229 & @xmath230 & 29.27/52 + 7 & @xmath231 & @xmath232 & @xmath233 & @xmath211 & @xmath234 & @xmath235 & @xmath236 & @xmath237 & 35.51/52 + cllllllllc interval a & & & & & & & & & + 1 & @xmath238 & @xmath239 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & @xmath244 & @xmath245 & 55.67/52 + 2 & @xmath246 & @xmath247 & @xmath248 & @xmath249 & @xmath250 & @xmath251 & @xmath252 & @xmath253 & 54.44/52 + 3 & @xmath254 & @xmath255 & @xmath256 & @xmath257 & @xmath258 & @xmath259 & @xmath260 & @xmath261 & 55.14/53 + 4 & @xmath262 & @xmath263 & @xmath264 & @xmath265 & @xmath266 & @xmath267 & @xmath268 & @xmath269 & 38.60/52 + 5 & @xmath270 & @xmath271 & @xmath272 & @xmath273 & @xmath274 & @xmath275 & @xmath276 & @xmath277 & 25.63/52 + 6 & @xmath278 & @xmath279 & @xmath280 & @xmath281 & @xmath282 & @xmath283 & @xmath284 & @xmath285 & 42.23/52 + 7 & @xmath286 & @xmath287 & @xmath288 & @xmath289 & @xmath290 & @xmath291 & @xmath292 & @xmath293 & 57.95/52 + 8 & @xmath294 & @xmath295 & @xmath296 & @xmath297 & @xmath298 & @xmath299 & @xmath300 & @xmath301 & 28.87/52 + 9 & @xmath302 & @xmath303 & @xmath304 & @xmath305 & @xmath306 & @xmath307 & @xmath308 & @xmath309 & 38.49/52 + 10 & @xmath310 & @xmath311 & @xmath312 & @xmath313 & @xmath314 & @xmath315 & @xmath316 & @xmath317 & 33.28/52 + 11 & @xmath318 & @xmath319 & @xmath320 & @xmath305 & @xmath321 & @xmath307 & @xmath322 & @xmath323 & 58.64/52 + 12 & @xmath324 & @xmath325 & @xmath326 & @xmath327 & @xmath328 & @xmath329 & @xmath330 & @xmath331 & 50.93/52 + 13 & @xmath332 & @xmath333 & @xmath334 & @xmath335 & @xmath336 & @xmath337 & @xmath338 & @xmath339 & 50.70/48 + interval b & & & & & & & & & + 1 & @xmath340 & @xmath341 & @xmath342 & @xmath343 & @xmath344 & @xmath345 & @xmath346 & @xmath347 & 37.17/52 + 2 & @xmath348 & @xmath349 & @xmath350 & @xmath351 & @xmath352 & @xmath353 & @xmath244 & @xmath354 & 56.23/52 + 3 & @xmath355 & @xmath356 & @xmath357 & @xmath358 & @xmath359 & @xmath360 & @xmath361 & @xmath362 & 37.67/52 + 4 & @xmath363 & @xmath364 & @xmath365 & @xmath366 & @xmath367 & @xmath368 & @xmath369 & @xmath370 & 33.10/52 + 5 & @xmath371 & @xmath372 & @xmath373 & @xmath374 & @xmath375 & @xmath376 & @xmath377 & @xmath378 & 43.26/52 + 6 & @xmath379 & @xmath380 & @xmath381 & @xmath366 & @xmath382 & @xmath383 & @xmath384 & @xmath385 & 29.02/52 + 7 & @xmath386 & @xmath387 & @xmath388 & @xmath389 & @xmath390 & @xmath391 & @xmath392 & @xmath393 & 35.53/52 + llll interval a & & & + @xmath394 & @xmath395 & @xmath396 & @xmath397 + @xmath398 & @xmath399 & @xmath400 & @xmath401 + @xmath402 & @xmath403 & @xmath404 & @xmath405 + @xmath406 & @xmath407 & @xmath408 & @xmath409 + @xmath410 & @xmath411 & @xmath412 & @xmath413 + @xmath414 & @xmath415 & @xmath416 & @xmath417 + @xmath418 & @xmath419 & @xmath420 & @xmath421 + @xmath422 & @xmath423 & @xmath424 & @xmath425 + @xmath426 & @xmath427 & @xmath428 & @xmath429 + @xmath430 & @xmath431 & @xmath432 & @xmath433 + @xmath434 & @xmath435 & @xmath436 & @xmath437 + @xmath438 & @xmath439 & @xmath440 & @xmath441 + @xmath442 & @xmath443 & @xmath444 & @xmath445 + interval b & & & + @xmath446 & @xmath447 & @xmath448 & @xmath449 + @xmath450 & @xmath451 & @xmath452 & @xmath453 + @xmath454 & @xmath455 & @xmath456 & @xmath457 + @xmath458 & @xmath459 & @xmath460 & @xmath461 + @xmath462 & @xmath463 & @xmath464 & @xmath465 + @xmath466 & @xmath467 & @xmath468 & @xmath469 + @xmath470 & @xmath471 & @xmath472 & @xmath473 + cccc stage i & & & + @xmath474 & @xmath475 & @xmath476 & @xmath477 + @xmath478 & @xmath479 & @xmath480 & @xmath481 + @xmath482 & @xmath483 & @xmath484 & @xmath485 + @xmath486 & @xmath487 & @xmath488 & @xmath489 + @xmath490 & @xmath491 & @xmath492 & @xmath493 + @xmath494 & @xmath495 & @xmath496 & @xmath497 + @xmath498 & @xmath499 & @xmath500 & @xmath501 + @xmath502 & @xmath503 & @xmath504 & @xmath505 + @xmath506 & @xmath507 & @xmath508 & @xmath509 + stage ii & & & + @xmath510 & @xmath511 & @xmath512 & @xmath513 + @xmath514 & @xmath515 & @xmath516 & @xmath517 + @xmath518 & @xmath519 & @xmath520 & @xmath521 + @xmath522 & @xmath523 & @xmath524 & @xmath525 + @xmath526 & @xmath527 & @xmath528 & @xmath529 + @xmath530 & @xmath531 & @xmath532 & @xmath533 + @xmath534 & @xmath535 & @xmath536 & @xmath537 + @xmath538 & @xmath539 & @xmath540 & @xmath541 + @xmath542 & @xmath543 & @xmath544 & @xmath545 + @xmath546 & @xmath547 & @xmath548 & @xmath549 + @xmath550 & @xmath551 & @xmath552 & @xmath553 + @xmath554 & @xmath555 & @xmath556 & @xmath557 + clllllll stage i & & & & & + @xmath558 & @xmath559 & @xmath560 & @xmath561 & @xmath562 & @xmath563 & @xmath564 & @xmath565 + @xmath566 & @xmath567 & @xmath568 & @xmath569 & @xmath570 & @xmath571 & @xmath572 & @xmath573 + @xmath574 & @xmath575 & @xmath576 & @xmath577 & @xmath578 & @xmath579 & @xmath580 & @xmath581 + @xmath574 & @xmath582 & @xmath583 & @xmath584 & @xmath585 & @xmath586 & @xmath587 & @xmath588 + @xmath574 & @xmath589 & @xmath590 & @xmath591 & @xmath592 & @xmath563 & @xmath593 & @xmath594 + @xmath574 & @xmath595 & @xmath596 & @xmath597 & @xmath598 & @xmath599 & @xmath600 & @xmath601 + @xmath602 & @xmath603 & @xmath604 & @xmath605 & @xmath606 & @xmath607 & @xmath608 & @xmath609 + @xmath602 & @xmath610 & @xmath611 & @xmath612 & @xmath613 & @xmath563 & @xmath614 & @xmath615 + @xmath602 & @xmath616 & @xmath617 & @xmath618 & @xmath619 & @xmath620 & @xmath621 & @xmath622 + stage ii & & & & & + @xmath623 & @xmath624 & @xmath625 & @xmath626 & @xmath627 & @xmath628 & @xmath629 & @xmath630 + @xmath631 & @xmath632 & @xmath633 & @xmath634 & @xmath635 & @xmath636 & @xmath637 & @xmath638 + @xmath639 & @xmath640 & @xmath633 & @xmath634 & @xmath641 & @xmath642 & @xmath643 & @xmath644 + @xmath645 & @xmath646 & @xmath625 & @xmath647 & @xmath648 & @xmath649 & @xmath650 & @xmath651 + @xmath652 & @xmath653 & @xmath654 & @xmath626 & @xmath655 & @xmath636 & @xmath656 & @xmath657 + @xmath658 & @xmath659 & @xmath660 & @xmath661 & @xmath662 & @xmath628 & @xmath663 & @xmath664 + @xmath665 & @xmath666 & @xmath667 & @xmath668 & @xmath669 & @xmath670 & @xmath637 & @xmath671 + @xmath672 & @xmath673 & @xmath674 & @xmath668 & @xmath675 & @xmath676 & @xmath677 & @xmath678 + @xmath679 & @xmath680 & @xmath681 & @xmath682 & @xmath683 & @xmath684 & @xmath685 & @xmath686 + @xmath665 & @xmath687 & @xmath688 & @xmath689 & @xmath690 & @xmath691 & @xmath692 & @xmath693 + @xmath694 & @xmath695 & @xmath625 & @xmath696 & @xmath697 & @xmath698 & @xmath699 & @xmath700 + @xmath672 & @xmath701 & @xmath633 & @xmath668 & @xmath702 & @xmath636 & @xmath703 & @xmath704 + @xmath679 & @xmath705 & @xmath706 & @xmath682 & @xmath707 & @xmath708 & @xmath709 & @xmath710 + bardeen , j.m . , &
petterson , j.a . , 1975 , , 195 , l65 belloni t. , psaltis , d. , & van der klis , m. , 2002 , , 572 , 392 belloni t. , homan , j. , casella p. , et al . , 2005 , , 440 , 207 belloni t. et al . , 2006 , , 369 , 305 belloni t. , motta , s.e . ,
& muoz - darias , 2011 , bull . astr .
india , 39 , 409 casella , p. , belloni , t. , & stella , l. , 2005 , , 629 , 403 church et al .
, 2006 , , 460 , 233 church et al . , 2010 , , 512 , a9 cottam , j. , paerels , f. , & mendez , m. , 2002 , , 420 , 51 ding , g. q. , zhang , s. n. , wang , n. , qu , j. l. , & yan , s. p. 2011
, , 142 , 34 gilfanov m. , & revivtsev m. , 2005 , astro .
nachr . , 326 , 812 guillot s , servillat m , webb n a , et al . , 2013 , , 2013 , 772 , 7 hasinger , g. , & van der klis , m. 1989 , , 225 , 79 hasinger g. , van der klis m. , ebisawa k. , dotani t. , & mitsuda k. , 1990 , 235 , 131 homan , j. , et al .
, 2002 , , 568 , 878 homan , j. et al .
, 2007 , , 656 , 420 homan , j. et al .
, 2010 , , 719 , 201 homan , j. 2012 , , 760 , l30 ingram , a. , done , c. , & fragile , p.c . , 2009 , 397 , l101 kalemci e. , tomsick j. a. , rothschild r. e. , pottschmidt k. , corbel s. , kaaret p. , 2006
, , 639 , 340 lattimer , j. m. , & steiner , a. w. , 2013 , arxiv:1305.3242 lense , j.&thirring , h. , 1918 , phys.z . , 19 , 156 lin , d. c. , remillard , r. a. , & homan , j. , 2009 , , 696 , 1257 lin , d. , altamirano , d. , homan , j. , remillard , r. a. , wijnands , r. , & belloni , t. , 2009 , , 699 , 60 .
lin , d. c. , remillard , r. a. , homan , j. , & barret , d. , 2012 , , 756 , 34 li , z. b. , qu , j. l. , song , l. m. , ding , g. q. , zhang , c. m. , 2013 , , 428 , 1704 li , z. b. , zhang , s. , qu , j. l. , gao , h. q. , zhao , h. h. , huang , c. p. , & song , l. m. , 2013 , , 431 , 2285 li , z. s. , chen l , qu j.l , et al .
, 2013 , , 767 , 167 .
qu , j. l. et al . , 2010 , , 710 , 836 remillard , r. a. , & lin , d. , the asm team at mit and nasa / gsfc 2006 , astron .
, 696 sala , g. , haberl , f. , jose , j. , & parikh , a. et al . , 2012 , , 752 , 158 stella , l. , & vietri , m. , 1998 , , 492 , l59 stella , l. , vietri , m. , & morsink , s.m . , 1999 , , 524 , l63 sunyaev , r. , & revnivtsev , m. , 2000 , , 358 , 617 titarchuk , l. and shaposhnikov , n. , 2005 , , 626 , 298 titarchuk , l. , kuznetsov , s. , shaposhnikov , n. , 2007 , , 667 , 404 van straaten s. , van der klis m. , & wijnands , r. , 2005 , , 619 , 455 van der klis , m. , hasinger , g. , stella , l. , langmeier , a. , van paradijs , j. , lewin , & w. h. g. 1987 , , 319 , l13 vrtilek , s. d. , et al . 1990 , , 235 , 162 wijnands , r. , & van der klis , m. , 1999 , , 514 , 939 yan , s.p . , ding , g.q . , wang , n . , et al . , 2013 , ,
1739 zhang c.m .
, 2009 , astronomische nachrichten , 330 , 398 | we study the temporal and energy spectral properties of the unique neutron star low - mass x - ray binary . after assuming the hb / nb vertex as a reference position of accretion rate ,
the horizontal branch oscillation ( hbo ) of the hb / nb vertex is roughly 50 hz .
it indicates that the hbo is independent with the accretion rate or the source intensity .
the spectral analysis shows @xmath0 in the hb / nb vertex and @xmath1 in the nb / fb vertex , which implies that different accretion rate may be produced in the hb / nb vertex and the nb / fb vertex .
the comptonization component could be fitted by constrained broken power law ( cbpl ) or nthcomp .
different with , the frequencies of hbo positively correlate with the inner disk radius , which contradict with the prediction of lense - thirring precession model .
, both in the cyg - like phase and in the sco - like phase , follows a positive correlation between the break frequency of broad band noise and the characteristic frequency of hbo , which is called the w - k relation .
an anticorrelation between the frequency of hbo and photon energy is observed .
moreover , the rms of hbo increases with photon energy till @xmath2 .
we discuss the possible origin of hbo from corona in . |
it is well established that interactions and/or galaxy collisions represent an important stage in the evolution of galaxies ( e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
numerous studies show clear indications of the importance of those external mechanisms for the enhancement of star - formation and its effects in the chemical enrichment of galaxies .
in particular , in major merger of massive galaxies the preexisting gas metallicity can be substantially diluted by the inflow of metal poor gas from the outskirts to the nucleus ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in the case of low - mass , low - metallicity ( 7.0 @xmath6 12 + log(o / h ) @xmath6 8.4 ; * ?
* ) and star - forming dwarf galaxies the effects of tidal interactions and/or mergers ( e.g. * ? ? ?
* ; * ? ? ?
* ) also has a huge impact in their evolution .
observational evidences suggest that hii / bcd galaxies arise from the interactions or accretion of extended hi cloud complexes ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . however , the triggering mechanisms of the current burst of star - formation in those objects is not yet clear since most are , apparently , isolated systems ( e.g. * ? ? ?
* ; * ? ? ?
thus , if not triggered by external agents star - formation is likely produced by internal processes ( e.g. gravitational cloud collapse , infall of gas in conjunction with small perturbations ) and/or minor mergers ( see * ? ? ? * and references therein ) . as described above
, a considerable fraction of these galaxies has been associated with hi clouds @xcite or low - mass and undetected companions in the optical ( e.g. * ? ? ?
* ) , which could rule out the idea of bcd galaxies as isolated systems @xcite .
in fact , a significant fraction of bcds do show signs of extensions or tails in their outer envelopes , suggesting a tidal origin .
many of these low - metallicity galaxies that show cometary " or elongated shapes show values of 12+log(o / h ) @xmath7 7.6 ( e.g. * ? ? ?
* ) . within this subsample of bcds or extremely metal poor ( xmp ) bcd galaxies we found the
least chemically evolved galaxies in the local universe @xcite .
this particular morphology has been interpreted for high redshift galaxies in the hubble deep field as the result of weak tidal interactions @xcite , gravitational instabilities in gas - rich and turbulent galactic disks in formation at high redshift @xcite and stream - driven accretion of metal - poor gas from the cosmic web @xcite .
@xcite argue that weak interactions between low - mass stellar or gaseous companions , or propagating shock waves , lead to a bar - like gas distribution triggering the star - formation that by propagation could subsequently produce a cometary morphology in xmp bcds .
recently , @xcite interpret the metallicity variation in a sample of low metallicity galaxies with cometary morphology as a sign of external gas accretion / infall of metal poor gas .
they argue that these results are consistent with the local tadpole " galaxies being disks in early stages of assembling , with their star - formation sustained by pristine gas infall . in any case ,
dwarf galaxies tend to show flat abundance ( o / h ,
n / o ) gradients ( e.g. * ? ? ?
? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , suggesting efficient dispersion and mixing of metals in the interstellar medium ( ism ) by expanding starburst - driven superbubbles ( e.g. * ? ? ?
* ) , afterward the gas begins to cool down by radiation and gravity , and/or external gas infall ( e.g. * ? ? ?
* ; * ? ? ?
these mechanisms have been put forth as potential causes for the observed flat metal distributions in local dwarf galaxies .
while in massive star - forming and/or interacting galaxies , bar - induced rotation or shear ( e.g. * ? ?
? * ) and merger - induced gas flows ( e.g. * ? ? ?
* ) could produce the metal dispersal and mixing . as expressed above ,
local hii / bcd and xmp bcd galaxies are considered chemically homogeneous and only in a few isolated cases we observed indications of variation of 12+log(o / h ) over the ism ( e.g. sbs 0335 - 052e , haro 11 , hs 2236 + 1344 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in addition to the expected low metal content in young galaxies at high redshift , according to theoretical models these objects should produce strong ly@xmath0 ( 1216 @xmath8 ) emission as the result of their intense star - formation activity ( e.g. * ? ? ? * ; * ? ? ?
however , the absence and/or diminished ly@xmath0 emission in these galaxies , which is significantly lower than the theoretical recombination ratio , indicate that the ly@xmath0 photons are likely redistributed by multiple scattering in the hi envelope , or are absorbed by dust
. examples of the detection of ly@xmath0 halos produced by hi scattering envelopes can be seen in the literature ( e.g. * ? ? ?
* ; * ? ? ?
it has been suggested in the literature ( e.g. * ? ? ? * ) that there should be an increase in the ly@xmath0/h@xmath9 flux ratio as the metallicity of the galaxy decreases , since presumably low - metallicity objects contain less dust and hence suffer less ly@xmath0 photon destruction .
ly@xmath0 can also be enhanced at low gas metallicity due to collisional excitation @xcite . as pointed by @xcite , the fact that there is no ly@xmath0 emission in the two most metal - deficient bcds known ,
i zw 18 @xcite and sbs 0335 - 052 @xcite , and also in tol 65 , argues against the existence of some correlation between the ly@xmath0 emission , metallicity and dust @xcite .
therefore , an important issue in the understanding of ly@xmath0 emission in galaxies is the study of the spatial distribution of properties in the ism of those objects in order to see the different regulation mechanisms involved in the detectability of ly@xmath0 emission .
although during the last years some progress has been made in this field , many questions remain about the metal content , the mechanism involved in the transport and mixing of metals and the star - formation activity in hii / bcd and xmp galaxies .
the morphologically diverse nature of hii / bcd , and xmp bcds , galaxies allows us to consider the role played by galaxy interactions and the feedback between the star - formation and the ism in the observed metal distributions .
our main objective in this paper is to carry out a spatial investigation of the warm gas properties in the cometary " and ly@xmath0 absorbing xmp hii / bcd galaxy tol 65 .
these include , the spatial distribution of emission lines , equivalent width ew(h@xmath9 ) , extinction c(h@xmath9 ) , ionization ratios ( [ oiii]@xmath105007/h@xmath9 , [ sii]@xmath116717,6731/h@xmath0 and [ nii]@xmath106584/h@xmath0 ) , kinematics , the chemical pattern ( e.g. o / h , n / h and n / o , etc . ) and also the possible dependence between these properties .
our aim is to search for the existence of metallicity inhomogeneities as expected if the ongoing star - formation activity is sustained by the infall or accretion of metal - poor gas . to this end , we use high resolution integral field unit ( ifu ) spectroscopy observations . the paper is organized as follows : sect .
[ sect_tol65 ] describes the most important properties of our analyzed galaxy .
[ sect_obser_reduc ] contains the technical details regarding observations and the data reduction .
[ sect_results ] describes our results : the ionized gas structure as well as the physical and kinematic properties of the ionized gas .
[ sect_discussion ] discusses the results .
finally , sect .
[ conclusions ] itemizes the conclusions .
[ cols="<,^,^,^,^,^,^,>,<,>",options="header " , ] in this context , an estimate of the mass of emitting ionized hydrogen in the tail of the galaxy , is an important indicator of the amount of material present in this structure .
this can be made using the h@xmath0 emission - line luminosity l(h@xmath0)=7.14@xmath210@xmath12 erg s@xmath13 and the expression for the mass of the ionized gas from @xcite ; @xmath14 where m@xmath15 is the mass of the proton and n@xmath16 is the electron density .
we use the effective recombination coefficient @xmath17 for h@xmath0 and case b emission at t=2@xmath210@xmath18 k @xcite .
thus , assuming an average electron density of @xmath510 @xmath19 , we find a total mass of ionized hydrogen in the tail of m(hii)@xmath51.70@xmath210@xmath3 m@xmath20 .
this mass corresponds with @xmath524 per cent of the total ionized mass flux of the galaxy summing all the spaxels over the fov .
thus , f(h@xmath0)/f(h@xmath9)=3.4899 with f(h@xmath9)=47.48@xmath210@xmath21 erg @xmath22 s@xmath13 . ] of the galaxy ( or @xmath53 per cent
if we consider @xmath5100 @xmath19 ) with m(hii)@xmath57.00@xmath210@xmath3 m@xmath20 .
although this mass estimate is approximate , it could clearly indicate that there is a large amount of hi in this extended structure .
the star - formation rate ( sfr ) inferred using the @xcite formula after correction for a kroupa imf @xcite , is 0.152 m@xmath4yr@xmath13 for the main body of the galaxy and 0.090 m@xmath4yr@xmath13 and 0.016 m@xmath4yr@xmath13 for regions nos . 1 and 2 , respectively .
this global sfrs are low in absolute terms .
however , its starbursting nature appears when one computes the specific sfr , i.e. the sfr per unit of mass .
somewhat equivalently , their sfr per unit of area ( @xmath23 ) .
we can translate the integrated sfr of the main body into an integrated @xmath23 assuming an aperture of @xmath51.95 kpc@xmath24 that corresponds to the total area of this region .
thus , we obtain that @xmath250.078 m@xmath4 yr@xmath13kpc@xmath26 . if we compare this result with the one obtained in our previously analyzed xmp bcd galaxy , using ifu spectroscopy , ( hs 2236 + 1344 , * ? ? ?
* ) we found that in tol 65 the @xmath23 is 2 times lower , while the @xmath27 is comparable to the star - formation per unit of area of local tadpole galaxies @xcite . on the other hand ,
the @xmath23 in tol 65 , and also in hs 2236 + 1344 , is above the sfrs found in normal " or more metal rich hii / bcd galaxies . from fig .
15 in @xcite it is evident that the sfr per unit of area in our xmp galaxies studied so far are higher than the ones found in most of their sample of hii galaxies .
this is explained by the high concentration of hi gas found in xmps ( e.g. * ? ? ?
* ; * ? ? ?
* ) . using the results obtained in sect .
[ sect_abundances ] we show , in fig . [ figure_sfr_oh ] , the spatially resolved relation between the @xmath23 and the oxygen abundance 12+log(o / h ) in tol 65 .
even this relation at spaxel scales is dubious , our findings show a marginal anticorrelation between these quantities in the sense that high spatial star - formation is found in regions of lower metallicities ( see the linear fit to this relation in fig [ figure_sfr_oh ] ) .
the calculation of pearson s correlation for these data gives a value of -0.41 , that indicate a moderate monotonically decreasing relationship between these quantities .
if we interpret this result as the consequence of an ongoing infall of metal - poor gas from the outskirts the scatter ( @xmath28=0.13 dex ) in fig .
[ figure_sfr_oh ] clearly indicates that the metals in the ism are almost fully diluted .
therefore , from an statistical point of view the ism of the galaxy can be considered chemically homogeneous .
we will come back on this discussion later in sect .
[ sect_disc_properties ] . and
oxygen abundance 12+log(o / h ) at spaxel scales .
the cyan line represents the linear fit to this relation .
the 12+log(o / h)=7.56 obtained in the main body of the galaxy is represented by the continuous line , while the errors at 1@xmath29 level are in dotted lines.,width=321 ] it is however worth noting that the age ( @xmath53 - 5 myr ; see sect .
[ sect_i_c_ew ] ) of the current burst of tol 65 agrees with the idea proposed by @xcite , in the sense that this galaxy is a young starburst embedded in a static hi cloud , which produces a damped ly@xmath0 absorption .
@xcite found even with the poor quality of their spectrum , that the oi @xmath101302 absorption line in tol 65 is blue shifted by @xmath5200 km s@xmath13 with respect to the emission lines and the siii@xmath101304 absorption line .
they argue that if this is true , the oi absorption would be produced in gas shells moving outward from the central star clusters @xcite .
the spatial distribution of electron density n@xmath16(sii ) , in fig .
[ figure_temp_den_galaxies ] , shows a clear area of constant values in region no .
1 with n@xmath16(sii)@xmath5200 @xmath19 surrounded by relatively higher density values @xmath30200 @xmath19 . in particular
, the area in between the two regions shows spaxels which reach values of n@xmath16(sii)@xmath5500 @xmath19 and the [ sii]@xmath116717,6731/h@xmath0 ratio ( see fig .
[ figure_ratios_galaxies ] ) is slightly enhanced in this inter - cluster region which would be consistent with an enhanced contribution to the ionization / excitation by shocks .
therefore , the gas ejected by massive stars within the star clusters grows close to the centre creating a high density region at the same time that the gas expands with velocities close or higher than the supersonic values ( with @xmath3110 km s@xmath13 ) into the surrounding region , then compressing the gas into high - density condensations and creating a low density region between the centre and the expanding high density shells .
we found high fwhm values in the same area of high electron density in between the ghiirs ( see fig .
[ figure_velocity_field_tol65 ] ) indicating that the turbulent motions in the ionized gas could be produced by winds from the main star cluster complexes .
likewise , the expanding gas would cause the expansion in the surrounding material ( hi ) , so moving it out of the galactic plane .
this agrees with the arguments proposed by @xcite and the evolutionary models by @xcite in order to produce a damped ly@xmath0 absorption .
in addition , if the wind material contains enough dust , this outflow of gas would give rise to blueshifted velocities ( compared to the clusters ) observed in between the ghiirs because redshifted emission would be preferentially extinguished by the dust .
in fact , the lack of ly@xmath0 emission ( see fig . 2 in * ? ? ? * ) and the detection of relatively high c(h@xmath9 ) associated with the most intense star - forming region in this galaxy , region no .
1 , suggest that the ly@xmath0 photons have likely been destroyed by dust .
this is in good agreement with previous findings by @xcite in the sense that dust extinction is an important ly@xmath0 escape regulator .
a number of recent studies ( e.g. * ? ? ?
* ; * ? ? ?
* ) have suggested that local galaxies with cometary or tadpole morphology could have a variety of origins .
@xcite suggest that most local tadpoles are bulge - free galaxy disks with lopsided star - formation , likely from environmental effects ( e.g , ram pressure , disk impacts or random collapse of local disk gas with an unstable jeans length ) .
alternatively , propagating star - formation along the tail has been proposed by @xcite as one of the mechanisms that lead to the formation of the elongated underlying component or stellar tail of cometary hii / bcds .
given that the starburst produces enough energy from stellar winds and supernovae ( sne ) , could the stellar feedback from the starburst have created the stellar tail in this galaxy ? to create a @xmath51.5 kpc long extension in @xmath55 myr ( age of the current burst ) would require an outflow velocity @xmath5293 km s@xmath13 , while the difference between the velocities found in the field is @xmath5 50 km s@xmath13 .
therefore , the current star - formation activity appears to be too young to have pulled out the surrounding gas and stars , driven a galactic outflow , to form the tail . on the other hand ,
if the global starburst take longer @xmath510@xmath32 yr ( e.g. * ? ? ?
* ) as compared to the current burst , certainly it may promote the formation of gas outflows .
after that some part of the gas could decelerate and collapse to form the stellar tail .
if so , the close similarity between @xmath33 and @xmath34 agrees with this scenario , given that the diffuse gas will acquire the kinematical information of the system which is transferred into the ism by the star - formation activity ( e.g. * ? ? ?
@xcite found that the relatively red colors of the underlying stellar component , in tol 65 , suggest a stellar population with a mean age of @xmath351 gyr .
this stellar age is compatible with this idea , but the intrinsic youth of xmps is questionable @xcite and the evidence of such strong winds and the presence of extended ( @xmath361 kpc ) super - shells in low luminosity dwarf galaxies is also sparse . finally , we did not observe evidences of propagating star - formation along the galaxy and the presence of star clusters in the tail .
thus , there seems to be no correlation between the star - forming activity and the formation of the tail in this galaxy , but from our present data we can not rule out completely this possibility . in many respects ,
the tidal hypothesis is an attractive one for the origin of the tail in tol 65 , given that tidal interactions enhance lopsidedness in low mass galaxies @xcite .
however , hii / bcds tend to populate low - density environments and are not associated with massive galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ) suggesting that tidal interactions with massive companions could not be the dominant starburst triggering mechanism .
thus , the cometary or tadpole shapes among these may not generally be major mergers .
we used the nasa / ipac extragalactic database ( ned ) to search for nearby objects with measured systemic velocities .
figure [ figure_field_tol65 ] shows the field on the plane of the sky of the galaxy .
we find two galaxies , with measured velocities , in the field of tol 65 .
the first one j122531.50 - 360715.3 is an irregular ( dirr ) and likely dwarf galaxy with a difference between their respective systemic velocities of @xmath37v = -4 km s@xmath13 .
eso digitized sky survey ( dss ) images show that this object is highly disturbed with an apparent size of @xmath530@xmath38 equivalent with @xmath56.2 kpc , which is two times the apparent size of tol 65 .
the projected distance between these galaxies is r@xmath39100 kpc .
the other object in the field is the spiral galaxy eso 380-g029 . for this galaxy we found a @xmath37v = 1249 km s@xmath13 ,
thus it is unlikely that this galaxy is producing the tidally - disturbed tail present in the galaxy , because companions with @xmath37v@xmath30 500 km s@xmath13 have not a significant dynamical influence @xcite .
it is interesting to note that the tail of the galaxy is oriented , in the plane of the sky , to this spiral galaxy .
possibly eso 380-g029 had an stronger influence over tol 65 in the past ( several gyr ago ) , thus the tail in the latter could be a remnant of that tidal interaction .
however , there is no observational evidence showing that this galaxy had an interaction in the past with tol 65 to produce a tidal tail . @xcite using vlt images suggested the presence of a low - surface - brightness companion galaxy of tol 65 , but it was discarded by @xcite .
they found this object is in fact a background galaxy .
thus , if this extended structure has a tidal origin with a close companion(s ) it may be related with a faint and massive one that have not been identified by optical surveys . in any case , j122531.50 - 360715.3 remains as the only potential perturber . in this sense , the slight enhancement of the sfr in region no .
1 with respect to region no .
2 and the lopsidedness of those star - forming regions could be the consequence of this tidal interaction . although it is unlikely that another dwarf galaxy of low mass , as it seems , could have a tidal effect at this distance . from the present observations we can not say much about the impact of j122531.50 - 360715.3 over the physical properties of tol 65
but we speculate based on the aforementioned arguments that the co - evolution on these systems play an important role in the enhancement of star - formation ( e.g. * ? ? ?
* ) and evolution in many of these galaxies @xcite .
several studies have shown that a number of star - forming dwarf galaxies present filamentary hi and/or extended structures , which may indicate a recent cold gas infall / accretion ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and/or minor merger / interaction ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in this sense
, the extended structure of this galaxy could be a remnant of one of these processes .
@xcite argue that if the stars are being formed as a consequence of the infall they will be formed along the path of the accretion .
but , we did not observe the presence of star clusters in the tail of the galaxy . then , alternatively , the effect of these interactions has carried away in the recent past an important fraction of gas and stars to form the stellar tail . the fact that @xmath40 ( see sect .
[ sect_velocity_field ] ) implies that the ionized gas still , on a global scale , retains the kinematic memory of its parental cloud and likely a common origin . while extended regions of diffuse gas are highly disturbed , probably due to unresolved expanding shells and the effects of massive star evolution .
therefore , we interpreted the observed properties of the ism and the morphology of the galaxy as due to a late - stage minor merger and/or inflow of metal - poor gas in the recent past of the galaxy . if the current star - forming episode in xmps is the consequence of the infall of metal - poor gas , the less enriched gas dilutes the preexisting nuclear gas to produce a lower metallicity than would be obtained prior to the accretion .
@xcite argue that such accretion flows could explain all the major xmp properties such as , i.e. , isolation , lack of interaction / merger signatures , metal - poor hi gas , metallicity inhomogeneities , etc .
in this sense , @xcite argue that the variation of oxygen abundance along the major axis of two xmp star - forming dwarf galaxies can be interpreted as an early stage of assembling in disk galaxies with the star - formation sustained by external metal - poor gas accretion likely from the cosmic web @xcite . in @xcite
we studied one of the objects analyzed by @xcite , the xmp galaxy hs 2236 + 1344 . in this study
, we found indications of variation of o and n / o associated with the less luminous star - formation region of that galaxy . but given the uncertainties related to those measurements we considered the abundances across the galaxy as fairly uniform .
since most of our hii / bcd galaxies @xcite , and xmp / bcds from the literature ( * ? ? ?
* see references therein ) studied so far are chemically homogeneous it seems that these galaxies have mixed up the metals with the pre - existing ism before the current starburst appears .
the newly synthesized metals from the current star - formation episode reside in a hot gas phase ( t @xmath41 10@xmath42 k ) and those will be dispersed probably in the whole galaxy by the expansion of starburst - driven outflows .
the energy injected by stellar winds and sne may be able to eject part of the enriched gas into the intergalactic medium , but in most cases the expanding velocity ( e.g. * ? ? ? * and references therein ) are not sufficient to allow the gas to escape from the potential well of the galaxies . in addition ,
inflows or accretion of relatively low metallicity gas could possibly dilute more easily the abundance of the gas on large scales . assuming a sound speed of @xmath512 km s@xmath13 ( see sect . [ sect_velocity_field ] ) as the expected speed
at which mixing occurs in tol 65 , we found that over 1.5 - 2.0 kpc it yields @xmath510@xmath32yr , which is similar to the timescale required for cooling and dispersal of metals produced by massive stars ( e.g. * ? ? ? * ) as observed in compact hii / bcd galaxies ( e.g. um 408 ; * ? ? ? * ) .
consequently , the predominant mechanism for metal transport may not be resolved in star - forming dwarf galaxies .
however , if we interpreted the observed properties of the ism as due to a late stage - merger and/or inflow of metal - poor gas , the accreted gas must be almost fully diluted as seen in fig .
[ figure_sfr_oh ] .
therefore , the final mixture of gas thus depends on the relative enrichment of the acquired and pre - enriched gas .
@xcite showed that the neutral gas metallicity , in a sample of bcds , is equal or lower than the ones found in the hii region .
this agree with the results obtained by @xcite , using the cosmic origin spectrograph ( cos ) onboard hst , in the sense that the hi gas abundances is factor @xmath52 lower than the warm gas abundance in i zw18 indicating that infall of , non fully pristine , metal - poor gas could be the responsible for its low metallicity .
therefore , we can not exclude that the tail in this xmp galaxy could be made from a gas with much lower metallicity than that in the warm gas phase .
however , we still lack an adequate explanation about the evolutionary status of tol 65 since no spatially resolved hi observations are available , but the infall / accretion of cold gas from the outskirts of the galaxy and/or minor merger / interaction with an small companion recently in the past of the galaxy could explain the main properties of the galaxy . as mentioned above , the tail and the _ cometary morphology _ could be the remnant of these interactions in which an appreciable fraction of gas has been carried away as the result of the tidal torques , then producing the mixing of metals through the ism at large scales . in this sense , the scatter in fig .
[ figure_sfr_oh ] indicates that the metals in the ism were diluted in a relatively small timescale .
if the past infall of metal poor gas do not explain the low metallicity of tol 65 , alternatively , it could be merely a consequence of the low star - formation efficiency due by the cosmic downsizing .
in this work we have presented vimos - ifu spectroscopy of the xmp hii / bcd galaxy tol 65 .
we studied the spatial distribution of properties ( i.e. , emission lines , abundances , kinematics , etc ) through the ism of the galaxy in an extended area of 13@xmath213@xmath38 encompassing the two ghiirs and most of the extended stellar tail .
below , we summarize our results and conclusions : 1 .
this galaxy shows a clear cometary shape with a bright main body and an extended and diffuse stellar tail .
we found that the current star - formation activity in tol 65 started recently about 3 - 5 myr ago .
our observations show the presence of an extended h@xmath0 emission in the tail of the galaxy .
the mass of the ionized gas in the tail corresponds to @xmath524 per cent of the total mass of the ionized gas in the galaxy .
+ 2 . the integrated properties in the main body of the galaxy , it to say ,
the extinction coefficient c(h@xmath9)=0.29 , oxygen abundance 12+log(o / h)[email protected] , velocity dispersion @[email protected] kms@xmath13 obtained in this study agree , within the uncertainties , with the ones determined in the literature ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the velocity field v@xmath44 in tol 65 is highly disturbed and it shows no global rotation pattern .
we found @xmath45 suggesting that the gas still retains the kinematic memory of its parental cloud and a common origin . while extended regions of diffuse gas are highly disturbed , probably due to unresolved expanding shells and the effects of massive star evolution .
4 . we did not observe a spatial variation of metal ( o , n , ne , ar and s ) in the ism of the galaxy .
this agrees with previous observations in other hii galaxies in the literature ( * ? ? ?
* and references therein ) .
although the evidence is far from conclusive , we favour the idea that the most likely mechanisms to produce the flat abundance gradient and the cometary morphology in this xmp hii / bcd is the infall / accretion of cold gas from the outskirts of the galaxy or minor merger / interaction with an small companion recently in the past of the galaxy .
this is in agreement with the scatter found between the gas metallicity ( 12+log(o / h ) ) and star - formation rate at spaxel scales , in fig .
[ figure_sfr_oh ] , which clearly indicates that the metals in the ism are almost fully diluted .
+ our vimos - ifu observation of tol 65 provides a picture in which a late - stage minor merger and/or infall of gas explain most of the observed properties over the ism .
therefore , we suggest that these global effects might be attributed as the main mechanism diminishing the preexisting metal of the galaxy ( e.g. * ? ? ?
* ) , then keeping the oxygen abundance ( and other @xmath0 elements ) constant through the ism at large scales .
clearly , hi observations of tol 65 with high spatial resolution and hst cos observations are needed in order to study the cool gas component and check whether the proposed mechanism are capable to produce the homogeneity of the ism , the difference between the neutral and warm gas metallicity and the morphology of this xmp galaxy .
we would like to thank the referee , dr .
daniel kunth , for his very valuable comments and suggestions .
this work was supported by fundao para a cincia e a tecnologia ( fct ) through the research grant uid / fis/04434/2013 .
is supported by a post - doctoral grant sfrh / bpd/72308/2010 , funded by fct ( portugal ) .
r.d . gratefully acknowledges the support provided by the basal center for astrophysics and associated technologies ( cata ) , and by fondecyt grant n. 1130528 .
, n.r . and j.m.g .
acknowledge support by the fct under project fcomp-01 - 0124-feder-029170 ( reference fct ptdc / fis - ast/3214/2012 ) , funded by the feder program .
is supported by fct through the investigador fct contract no .
if/01220/2013 and poph / fse ( ec ) by feder funding through the program compete .
is supported by a postdoctoral grant sfrh / bpd/66958/2009 , funded by fct ( portugal ) .
we acknowledge support by the exchange programme ` study of emission - line galaxies with integral- field spectroscopy ' ( selgifs , fp7-people-2013-irses-612701 ) , funded by the eu through the irses scheme . 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 .
filho , m. e. , winkel , b. , snchez almeida , j. , aguerri , j. a. , amorn , r. , ascasibar , y. , elmegreen , b. g. , elmegreen , d. m. , gomes , j. m. , humphrey , a. , lagos , p. , morales - luis , a. b. , muoz - tun , c. , papaderos , p. , vlchez , j. m. 2013 , a&a , 558 , 18 | in this study we present high - resolution visible multi - object spectrograph integral field unit spectroscopy ( vimos - ifu ) of the extremely metal - poor hii / blue compact dwarf ( bcd ) galaxy tol 65 .
the optical appearance of this galaxy shows clearly a cometary morphology with a bright main body and an extended and diffuse stellar tail .
we focus on the detection of metallicity gradients or inhomogeneities as expected if the ongoing star - formation activity is sustained by the infall / accretion of metal - poor gas .
no evidences of significant spatial variations of abundances were found within our uncertainties . however , our findings show a slight anticorrelation between gas metallicity and star - formation rate at spaxel scales , in the sense that high star - formation is found in regions of low - metallicity , but the scatter in this relation indicates that the metals are almost fully diluted .
our observations show the presence of extended h@xmath0 emission in the stellar tail of the galaxy .
we estimated that the mass of the ionized gas in the tail m(hii)@xmath11.7@xmath210@xmath3 m@xmath4 corresponds with @xmath524 per cent of the total mass of the ionized gas in the galaxy .
we found that the h@xmath0 velocity dispersion of the main body and the tail of the galaxy are comparable with the one found in the neutral gas by previous studies .
this suggests that the ionized gas still retains the kinematic memory of its parental cloud and likely a common origin . finally , we suggest that the infall / accretion of cold gas from the outskirts of the galaxy and/or minor merger / interaction may have produced the almost flat abundance gradient and the cometary morphology in tol 65 .
[ firstpage ] galaxies : dwarf galaxies : individual : tol 65 ( tol 1223 - 359 , eso380-g027 ) galaxies : ism galaxies : abundances . |
type ia supernovae ( sne - ia ) are thought to be standardizable " candles capable of measuring cosmological distances .
their utility in this sense has been used over the past two decades to first discover ( riess et al .
1998 ; perlmutter et al . 1999 ) and then study the accelerated expansion of the universe ; the results from which have deeply shaken our perception of nature . despite their overwhelming acceptance as standardizable " candles , the mechanism and nature of the progenitors behind the sn - ia explosion are poorly understood .
we understand from their spectra that sne - ia are explosions of carbon / oxygen white dwarfs ( cowds ) ; how they actually explode is still an area of intense research .
two theories lead the fray , both requiring the accumulation of mass onto the white dwarf ( wd ) leading to a thermonuclear explosion : the single - degenerate ( sd ) channel is the accretion onto the wd from a non - degenerate companion ( whelan & iben 1973 ) and the double - degenerate ( dd ) channel provides the mass from the merger of two wds ( iben & tutukov 1994 ; webbink 1984 ) .
in the past year a new channel for the explosion of a wd has been presented ( ouyed & staff 2013 ) , involving the explosion of a companion neutron star ( ns ) as a quark - nova ia ( qn - ia ) .
interestingly sne - ia formed through this channel are not standardizable using usual methods ( ouyed et al .
2013a ) leading to profound implications for cosmology if the majority are formed in this way . resolving the progenitor /
mechanism debate is therefore of the utmost importance for sne - ia to become more accurate tools for measuring cosmological distances , or even for determining if they can be used for such purposes at all .
each of the aforementioned theories has unique signatures , that if observed would provide strong evidence for their cause .
however , sne - ia are difficult to study because they are rare and hence usually discovered at large distances .
the detection of nearby sne - ia therefore offers a unique opportunity to study these explosions in exquisite detail and perhaps lead to a resolution of the progenitor / mechanism debate .
the closest modern - era sne - ia observed prior to 2014 were sn 1972e in ngc 5253 at @xmath0 2.5 - 8 mpc ( e.g. , phillips et al .
1992 ; sandage & tammann 1975 ; della valle & melnick 1992 ; branch et al . 1994 ; sandage et al . 1994 ) and sn 1986 g in ngc 5128 at @xmath0 3 - 5 mpc .
unfortunately at the time of these sne , we did not have the space telescope capabilities we have today .
sn 2014j ( goobar et al .
2014 ) was discovered on jan 21 2014 by astronomer stephen j. fossey in m82 at @xmath0 3.5 mpc , making it one of the closest sne - ia observed in the modern era . with many of the most sophisticated telescopes trained on this object ,
sn 2014j provides an unprecedented opportunity to shed light on the mechanism and nature of the progenitors behind sn - ia explosions . in the coming months and years we expect a deluge of papers on sn 2014j spanning the entire electromagnetic spectrum .
in this paper we present observable predictions for sn 2014j in the context of the qn - ia model in an attempt to make sense of what may or may not be seen and to provide motivation for upcoming observing proposals .
ouyed & staff ( 2013 ) considered the scenario in which a ns and cowd form a tight binary system , where mass transfer from the wd to the ns would occur .
the accreted mass would drive the ns above the critical mass sustainable by neutron matter and it would undergo a quark nova ( qn ) explosion to a quark star ( qs ) ( ouyed et al . 2002 ; vogt et al .
2004 ; niebergal et al . 2010 ; ouyed et al .
the qn ejects the outermost layers of the ns at relativistic speeds with a lorentz factor @xmath1 . on average @xmath2 of iron - rich and neutron - rich material
is ejected during a qn ( kernen et al . 2005 ) equalling about @xmath3 erg in kinetic energy .
this ejecta hits the wd fractions of a second after the qn explosion , leading to the thermonuclear explosion of the wd ; the qn - ia .
the properties of the qn ejecta as it hits the wd have been presented in 2.3 in ouyed & staff ( 2013 ) .
this external triggering mechanism and the induced shock compression implies that even low - mass wds ( i.e. @xmath4 ) will explode in the qn - ia model .
a qn - ia , in addition to the energy from the @xmath5ni decay , is also powered by spin - down energy of the newly born qs .
this results in the qn - ia obeying a phillips - like ( calibration ) relation where the variation in luminosity is due to spin - down power ( see 4 in ouyed et al
. 2013a ) .
we also find the calibration relation to be redshift - dependent which means that sne - ia are not standard candles ( see ouyed et al .
2013a ) making their utility as distance indicators unreliable .
if sn 2014j is a qn - ia explosion , several unique signatures may have already been observed prior to and in the few weeks following its discovery . 1
. the hyper - accretion rate onto the ns just prior to the qn explosion should generate temperatures high enough for strong neutrino emission .
a luminosity on the order @xmath6-@xmath7 erg s@xmath8 in tens of mev neutrinos would be expected . for an @xmath9 erg s@xmath8 and @xmath10 mev neutrinos
this would correspond to a flux of @xmath11s @xmath12 s@xmath8 .
this is clearly much below icecube sensitivity ( abassi et al .
2011 ) but worth mentioning here .
just prior to the neutrino dominated hyper - accretion phase , we expect a brief accretion phase ( @xmath13 day ) set by the photon eddington limit ( @xmath14 erg s@xmath8 ) . in the case of sn 2014j
this would correspond to a flux of @xmath15 erg @xmath12 s@xmath8 which is detectable by chandra ( http://cxc.harvard.edu/ ) .
3 . the ns will spin - up to millisecond periods due to accretion from the wd .
if the viewing angle is fortuitous , and the surrounding electron density low enough , this will have been observed as a radio pulsar in the days prior to the qn - ia explosion .
4 . _ gravitational wave _ ( gw ) detectors should see signatures of two explosions , the qn explosion and the wd detonation ( a fraction of a second apart ) .
the qn gw signatures have been investigated in staff et al .
many signatures of the qn - ia are not evident until the explosion becomes transparent to radiation .
the following are a list of unique signatures that might be observed in sn 2014j in the months and years to come . 1 .
the light curve of sn 2014j is expected , in the qn - ia model , to undergo three distinct phases ( see figure 2 in ouyed et al .
2013a ) : ( i ) spin - down dominated which lasts tens of days ; ( ii ) ( @xmath5ni + @xmath5co ) decay dominated ; ( iii ) a return to spin - down dominated emission starting a few hundred days after the explosion .
we therefore expect to see a glitch " in the light curve of sn 2014j a few hundred days after the explosion , assuming a fiducial spin period of @xmath0 20 ms , as the main energy source changes from ( @xmath5ni + @xmath5co ) decay to spin - down .
if the qs later collapses into a black hole ( bh ) , a second glitch " will be observed as the spin - down energy will suddenly be extinguished .
2 . at @xmath16 year after the explosion we estimate the spin - down luminosity to be @xmath17 erg s@xmath8 for a qs born with an initial period of @xmath18 ms and an initial magnetic field of @xmath19 g. assuming an x - ray efficiency of @xmath16% this would correspond to a flux of @xmath20 erg @xmath12 s@xmath8 .
thus the compact remnant in sn 2014j should appear as a bright x - ray source @xmath16 year after the explosion .
the qs could also be a marginally detectable fermi source ( in the 10 - 100 gev band ; http://fermi.gsfc.nasa.gov ) if a @xmath21-ray efficiency of @xmath16% is assumed . 3 .
the qn explosion proper ejects a very dense , ultra - relativistic ejecta with mass @xmath22 and lorentz factor @xmath23 .
the portion of the qn ejecta that will impact the wd is @xmath24 where @xmath25 is the wd radius and @xmath26 the binary separation when the wd starts to fill its roche - lobe .
this means that at most @xmath27 will collide with the wd while the rest of the qn ejecta expands freely .
thus most of the qn ejecta with its @xmath3 erg of kinetic energy will expand freely outwards with unique implications if it couples to the surrounding environment .
for example , the qn ejecta may carve out a superbubble that could reach out to a parsec in a few years assuming a typical number density of @xmath16 particle per cc in the surrounding environment prior to the explosion .
hst should be able to resolve such a superbubble .
the neutron- and iron - rich qn ejecta was shown to be an ideal site for the nucleo - synthesis of heavy elements , in particular nuclei with atomic weight @xmath28 ( jaikumar et al .
compared to the burnt wd material , these nuclear proxies should be distinguishable in the late spectrum of sn 2014j .
however we predict at most @xmath29 of @xmath28 radioactive material to be mixed with the burnt co ejecta .
this is far too small to be detectable by fermi and nustar at the distance of sn 2014j .
the qs is likely to be born as an aligned rotator ( ouyed et al .
2004 ; ouyed et al .
2006 ) and as such no radio pulsar should be seen in future observations of sn 2014j . 6 .
we have argued in previous work that the qn compact remnant shows properties reminiscent of soft gamma - ray repeaters ( sgrs ) and anomalous x - ray pulsars ( axps ) ( see ouyed et al .
2010 and references therein ) .
we therefore expect magnetar - like behaviour from the location of sn 2014j in the future .
the proximity of sn 2014j offers us an unparalleled opportunity to study a sn - ia which may reveal clues as to the nature of the progenitors and the explosion mechanism .
the two leading explosion scenarios ( sd and dd channels ) have recently been joined by a new intriguing possibility ; the qn - ia .
the relative unfamiliarity of the qn - ia model makes it easy to dismiss .
however , the qn has been successfully applied to a plethora of other astronomical phenomena including sgrs and axps ( e.g. ouyed et al 2010 ) , gamma - ray bursts ( e.g. ouyed et al .
2011 ) and superluminous - supernovae ( e.g. ouyed et al . 2012 ) .
in fact the double - humped light curve observed in sn 2009ip and sn 2010mc ( modelled and well fit by ouyed et al .
2013c ) was first predicted by the qn model in 2009 ( ouyed et al .
2009 ) , four years prior to its discovery .
successes put aside , the qn - ia model does make bold claims that if true could once again alter our perception of nature .
it is therefore imperative that we either rule out this scenario , or confirm it with observations from near - by sne - ia such as sn 2014j . in this paper
we have provided the observer with a list of qn - ia signatures which , if observed in sn 2014j , would support the existence of qne - ia .
we do note , however , that many of the late - time predictions rely on the existence of a qs .
it is entirely possible that the qs collapses to a bh during the ( @xmath5ni + @xmath5co ) decay phase ( see point 1 in 4 ) thereby providing an ( albeit unintended ) hedge .
if the qs - bh transition occurs at any other time , however , we should see this in the light curve of sn 2014j ( e.g. if it occurs immediately , no spin - down energy will be deposited and the light curve will be purely due to ( @xmath5ni + @xmath5co ) decay ) .
ouyed , r. , leahy , d. , & jaikumar , p. 2009 ,
predictions for signatures of the quark - nova in superluminous supernovae " in proceedings of the compact stars in the qcd phase diagram ii ( csqcd ii ) " , may 20 - 24 , 2009 , kiaa at peking university , beijing- p. r. china , eds .
r. ouyed & r. xu , http://www.slac.stanford.edu/econf/c0905202/ [ arxiv:0911.5424 ] ouyed , r. , niebergal , b. , & jaikumar , p. 2013b , explosive combustion of a neutron star into a quark star : the non - premixed scenario " in proceedings of the compact stars in the qcd phase diagram iii ( csqcd iii ) conference , december 12 - 15 , 2012 , guaruj , brazil . eds .
l. paulucci , j. e. horvath , m. chiapparini , and r. negreiros , http://www.slac.stanford.edu/econf/c121212/ [ arxiv:1304.8048 ] | the exact mechanism behind type ia supernovae ( sne - ia ) and the nature of the progenitors is poorly understood , although several theories vie for supremacy .
their secured importance to the study of cosmology necessitates a resolution to this problem .
observations of nearby sne - ia are therefore critical to determine which theory , if any , is the correct one .
sn 2014j discovered at a relatively close 3.5 mpc in the galaxy m82 provides such an opportunity . in this paper
we give specific predictions for sn 2014j in the context of the quark nova ia ( qn - ia ) model .
predictions include x - ray luminosities just prior and hundreds of days after the explosion , light curve glitches " , neutrino emission , heavy element creation , and gravitational signatures . |
while advances in two - dimensional ( 2d ) percolation have recently allowed to determine the site percolation threshold on the square lattice with an astonishing accuracy of 14 significant digits @xcite and many critical exponents in 2d have been known exactly for decades @xcite , the progress in higher dimensions is far slower .
the main reason for this is that the two theoretical concepts that proved particularly fruitful in percolation theory , conformal field theory and duality , are useful only in 2d systems , and the thresholds in higher dimensions are known only from simulations .
the site and bond percolation thresholds in dimensions @xmath5 are known with accuracy of at least 6 significant digits @xcite , but for more complicated lattices , e.g. fcc , bcc or diamond lattices @xcite , complex neighborhoods @xcite , or continuum percolation models @xcite this accuracy is often far from satisfactory . moreover , even though the upper critical dimension is known to be @xmath6 @xcite , numerical estimates of the critical exponents for @xmath7 are still rather poor .
continuous percolation of aligned objects can be regarded as a limit of a corresponding discrete model .
using this fact , we recently improved the accuracy of numerical estimates of continuous percolation of aligned cubes ( @xmath8 ) @xcite .
we also generalized the excluded volume approximation @xcite to discrete systems and found that the limit of the continuous percolation is controlled by a power - law dependency with an exponent @xmath9 valid for both @xmath10 and @xmath11 .
the main motivation behind the present paper is to verify whether the relation @xmath9 holds also for higher dimensions and if so , whether it can be used to improve the accuracy of continuous percolation thresholds in the model of aligned hypercubes in dimensions @xmath1 . with this selection
, the conjecture will be verified numerically for all dimensions @xmath12 as well as in one case above @xmath6 , which should render its generalization to all @xmath13 plausible .
answering these questions required to generate a lot of data , from which several other physically interesting quantities could also be determined .
in particular , we managed to improve the accuracy of the correlation length critical exponent @xmath2 in dimensions @xmath7 and to determine the values of various universal wrapping probabilities in dimensions @xmath4 .
we consider a hypercubic lattice of the linear size @xmath14 lattice units ( l.u . ) in a space dimension @xmath13 .
this lattice is gradually filled with hypercubic `` obstacles '' of linear size @xmath15 l.u .
( @xmath16 ) until a wrapping percolation has been found ( for the sake of simplicity , henceforth we will assume that @xmath14 , @xmath15 are dimensionless integers ) .
the obstacles , aligned to the underlying lattice and with their edges coinciding with lattice nodes , are deposited at random into the lattice and the periodic boundary conditions in all directions are assumed to reduce finite - size effects . during this process the deposited hypercubes are free to overlap ; however , to enhance the simulation efficiency , no pair of obstacles is allowed to occupy exactly the same position . as illustrated in figure
[ fig : model ] , construction of the model in the space of dimension @xmath10 .
an empty regular lattice of size @xmath17 lattice units ( l.u . ) with periodic boundary conditions ( a ) is filled at random with square obstacles of size @xmath18 l.u .
aligned to the lattice axes ( b ) and the elementary cells occupied by the obstacles are identified ( c ) ; finally , a wrapping path through the occupied elementary cells ( site percolation ) is looked for ( d ) .
the same method was used for larger @xmath13 . ] the volume occupied by the obstacles can be regarded as a simple union of elementary lattice cells and the model is essentially discrete .
two elementary cells are considered to be connected directly if and only if they are occupied by an obstacle and share the same hyperface of an elementary cell .
we define a percolation cluster as a set of the elementary cells wrapping around the system through a sequence of directly connected elementary cells .
thus , the model interpolates between the site percolation on a hypercubic lattice for @xmath19 and the model of continuous percolation of aligned hypercubes @xcite in the limit of @xmath20 .
the percolation threshold is often expressed in terms of the volume fraction @xmath21 defined as the ratio of the number of the elementary cells occupied by the obstacles to the system volume , @xmath22 . what is the expected value of @xmath21 after @xmath23 hypercubes
have been placed at random ( but different ) positions ? to answer this question , notice that while the obstacles can overlap , they can be located at exactly @xmath24 distinct locations and so @xmath25 . moreover , owing to the periodic boundary conditions , any elementary cell can be occupied by exactly @xmath26 different hypercubes , where @xmath27 is the volume of a hypercube .
thus , the probability that an elementary cell is not occupied by an obstacle , @xmath28 , is equal to the product of @xmath26 probabilities that no hypercubes were placed at @xmath26 locations .
this implies that @xmath29 for @xmath19 this formula reduces to @xmath30 irrespective of @xmath13 . in the limit of @xmath20 equation ( [ eq : def - phi ] ) reduces to @xmath31 , where @xmath32 is the reduced number density @xcite .
the number of lattice sites in a cluster of linear size @xmath14 is of order of @xmath24 , a quantity rapidly growing with @xmath14 in high dimensions @xmath13 .
this imposes severe constraints on numerical methods . on the one hand
, one would like to have a large @xmath14 to minimize finite - size effects , which are particularly important near a critical state ; on the other hand , dealing with @xmath24 objects exerts a pressure on the computer storage and computational time . to mitigate this problem ,
special algorithms were developed that focus on the efficient use of the computer memory .
for example , leath s algorithm @xcite , in which a single cluster is grown from a single - site `` seed '' , turned out very successful in high - dimensional simulations of site and bond percolation @xcite .
however , the use of such algorithms in the present model would be impractical , as the obstacle linear size @xmath15 is now allowed to assume values as large as 1000 , which greatly complicates the definition of leath s `` active neighborhood '' of a cluster . therefore we used a different approach , with data structures typical of algorithms designed for the continuous percolation : each hypercube is identified by its coordinates , i.e. , by @xmath13 integers , and the clusters are identified using the union - find algorithm . with this choice , the computer memory storage , as well as the simulation time of each percolation cluster ,
is @xmath33 , which enables one to use large values of @xmath14 and @xmath15 .
we were able to run the simulations for @xmath34 ( @xmath8 ) , @xmath35 ( @xmath36 ) , @xmath37 ( @xmath38 ) , @xmath39 @xmath40 , and @xmath41 @xmath42 , and the maximum values of @xmath43 were limited by the acceptable computation time rather than the storage .
the simulation time in our method is determined by how quickly one can identify all obstacles connected to the next obstacle being added to the system . to speed this step up , we divided the system into @xmath44 bins of linear size @xmath43 .
each obstacle was assigned to exactly one bin and for each bin we stored a list of obstacles already assigned to it . in this way , upon adding a new obstacle , the program had to check @xmath45 neighboring bins to identify all obstacles connected to the just added one .
as @xmath13 increases , this step becomes the most time - consuming part of the algorithm .
fortunately , the negative impact of the @xmath45 factor is to some extent mitigated by the fact that the critical volume fraction , @xmath46 , is much smaller for @xmath47 than for @xmath19 , so that for @xmath47 one needs to generate a relatively small number of obstacles to reach the percolation .
this is related to the fact that for @xmath48 the value of @xmath49 @xcite , whereas @xmath50 @xcite .
the case @xmath19 is special in that one has to check only @xmath51 neighboring sites of a given obstacle , a value much smaller than @xmath45 .
thus , the total simulation time at a high space dimension for a fixed value of @xmath43 can be approximated as @xmath52 for @xmath19 and @xmath53 for @xmath54 . in practice , simulations in the space dimension @xmath55 ( with @xmath43 fixed ) are between 8 to 13 times faster for @xmath19 than for @xmath47 .
this allowed us to run more simulations and obtain more accurate results for @xmath19 than for @xmath54 .
we assumed periodic boundary conditions along all @xmath13 main directions of the lattice .
the number of independent samples varied from @xmath56 for very small systems ( e.g. , @xmath8 , @xmath19 , @xmath57 ) to @xmath58 for larger @xmath14 and @xmath13 ( e.g. , @xmath55 , @xmath19 , @xmath59 ) . starting from an empty system of volume @xmath24 , we added hypercubes of volume @xmath60 at different random locations until we have detected wrapping clusters in all @xmath13 directions .
thus , for each simulation we stored @xmath13 numbers @xmath61 equal to the number of hypercubes for which a wrapping percolation was first detected along cartesian direction @xmath62 .
having determined all @xmath61 in a given simulation , we can use several definitions of the onset of percolation in a finite - size system @xcite .
for example , one can assume that the system percolates when there is a wrapping cluster along some preselected direction @xmath63 , say , @xmath64 . we shall call this definition ` case a ' .
alternatively , a system could be said to be percolating when there is a wrapping cluster along _ any _ of the @xmath13 directions .
we shall call this ` case b '
. another popular definition of a percolation in a finite - size system is the requirement that the wrapping condition must be satisfied in _ all _ @xmath13 directions .
this will be denoted as ` case c ' .
next , for each of the three percolation definitions , a , b and c , we determined the probability @xmath65 that a system of size @xmath14 , obstacle size @xmath15 , and volume fraction @xmath21 contains a percolating ( wrapping ) cluster .
this step is based on a probability distribution function constructed from from all @xmath61 in case a , from all @xmath66 in case b , and from all @xmath67 in case c. notice that in case a we take advantage of the symmetry of the system which ensures that all main directions are equivalent so that after @xmath23 simulations we have @xmath68 pieces of data from which a single probability distribution function can be constructed . in doing this
we implicitly assume that all @xmath61 are independent of each other , which may improve statistics .
fixing @xmath69 and @xmath15 , and using ( [ eq : def - phi ] ) , we can write @xmath70 as a function of @xmath23 , @xmath71 . in accordance with the finite - size scaling theory , @xmath70 is expected to scale with @xmath14 and the deviation @xmath72 from the critical volume fraction @xmath73 as @xcite @xmath74l^{1/\nu } \right ) , \quad l / k \gg 1,\ ] ] where @xmath2 is the correlation length exponent , @xmath75 is a scaling function , and @xmath21 is related to @xmath23 through ( [ eq : def - phi ] ) .
this formula describes the probability that there is a percolation cluster in a system containing exactly @xmath23 obstacles , i.e. , is a quantity computed for a `` microcanonical percolation ensemble '' .
the corresponding value in the canonical ensemble is @xcite @xmath76 where @xmath77 is the probability that there is an obstacle assigned to a given location ; this quantity is related to the mean occupied volume fraction through @xmath78 while @xmath79 is a discrete function , @xmath80 is defined for all @xmath81 and has a reduced statistical noise . by using ( [ eq : phi - p ] ) , @xmath82 can be regarded as a continuous function of @xmath21 for @xmath83 .
an effective , @xmath14-dependent volume fraction @xmath84 was then determined numerically as the solution to @xmath85 where @xmath86 is a fixed parameter , and we chose @xmath87 in all our calculations . the critical volume fraction , @xmath73 , as well as the critical exponent @xmath2
are then determined using the scaling relation @xcite @xmath88 where @xmath89 are some @xmath15- and @xmath13-dependent parameters and @xmath90 is the cutoff parameter .
recently a more general scaling ansatz for the form of the probability @xmath82 in the vicinity of the critical point was proposed @xcite , @xmath91 where @xmath92 is a universal constant , @xmath93 is the leading correction exponent , and @xmath94 are some nonuniversal , model - dependent parameters . at the critical point
this reduces to @xmath95 we used this ansatz to determine the universal , @xmath15-independent constant @xmath92 @xcite representing the probability that a wrapping cluster exists at the critical point .
relation ( [ eq : phi - scaling ] ) contains @xmath96 unknowns : @xmath73 , @xmath2 , and @xmath89 .
the critical exponent @xmath2 can be also estimated from an alternative relation containing @xmath97 unknowns by noticing that ( [ eq : scaling ] ) leads to @xmath98 where @xmath99 . to take into account finite - size corrections , we used a formula @xmath100 where @xmath101 are some parameters and we used @xmath102 . actually , since @xmath103 is a quickly growing function near @xmath73 , we calculated the derivative of its inverse , @xmath104 using a five - point stencil , @xmath105/12h$ ] with @xmath106 .
in @xcite we conjectured that for sufficiently large @xmath15 @xmath107 \propto k^{-\theta},\ ] ] where @xmath108 is the critical volume fraction for the continuous percolation of aligned hypercubes and @xmath0 .
the left - hand side of ( [ eq : approx - k32 ] ) was derived using the excluded volume approximation applied to discrete systems , whereas its right - hand - side was obtained numerically and verified for @xmath109 .
this formula enables one to estimate @xmath110 from the critical volumes @xmath46 obtained for discrete ( lattice ) models with finite @xmath15 by investigating the rate of their convergence as @xmath20 .
its characteristic feature is the conjectured independence of @xmath111 on @xmath13 that we verify in this report .
the uncertainties of the results were determined as follows .
first , the percolation data were divided into 10 disjoint groups .
then for each group the value of @xmath84 was calculated in the way described above .
the value of @xmath84 was then assumed to be equal to their average value , and its uncertainty to the standard error of the mean .
next , the value of @xmath73 was obtained from a non - linear fitting to eq .
( [ eq : phi - scaling ] ) using the levenberg - marquardt algorithm , with the errors on the parameters estimated from the square roots of the diagonal elements of the covariance matrix , multiplied by @xmath112 , where @xmath113 is the reduced chi - square statistic .
the same method was used to estimate the value and uncertainty of @xmath110 from eq .
( [ eq : approx - k32 ] ) . finally , making up the sum in eq .
( [ eq : q ] ) is potentially even more tricky than in site percolation , as in our model @xmath114 can be as large as @xmath115 . in solving this technical problem
we followed the method reported in @xcite if @xmath114 could be stored in a 64-bit integer , otherwise we approximated the binomial distribution with the normal distribution .
another point worth noticing is that @xmath77 in eq .
( [ eq : q ] ) can be as small as @xmath116 .
in such a case expressions like @xmath117 should be computed using appropriate numerical functions , e.g. , log1p from the c++ standard library , which is designed to produce values of @xmath118 with @xmath119 without a potential loss of significance in the sum @xmath120 .
we start our analysis from the particular case @xmath19 , in which the model reduces to the standard site percolation .
as site percolation has been analyzed extensively with many dedicated methods , we were going to use the case @xmath19 only to test the correctness of our computer code , but as it turned out , we have managed to obtain some new results , too .
the main results are summarized in table [ tab : site - percolation ] .
@l*15@l @xmath13 & case & & + & & & + & & best known & present & best known & present ( i ) & present ( ii ) + 3 & a & 0.311 607 68(15)@xmath121 & 0.311 608 8(57 ) & 0.876 19(12)@xmath121 & 0.873 6(35 ) & 0.877 3(12 ) + & b & & 0.311 608 0(42 ) & & 0.855(18 ) & 0.874 3(15 ) + & c & & 0.311 601 7(47 ) & & 0.878 7(16 ) & 0.878 31(80 ) + & & & & + & & & 0.311 606 0(48 ) & & + 4 & a & 0.196 886 1(14)@xmath122 & 0.196 890 8(60 ) & 0.689(10)@xmath123 & 0.683 0(59 ) & 0.682 2(41 ) + & b & & 0.196 891 9(55 ) & & 0.674(18 ) & 0.682 7(34 ) + & c & & 0.196 885(10 ) & & 0.687 9(61 ) & 0.687 5(23 ) + & & & & + & & & 0.196 890 4(65 ) & & + 5 & a & 0.140 796 6(15)@xmath122 & 0.140 796 7(22 ) & 0.569(5)@xmath124 & 0.574 3(50 ) & 0.572 5(38 ) + & b & & 0.140 795(10 ) & & 0.59(11 ) & 0.571 6(26 ) + & c & & 0.140 796 5(21 ) & & 0.573 3(29 ) & 0.572 0(14 ) + & & & & + & & & 0.140 796 6(26 ) & & + 6 & a & 0.109 017(2)@xmath122 & 0.109 011 3(14 ) & 1/2 & 0.58(39 ) & 0.495(19 ) + & b & & 0.109 017 5(26 ) & & & 0.497(33 ) + & c & & 0.109 009 9(16 ) & & 0.513(58 ) & 0.495(40 ) + & & & & + & & & 0.109 011 7(30 ) & & + 7 & a & 0.088 951 1(9)@xmath122 & 0.088 951 4(56 ) & 1/2 & 0.36(12 ) & 0.44(11 ) + & b & & 0.088 950(15 ) & & & + & c & & 0.088 945 7(35 ) & & 0.40(11 ) & 0.41(8 ) + & & & & + & & & 0.088 951 1(90 ) & & + @xmath121@xcite , @xmath122@xcite , @xmath123@xcite , @xmath124@xcite .
the values of the critical volume fraction , @xmath125 , were determined using ( [ eq : phi - scaling ] ) with @xmath102 for all three definitions ( a , b , and c ) of the onset of percolation in finite - size systems , as defined in section [ sec : numerical - details ] .
for @xmath126 we assumed that @xmath127 , whereas for @xmath128 we treated @xmath2 as an unknown , fitting parameter .
percolation thresholds obtained in cases a , b , and c are consistent with each other and with those reported in other studies @xcite .
we combined them into a single value using the inverse - variance weighting .
these combined values are listed in table [ tab : site - percolation ] as `` final '' values .
their uncertainty was determined as the square root of the variance of the weighted mean multiplied by a correction term @xmath129 , where @xmath130 is an additional , conservative correction term introduced to compensate for the possibility that the measurements carried out in cases a , b , and c are not statistically independent , as they are carried out using the same datasets .
the uncertainties obtained for @xmath131 are similar in magnitude to those obtained with leath s algorithm @xcite , even though our algorithm was not tuned to the numerical features of the site percolation problem .
it is also worth noticing that the uncertainties of @xmath125 for cases a , b and c are similar to each other even though case a utilizes a larger number of data .
this suggests that the numbers @xmath61 obtained in individual simulations are correlated .
the values of the critical exponent @xmath2 were obtained independently using either ( [ eq : phi - scaling ] ) , which we call `` method i '' , or ( [ eq : derivative ] ) ( `` method ii '' ) . obviously , in contrast to the use of ( [ eq : phi - scaling ] ) to estimate @xmath125 , in method i we treated @xmath2 as a fitting parameter for all @xmath13 .
just as for @xmath125 , we publish the values of @xmath2 for individual cases a , b , and c as well as their combined values obtained with the inverse - variance weighting .
we also present the results for @xmath132 , where the exact value of @xmath2 is known , as it helps to verify accuracy of the applied methods .
again , the results are consistent with the values of @xmath2 reported in previous studies @xcite .
method ii turned out to be generally more accurate than method i and the accuracy of both methods decreases with @xmath13 .
we attribute the latter phenomenon to a rapid decrease of the maximum system size that can be reached in simulations , @xmath133 , with @xmath13 .
for example , while for @xmath8 we used @xmath134 , for @xmath55 we had to do with @xmath135 , which certainly has a negative impact on power - law fitting accuracy . for @xmath126 equation ( [ eq : phi - scaling ] ) with @xmath2 treated as a fitting parameter leads to rather poor fits in which the uncertainty of some fitting parameters may exceed 100%
however , the same equation still gives good quality fits after fixing @xmath2 at its theoretical value @xmath136 , which justifies its use for @xmath137 in table [ tab : site - percolation ] .
combining the results from cases a , b , and c and methods i and ii , we found @xmath138 for @xmath36 and @xmath139 for @xmath38 , which are more accurate than those reported previously , @xmath140 for @xmath36 @xcite and @xmath141 for @xmath38 @xcite .
these improved values will be used in data analysis for the case @xmath47 .
next we verified eq .
( [ eq : approx - k32 ] ) for cases a , b , and c and @xmath142 . to this end
the values of @xmath73 were determined using ( [ eq : phi - scaling ] ) and @xmath2 fixed at the best value available , i.e. , @xmath143 for @xmath8 @xcite , our values reported in table [ tab : site - percolation ] for @xmath7 , and @xmath144 for @xmath137 @xcite .
the uncertainty of @xmath2 was included into the final uncertainties of the fitting parameters .
our results , depicted in figure [ fig : exponents-4 - 7 ] , the left hand side of ( [ eq : approx - k32 ] ) as a function of the obstacle size , @xmath15 .
symbols show numerical results for the space dimension @xmath8 ( circles ) , 4 ( pluses ) , 5 ( crosses ) , 6 ( stars ) , and 7 ( squares ) , whereas the lines are the best fits to ( [ eq : approx - k32 ] ) with @xmath145 . ]
confirm our hypothesis that @xmath0 irrespective of the space dimension ( similar scaling for @xmath109 but percolation defined through spanning clusters was reported in @xcite ) .
this opens the way to use eq .
( [ eq : approx - k32 ] ) , with @xmath0 , as a means of estimating the continuous percolation threshold of aligned hypercubes , @xmath146 . the results , presented in table [ tab : continuous - percolation ] , llllll @xmath13 & case & & & + & & & & + & & & best known & present & + 3 & a & 0.226 38 & 0.277 27(2)@xmath121 & 0.277 302 0(10 ) & 0.293 + & b & & & 0.277 300 9(10 ) & + & c & & & 0.277 302 61(79 ) & + & & & + & & & & 0.277 301 97(91 ) + 4 & a & 0.098 13 & 0.113 2(5)@xmath122 & 0.113 234 40(73 ) & 0.146 + & b & & & 0.113 233 90(91 ) & + & c & & & 0.113 237 9(13 ) & + & & & + & & & & 0.113 234 8(17 ) + 5 & a & 0.043 73 & 0.049 00(7)@xmath122 & 0.048 163 5(15 ) & 0.071 + & b & & & 0.048 165 8(14 ) & + & c & & & 0.048 162 1(13 ) & + & & & + & & & & 0.048 163 7(19 ) + 6 & a & 0.020 03 & 0.020 82(8)@xmath122 & 0.021 347 4(10 ) & 0.034 + & b & & & 0.021 344 6(27 ) & + & c & & & 0.021 347 9(10 ) & + & & & + & & & & 0.021 347 4(12 ) + 7 & a & 0.009 38 & 0.009 99(5)@xmath122 & 0.009 776 9(10 ) & 0.017 + & b & & & 0.009 782 0(27 ) & + & c & & & 0.009 773 1(10 ) & + & & & + & & & & 0.009 775 4(31 ) + @xmath121@xcite.@xmath122@xcite .
turn out far more accurate than those obtained with other methods .
however , they agree with the data reported in @xcite only for @xmath147 . in particular , for @xmath148 our value of the percolation threshold of aligned hypercubes , @xmath146 , is away from the value predicted in @xcite by @xmath149 , where @xmath150 is the sum of the uncertainities of @xmath146 found in our simulations and that reported in @xcite .
this indicates that for @xmath151 either our uncertainty estimates or those reported in @xcite are too small .
this discrepancy is very peculiar , because it exists only for @xmath151 , whereas the results for lower @xmath13 are in perfect accord .
this finding made us recheck our computations .
the main part of our code is written using c++ templates with the space dimension @xmath13 treated as a template parameter .
the raw percolation data is then analyzed using a single toolchain for which @xmath13 is just a parameter .
this implies that exactly the same software is used for any @xmath13 .
next , our results for different percolation definitions ( a , b , and c ) agree with each other well . moreover , our results for @xmath19 are in good agreement with all the results available for the site percolation , and for @xmath47 they satisfy the asymptotic scaling expressed in eq .
( [ eq : approx - k32 ] ) . also , as shown in table [ tab : continuous - percolation ] , all our results for @xmath146 lie between the lower and upper bounds , @xmath152 and @xmath153 , reported in @xcite .
we verified that the reduced chi - square statistic in practically all fits satisfies @xmath154 , which indicates that the data and their uncertainties fit well to the assumed models .
we also implemented the code responsible for the transition from the microcanonical to canonical ensemble , equation ( [ eq : q ] ) , in such a way that all floating - point operations could be performed either in the ieee 754 double ( 64-bit ) or extended precision ( 80-bit ) mode .
the results turned out to be practically indistinguishable , indicating that the code is robust to numerical errors related to the loss of significance .
an alternative verification of the results is presented in figure [ fig : verification ] .
percolation threshold @xmath46 in dimension @xmath38 for several values of the obstacle linear size @xmath15 .
pluses represent our numerical results , the dashed line was calculated from a fit to ( [ eq : approx - k32 ] ) for @xmath155 , the circle depicts the value extrapolated for @xmath20 from ( [ eq : approx - k32 ] ) , and the square reproduces the value of this limit as reported in @xcite . ]
it shows that our simulation data for @xmath46 and @xmath38 are in a very good agreement with ( [ eq : approx - k32 ] ) .
the reduced chi - square statistic , @xmath156 , indicates a good fit , even though the uncertainties of individual data points are very small , from @xmath157 ( @xmath158 ) to @xmath159 ( @xmath160 ) the value reported in @xcite for @xmath46 is clearly inconsistent with our data .
the situation for @xmath132 is similar ( data not shown ) .
it is also worth noticing that our results for @xmath8 are an order of magnitude more accurate than those obtained in @xcite using exactly the same method , but with the percolation defined through spanning rather than wrapping clusters .
this confirms a known fact that the estimates of the percolation threshold obtained using a cluster wrapping condition in a periodic system exhibit significantly smaller finite - size errors than the estimates made using cluster spanning in open systems @xcite .
one possible cause of the discrepancy between our results for continuous percolation of aligned hypercubes and those obtained in @xcite are the corrections to scaling due to the finite size of the investigated systems .
to get some insight into their role , we used ( [ eq : u0 ] ) to obtain the values of the universal constant @xmath92 for @xmath161 together with @xmath162 , @xmath163 , @xmath164 , which control the magnitude of the corrections to scaling .
first we focused on @xmath164 and found that it is impossible to find reliable values of this exponent from our data .
wang et al .
@xcite also reported difficulties in determining @xmath164 from simulations , but eventually found @xmath165 for @xmath8 .
as the values of this exponent for @xmath166 are unknown , and we checked that the value of @xmath92 obtained from ( [ eq : u0 ] ) is practically insensitive to whether one assumes that @xmath167 or @xmath168 , we chose the simplest option : @xmath167 for all @xmath13 , which turns ( [ eq : u0 ] ) into the usual taylor expansion in @xmath43 , @xmath169 where @xmath170 and @xmath171 .
figure [ fig : ypc - individual ] the probability of a wrapping cluster along a given axis ( case a ) at criticality , @xmath172 , as a function of the system size relative to the obstacle size , @xmath43 for selected values of the obstacle size , @xmath19 ( red crosses ) , 10 ( blue squares ) , and @xmath173 ( green circles ) in dimensions @xmath161 ( panels a, e , respectively ) .
the lines show the fits to ( [ eq : u01 ] ) for @xmath174 ( with @xmath175 ) .
the lines for @xmath176 would lie very close to those for @xmath177 and are hidden for clarity .
the error bars do not include the uncertainty of @xmath178 .
the regions filled with a pattern in panel ( e ) show how @xmath172 would change if the value of @xmath179 was allowed to vary by up to its three standard deviations , for @xmath19 ( red ) and @xmath176 ( green ) .
, title="fig : " ] the probability of a wrapping cluster along a given axis ( case a ) at criticality , @xmath172 , as a function of the system size relative to the obstacle size , @xmath43 for selected values of the obstacle size , @xmath19 ( red crosses ) , 10 ( blue squares ) , and @xmath173 ( green circles ) in dimensions @xmath161 ( panels a, e , respectively ) .
the lines show the fits to ( [ eq : u01 ] ) for @xmath174 ( with @xmath175 ) .
the lines for @xmath176 would lie very close to those for @xmath177 and are hidden for clarity .
the error bars do not include the uncertainty of @xmath178 .
the regions filled with a pattern in panel ( e ) show how @xmath172 would change if the value of @xmath179 was allowed to vary by up to its three standard deviations , for @xmath19 ( red ) and @xmath176 ( green ) .
, title="fig : " ] + the probability of a wrapping cluster along a given axis ( case a ) at criticality , @xmath172 , as a function of the system size relative to the obstacle size , @xmath43 for selected values of the obstacle size , @xmath19 ( red crosses ) , 10 ( blue squares ) , and @xmath173 ( green circles ) in dimensions @xmath161 ( panels a, e , respectively ) .
the lines show the fits to ( [ eq : u01 ] ) for @xmath174 ( with @xmath175 ) .
the lines for @xmath176 would lie very close to those for @xmath177 and are hidden for clarity .
the error bars do not include the uncertainty of @xmath178 .
the regions filled with a pattern in panel ( e ) show how @xmath172 would change if the value of @xmath179 was allowed to vary by up to its three standard deviations , for @xmath19 ( red ) and @xmath176 ( green ) .
, title="fig : " ] the probability of a wrapping cluster along a given axis ( case a ) at criticality , @xmath172 , as a function of the system size relative to the obstacle size , @xmath43 for selected values of the obstacle size , @xmath19 ( red crosses ) , 10 ( blue squares ) , and @xmath173 ( green circles ) in dimensions @xmath161 ( panels a, e , respectively ) .
the lines show the fits to ( [ eq : u01 ] ) for @xmath174 ( with @xmath175 ) .
the lines for @xmath176 would lie very close to those for @xmath177 and are hidden for clarity .
the error bars do not include the uncertainty of @xmath178 .
the regions filled with a pattern in panel ( e ) show how @xmath172 would change if the value of @xmath179 was allowed to vary by up to its three standard deviations , for @xmath19 ( red ) and @xmath176 ( green ) .
, title="fig : " ] + the probability of a wrapping cluster along a given axis ( case a ) at criticality , @xmath172 , as a function of the system size relative to the obstacle size , @xmath43 for selected values of the obstacle size , @xmath19 ( red crosses ) , 10 ( blue squares ) , and @xmath173 ( green circles ) in dimensions @xmath161 ( panels a, e , respectively ) .
the lines show the fits to ( [ eq : u01 ] ) for @xmath174 ( with @xmath175 ) .
the lines for @xmath176 would lie very close to those for @xmath177 and are hidden for clarity .
the error bars do not include the uncertainty of @xmath178 .
the regions filled with a pattern in panel ( e ) show how @xmath172 would change if the value of @xmath179 was allowed to vary by up to its three standard deviations , for @xmath19 ( red ) and @xmath176 ( green ) .
, title="fig : " ] shows @xmath172 as a function of @xmath43 for @xmath161 and selected values of @xmath15 ( case a ) .
as @xmath172 is expected to converge to a @xmath13-dependent limit @xmath92 as @xmath180 , inspection of its convergence rate can serve as an indicator of the magnitude of the corrections to scaling for the range of the @xmath14 values used in the simulations .
the plots for @xmath177 are very similar to those obtained for @xmath176 , which suggests that the behavior of @xmath172 for @xmath177 can be used as a good approximation of @xmath172 in the limit of the continuous system , @xmath181 . rather surprisingly ,
for @xmath8 this behavior is also similar to that observed in the site percolation ( @xmath19 ) . in higher dimensions the convergence patterns are different : the site percolation is characterized by a nonmonotonic dependence of @xmath172 on @xmath43 , whereas in continuous percolation this dependency is monotonic .
notice also the different scales used in the plots : the variability of @xmath172 increases with @xmath13 and at the same time the maximum value of @xmath43 attainable in simulations quickly decreases .
these two factors amplify each other s negative influence on the simulations , which hinders the usability of the method in higher dimensions .
looking at figures [ fig : ypc - individual ] ( c)-(e ) , one might doubt if they represent quantities converging to the same value irrespective of @xmath15 .
however , these curves turn out to be very sensitive to even small changes in @xmath179 , which are known with a limited accuracy , a factor not included into the error bars . to illustrate the magnitude of this effect , we show in figure [ fig : ypc - individual ] ( e ) how @xmath172 would change as a function of @xmath43 if @xmath179 was allowed to vary by up to three times its numerical uncertainty for @xmath55 and @xmath182 ( case a ) .
for @xmath176 the impact of the uncertainty of @xmath179 on @xmath172 turns out larger than the statistical errors , and if we take it into account , the hypothesis that the curves converge to the same value can no longer be ruled out . actually , the requirement that this limit is @xmath15-independent can be used to argue that our estimation of @xmath179 for @xmath55 , @xmath176 is larger than the value obtained from this condition by about twice its numerical uncertainty , which is an acceptable agreement .
while this idea could be used to improve the uncertainty estimates of @xmath179 ( see @xcite ) , we did not use it systematically in the present study .
the reason of high sensitivity of @xmath183 to changes in @xmath179 is related to the fact that the slope of @xmath65 at @xmath179 for the largest system sizes attainable in simulations quickly grows with @xmath13 and , to a lesser extent , with @xmath15 the probability that the system is at percolation , @xmath70 , as a function of the distance to the critical point , @xmath184 , for @xmath185 , @xmath182 , and the largest values of @xmath43 used in our simulations ( case a ) . ]
( figure [ fig : pdfs ] ) . for @xmath55 and @xmath176
this slope is as large as @xmath186 , so that in this case the uncertainty of @xmath179 of the order of @xmath187 translates into the uncertainty of @xmath172 of the order of @xmath188 .
the data in figure [ fig : pdfs ] allows one to make also another observation . using ( [ eq : tau ] ) with @xmath87 and the raw data for @xmath8 , one can estimate the percolation threshold with the accuracy of @xmath189 . using extrapolation ,
this can be improved by a factor of @xmath190 to reach the accuracy reported in table [ tab : continuous - percolation ] . for @xmath55 and @xmath191 the error from the raw data is already very small , @xmath192 .
extrapolation can be still used to reduce it further , but since now the data come from systems of smaller linear size ( @xmath193 rather than @xmath34 for @xmath8 ) , the reduction factor is also smaller , of the order of @xmath194 .
thus , the problems with convergence , which can be seen in panels ( c)-(e ) of figure [ fig : ypc - individual ] , are related to the difficulty in the determination of the universal constant @xmath92 , not @xmath179
. an independent method of evaluating @xmath92 for @xmath195 , even with a moderate precision , would give a powerful method of obtaining the percolation threshold in high dimensional spaces .
the values of @xmath92 , @xmath162 , and @xmath163 obtained from the fits of the data shown in figure [ fig : ypc - individual ] are presented in table [ tab : rx ] .
cr|lll @xmath13 & & & & + 3 & 1 & 0.2580(2 ) & @xmath196 & @xmath197 + 3 & 10 & 0.2581(6 ) & @xmath198 & @xmath199 + 3 & 100 & 0.2583(6 ) & @xmath200 & @xmath201 + 4 & 1 & 0.1786(7 ) & @xmath202 & @xmath203 + 4 & 10 & 0.1796(19 ) & @xmath204 & @xmath205 + 4 & 100 & 0.1796(16 ) & @xmath206 & @xmath207 + 5 & 1 & 0.167(1 ) & @xmath208 & @xmath209 + 5 & 10 & 0.166(8 ) & @xmath210 & @xmath211 + 5 & 100 & 0.165(9 ) & @xmath212 & @xmath213 + 6 & 1 & 0.206(3 ) & @xmath214 & @xmath215 + 6 & 10 & 0.199(19 ) & @xmath216 & @xmath217 + 6 & 100 & 0.202(18 ) & @xmath218 & @xmath219 + 7 & 1 & 0.313(8 ) & @xmath220 & @xmath221 + 7 & 10 & 0.229(50 ) & @xmath222 & @xmath223 + 7 & 100 & 0.253(32 ) & @xmath224 & @xmath225 + their inspection leads to several conclusions .
first , they agree with the hypothesis that @xmath92 is universal for a given space dimension @xmath13
. in particular , our value of the universal constant for @xmath8 , @xmath226 , agrees with @xmath227 reported in @xcite .
second , even though the uncertainties of @xmath162 and @xmath163 are typically high , often exceeding 100% @xcite , one can notice that their magnitude grows with @xmath13 , which means that the magnitude of the corrections to scaling also grows with @xmath13 .
this is particularly important for @xmath162 , which controls the main contribution to the corrections to scaling for large system sizes @xmath14 .
the absolute value of this parameter for @xmath19 is very likely to be at least two orders of magnitude larger for @xmath55 than for @xmath8 .
this translates into much slower convergence of @xmath172 for @xmath55 than for @xmath8 ( c.f .
figure [ fig : ypc - individual ] and @xcite ) .
actually , for @xmath8 the value of the linear coefficient , @xmath162 , is so close to zero that the convergence rate of @xmath172 in simulations is effectively controlled by the quadratic term , @xmath163 , an effect also reported in @xcite . once @xmath92 is known with sufficiently low uncertainty , one can try and use it to reduce the corrections to scaling by assuming @xmath228 in ( [ eq : tau ] ) .
this method turned out very successful for @xmath10 @xcite , but in this case @xmath92 is known exactly @xcite .
availability of the exact value of @xmath92 appears crucial , because the leading term in ( [ eq : phi - scaling ] ) has special properties only at @xmath228 so that small errors in @xmath92 may disturb the fitting .
we checked that , as expected , setting @xmath228 in dimensions @xmath229 significantly increased the convergence rate ; however , it did not result in more accurate values of the percolation threshold , probably due to the errors in @xmath92 and the fact that the uncertainty of the extrapolated value ( @xmath146 ) is closely related to the uncertainty of the data being extrapolated ( @xmath73 ) , which is independent of @xmath230 ( data not shown ) .
finally , we checked that the value of @xmath92 is universal for other definitions of percolation in finite - size systems .
if we assumed that a system percolates when a wrapping cluster appears in any direction ( case b ) , we obtained @xmath231 , @xmath232 , @xmath233 , @xmath234 , and @xmath235 for @xmath161 , respectively .
when we waited until a wrapping condition was satisfied along all @xmath13 directions ( case c ) , we obtained @xmath236 , 0.0291(3 ) , 0.0187(3 ) , 0.0224(9 ) , and 0.043(3 ) for @xmath161 , respectively .
the values for @xmath8 are consistent with those reported in @xcite , @xmath237 and @xmath238 .
treating continuous percolation of aligned objects as a limit of the corresponding discrete model turned out to be an efficient way of investigating the continuous model . using this approach
we were able to determine the percolation threshold for a model of aligned hypercubes in dimensions @xmath239 with accuracy far better than attained with any other method before .
actually , for @xmath4 the uncertainty of the continuous percolation threshold is now so small that it matches or even slightly surpasses that for the site percolation .
we were also able to confirm the universality of the wrapping probability @xmath92 and determine its value for @xmath4 for several definitions of the onset of percolation in finite - size systems .
the method proposed here has several advantages .
first , it allows one to reduce the statistical noise of computer simulations by transforming the results from the microcanonical to canonical ensemble .
second , it allows to exploit the universality of the convergence rate of the discrete model to the continuous one , which we found to be controlled by a universal exponent @xmath145 for all @xmath13 .
finally , it can be readily applied to several important shapes not studied here , like hyperspheres or hyperneedles .
one drawback of the method is that it does not seem suitable for continuous models in which the obstacles are free to rotate , e.g. randomly oriented hypercubes .
we also did not take into account logarithmic corrections to scaling at the upper critical dimension @xcite , which may render our error estimates at @xmath240 too optimistic .
our results for the continuous percolation threshold in dimensions @xmath241 are incompatible with those reported recently in @xcite .
the reason for this remains unknown , and we guess that they are related to corrections to scaling , which quickly grow with @xmath13 . finally , we have managed to improve the accuracy of the critical exponent @xmath2 measurement in dimensions @xmath7 .
the source code of the software used in the simulations is available at https://bitbucket.org/ismk_uwr/percolation .
the calculations were carried out in the wrocaw centre for networking and supercomputing ( http://www.wcss.wroc.pl ) , grant no . 356 .
we are grateful to an anonymous referee for pointing our attention to the gaussian approximation of the binomial distribution .
10 h.g .
ballesteros , l.a .
fernndez , v. martin - mayor , a. muoz sudupe , g. parisi , and j.j .
ruiz - lorenzo .
measures of critical exponents in the four - dimensional site percolation .
, 400(3 - 4):346 351 , 1997 . | we propose a method of studying the continuous percolation of aligned objects as a limit of a corresponding discrete model .
we show that the convergence of a discrete model to its continuous limit is controlled by a power - law dependency with a universal exponent @xmath0 .
this allows us to estimate the continuous percolation thresholds in a model of aligned hypercubes in dimensions @xmath1 with accuracy far better than that attained using any other method before .
we also report improved values of the correlation length critical exponent @xmath2 in dimensions @xmath3 and the values of several universal wrapping probabilities for @xmath4 .
_ keywords _ : percolation threshold ; continuous percolation ; critical exponents ; finite - size scaling , percolation wrapping probabilities |
conventional statistical mechanics addresses homogeneous macroscopic systems in the thermodynamic limit .
these are traditionally treated in canonical ensembles controlled by intensive temperature @xmath1 , chemical potential @xmath2 and/or pressure @xmath3 . in the canonical ensemble
the heat capacity is given by the fluctuation of the energy @xmath4 .
as in astro - physics the heat capacity is often negative it is immediately clear that astro - physical systems are not in the canonical ensemble .
this was often considered as a paradoxical feature of the statistics of self - gravitating systems . here
we will show that this is not a mistake of equilibrium statistics when applied to self - gravitating systems but is a generic feature of statistical mechanics of any many - body systems at phase separation , independently of the range of the interactions , ref.@xcite . as the original motivation of thermodynamics was the understanding of boiling water in steam - engines
, this points to a basic misconception of conventional canonical thermo - statistics . as additional benefit of our reformulation of the basics of statistical mechanics by microcanonical statistics
there is a rather simple interpretation of entropy , the characteristic entity of thermodynamics .
boltzmann , ref.@xcite , defined the entropy of an isolated system in terms of the sum of all possible configurations , @xmath5 , which the system can assume consistent with its constraints of given energy , volume , and further conserved constraints:@xmath6as written on boltzmann s tomb - stone , with @xmath7 in semi - classical approximation .
@xmath8 is the total energy , @xmath9 is the number of particles and @xmath10 the volume . or , more appropriate for a finite quantum - mechanical system : @xmath11\label{quantums }
= \sum{\scriptscriptstyle\begin{array}{ll}\mbox{all eigenstates
n of h with given n,$v$,}\\\mbox{and } e < e_n\le e+\epsilon_0\nonumber \end{array}}\ ] ] and @xmath12 the macroscopic energy resolution .
this is still up to day the deepest , most fundamental , and most simple definition of entropy .
_ there is no need of the thermodynamic limit , no need of concavity , extensivity , and homogeneity_. schrdinger was wrong saying that microcanonical statistics is only good for diluted systems , ref.@xcite .
it may very well also address the solid - liquid transition ref.@xcite and even self - gravitating systems as we will demonstrate in this article . in its semi - classical approximation , eq.([boltzmann ] ) , @xmath13 simply measures the area of the sub - manifold of points in the @xmath14-dimensional phase - space ( @xmath15-space ) with prescribed energy @xmath8 , particle number @xmath9 , volume @xmath10 , and some other time invariant constraints which are here suppressed for simplicity . because it was planck who coined it in this mathematical form ,
i will call it the boltzmann - planck principle .
the boltzmann - planck formula has a simple but deep physical interpretation : @xmath5 or @xmath16 measure our ignorance about the complete set of initial values for all @xmath14 microscopic degrees of freedom which are needed to specify the @xmath9-body system unambiguously , ref.@xcite . to have complete knowledge of the system we would need to know [ within its semiclassical approximation ( [ boltzmann ] ) ] the initial positions and velocities of all @xmath9 particles in the system , which means we would need to know a total of @xmath14 values
. then @xmath5 would be equal to one and the entropy , @xmath16 , would be zero . however , we usually only know the value of a few parameters that are conserved or change slowly with time , such as the energy , number of particles , volume and so on .
we generally know very little about the positions and velocities of the particles .
the manifold of all these points in the @xmath14-dim .
phase space , consistent with the given conserved macroscopic constraints of @xmath17 , is the microcanonical ensemble , which has a well - defined geometrical size @xmath5 and , by equation ( [ boltzmann0 ] ) , a non - vanishing entropy , @xmath18 .
the dependence of @xmath18 on its arguments determines completely thermostatics and equilibrium thermodynamics .
clearly , hamiltonian ( liouvillean ) dynamics of the system can not create the missing information about the initial values - i.e. the entropy @xmath18 can not decrease .
as has been further worked out in ref.@xcite and more recently in ref.@xcite the inherent finite resolution of the macroscopic description implies an increase of @xmath5 or @xmath16 with time when an external constraint is relaxed , c.f.chapter [ second ] . such is a statement of the second law of thermodynamics , ref.@xcite , which requires that the _ internal _ production of entropy be positive or zero for every spontaneous process .
analysis of the consequences of the second law by the microcanonical ensemble is appropriate because , in an isolated system ( which is the one relevant for the microcanonical ensemble ) , the changes in total entropy must represent the _ internal _ production of entropy , see above , and there are no additional uncontrolled fluctuating energy exchanges with the environment .
the weight @xmath19 of configurations with energy e in the definition of the canonical partition sum @xmath20 becomes here _ bimodal _ , at the transition temperature it has two peaks , the liquid and the gas configurations which are separated in energy by the latent heat .
consequently @xmath0 must be convex ( like @xmath21 ) and the weight in ( [ canonicweight ] ) has a minimum between the two pure phases .
of course , the minimum can only be seen in the microcanonical ensemble where the energy is controlled and its fluctuations forbidden .
otherwise , the system would fluctuate between the two pure phases ( inter - phase fluctuation ) by an , for macroscopic systems even macroscopic , energy @xmath22 of the order of the latent heat .
i.e. _ the convexity of @xmath0 is the generic signal of a phase transition of first order and of phase - separation , ref.@xcite_. this applies also to macroscopic systems coupled by short or long - range interactions .
of course , even there phase - separation exists and there is a minimum ( convexity ) in @xmath0 and , consequently , negative heat capacity@xcite .
it must be emphasized : in conventional canonical statistics phase - separated systems are treated by artificial constructions of two coexisting simple configurations of each phase like a single drop with vapor around ( wulff - construction ) .
but this is not a common ensemble of various droplets with its inter - phase fluctuations leading e.g. to a negative heat capacity .
such macroscopic energy fluctuations and the resulting negative heat capacity are already early discussed in high - energy physics by carlitz , ref.@xcite .
classical boltzmann - planck microcanonical statistics applies very well to the equilibrium of self - gravitating systems provided we put the system into a box ( ignoring slow evaporation ) and use a hard core in the interaction at short distances where anyhow non - gravitational physics like hydrogen burning dominates gravity .
padmanabhan in his stimulating review ref.@xcite discussed this point in detail .
we simplify the original many - body problem by the mean - field approximation , details in ref@xcite . here
the ( certainly also important ) many - body correlations are ignored .
the entropy is then a function of the one - body densities @xmath23 .
@xmath24 is maximized and a closed equation for the optimal one - body density distributions obtained . to get this result and to get finite @xmath23
the short - distance cut - off and a hard core in the two - body interaction @xmath25 is essential .
the resulting mean - field theory therefore does _ not _ lead to an isothermal sphere .
the latter is a solution of poisson s equation ( ref.@xcite ) with a strict @xmath26 interaction . moreover , for equilibrium one does not need any exotic q - deformed statistics , ref.@xcite .
the error of ref.@xcite is the use of a generalized variant of boltzmann - gibbs _ canonical _ ensemble instead of the basic microcanonical one , a point on which tsallis seems to agree in his discussion in ref.@xcite , see also ref.@xcite . in fig .
[ cover ] we show the equilibrium distributions of self - gravitating systems under large angular - momentum in the mean - field approximation : the necessity of using `` extensive '' instead of `` intensive '' control parameters is explicit in astrophysical problems .
e.g. : for the description of rotating stars one conventionally works at a given temperature and fixed angular velocity @xmath27 c.f .
of course in reality there is neither a heat bath nor a rotating disk .
moreover , the latter scenario is fundamentally wrong as at the periphery of the disk the rotational velocity may even become larger than velocity of light .
non - extensive systems like astro - physical ones do not allow a `` field - theoretical '' description controlled by intensive fields !
e.g. configurations with a maximum of random energy on a rotating disk , i.e. at fixed rotational velocity @xmath27 ( which is intensive ) : @xmath28 and consequently with the largest entropy are the ones with smallest moment of inertia @xmath29 , compact single stars .
just the opposite happens when the angular - momentum @xmath30 ( extensive ) and not the angular velocity @xmath27 is fixed:@xmath31then configurations with large moment of inertia are maximizing the phase space and the entropy .
i.e. eventually double or multi stars are produced , as observed in reality . in figure [ phased ]
one clearly sees the rich and realistic microcanonical phase - diagram of a rotating gravitating system controlled by the conserved `` extensive '' parameters energy and angular - momentum , ref.@xcite .
also the second law becomes significantly more transparent within microcanonical statistics : many formulations of the second law exist .
the understanding of entropy is sometimes obscured by frequent use of the boltzmann - gibbs canonical ensemble , and thermodynamic limit ; and the relationship of entropy to the second law is often beset with confusion between external transfers of entropy and its internal production .
perhaps the clearest statement is by prigogine , ref.@xcite : he distinguishes between internal and external entropy transfer .
whereas the transfer @xmath32 to an external heat bath can be positive , negative , or zero , the internal generation of entropy @xmath33 within the system is necessarily @xmath34 .
this is the sharpest formulation of the second law .
@xmath33 is best seen in the microcanonical ensemble , as there are no couplings to any heat bath and no sometimes uncontrolled exchanges of energy with it .
the second law is then most rigorously given by equation ( [ secondlaw ] ) : @xmath35 a geometric visualization how an equilibrium distribution in the @xmath14-dim phase space expands into a ( fat)-fractal ( gibbs ink lines ) after a constraint , here the wall between the phase - space volumes between @xmath36 and @xmath37 is suddenly released is given by figure ( [ box ] ) .
the liouville theorem tells us that the ( riemannian- ) size of it as given by eq.([boltzmann ] ) does not change . keeping in mind that due to the redundant information about the system ( c.f .
chapter [ entropy ] ) it is in general impossible to distinguish a point inside the fractal from a neighboring point outside .
therefore the proper way to `` measure '' the size of the manifold is by its box - counting `` measure '' .
i.e. the phase - space integrals in eq.([boltzmann ] ) or eq.([quantums ] ) have to be calculated by the _ box - counting volume _ @xmath38 as is explained in fig.([box ] ) ( details in refs.@xcite ) : @xmath39:=\underbar{$\lim$}_{\delta\to 0 } \delta^{6n } n_\delta[{\bf{m}}(e , n , t\gg t_0)]\label{boxvol}\ ] ] which is @xmath40 $ ] . that is the second law .
the story of applying microcanonical statistics to self - gravitating systems and eventually to cosmology is far open .
there are many questions to answer , e.g. : there is the problem of statistics in an expanding universe as discussed by padmanabhan ref.@xcite , the question of defining boltzmann - planck statistics in general relativity and connected the problem of avoiding field - theoretical descriptions with intensive statistical control - parameters or using them despite the fact that first order transitions demand extensive control parameters as discussed here .
the present author , coming from far away from nuclear and statistical physics , is open for any helpful suggestion .
an interdisciplinary meeting like this nbs2005 is here very important .
i like to thank francoise combes and raoul robert for this great opportunity .
conventional canonical statistics is insufficient to handle the original goal of thermodynamics , phase separations with their characteristic inter - phase fluctuations . here
the microcanonical ensemble is much richer and leads to _ negative specific heat_. this is thus _ not _ a characteristics of self - gravitating systems .
phase transitions can be found _ without the thermodynamic limit , without homogeneity , concavity , and extensivity_. microcanonical statistics opens to a much wider class of applications as e.g. astrophysical systems .
moreover it gives a simple formulation and proof of the second law .
negative heat - capacity at phase - separations in microcanonical thermostatistics of macroscopic systems with either short or long - range interactions .
http://arxiv.org/abs/condmat/0509234 , ( 2005 )
. d.h.e .
gross and j.f .
the microcanonical thermodynamics of finite systems : the microscopic origin of condensation and phase separations ; and the conditions for heat flow from lower to higher temperatures . , 122:224111 ; http://arxiv.org/abs/condmat/0503604 , ( 2005 ) . d.h.e .
ensemble probabilistic equilibrium and non - equilibrium thermodynamics without the thermodynamic limit . in andrei khrennikov , editor ,
_ foundations of probability and physics _ , number xiii in pq - qp : quantum probability , white noise analysis , pages 131146 , boston , october 2001 .
acm , world scientific .
votyakov , h.i .
hidmi , a. de martino , and d.h.e .
microcanonical mean - field thermodynamics of self - gravitating and rotating systems .
, 89:03110114 ; http://arxiv.org/abs/condmat/0202140 , ( 2002 ) . | conventional thermo - statistics address infinite homogeneous systems within the canonical ensemble .
( only in this case this is equivalent to the fundamental microcanonical ensemble . )
however , some 170 years ago the original motivation of thermodynamics was the description of steam engines , i.e. boiling water .
its essential physics is the separation of the gas phase from the liquid .
of course , boiling water is inhomogeneous and as such can not be treated by conventional thermo - statistics .
then it is not astonishing , that a phase transition of first order is signaled canonically by a yang - lee singularity . _
thus it is only treated correctly by microcanonical boltzmann - planck statistics_. it turns out that the boltzmann - planck statistics is much richer and gives fundamental insight into statistical mechanics and especially into entropy .
this can be done to a far extend rigorously and analytically .
as no extensivity , no thermodynamic limit , no concavity , no homogeneity is needed , it also applies to astro - physical systems .
the deep and essential difference between `` extensive '' and `` intensive '' control parameters , i.e. microcanonical and canonical statistics , is exemplified by rotating , self - gravitating systems . in the present paper
the necessary appearance of a convex entropy @xmath0 and negative heat capacity at phase separation in small as well macroscopic systems independently of the range of the force is pointed out .
thus the old puzzle of stellar statistics is finally solved , the appearance of negative heat capacity which is forbidden and can not appear in the canonical formalism .
presented at the nbs2005 at the observatoire de paris microcanonical statistics , first order transitions , phase separation , steam engines , negative heat capacity , self - gravitating and rotating stellar systems 01.55.+b,04.40.-b , 05.20.gg , 64.60.-i,65 , 47.27.eb , 97.80.-d |
ultra luminous compact x - ray sources ( ulxs ; makishima et al .
2000 ) are objects often seen in off - center regions of nearby spiral galaxies , and exhibit x - ray luminosities by far exceeding the eddington limit for a neutron star ( fabbiano . 1989 ; read et al .
1997 ; lira et al .
2000 ; roberts & warwick 2000 ; bauer et al . 2001 ; fabbiano et al . 2001 ; strickland et al . 2001 ) . as reported by many _
asca _ observers ( colbert & mushotzky 1999 ; kotoku et al . 2000 ; makishima et al . 2000 and references therein ) , the spectra of a majority of ulxs
are well fitted by multi - color disk blackbody model ( mcd model ; mitsuda et al .
1984 ; makishima et al . 1986 ; ebisawa et al . 1993 ) .
further considering their high luminosities ( @xmath2 ) , ulxs are hence inferred to be black - hole binaries ( bhbs ) of 10100 @xmath3 ( makishima et al .
2000 ; mizuno 2000 ; kubota et al .
2001 ) . a possible formation scenario of such massive ( @xmath4 ) bhs has been proposed by ebisuzaki et al .
alternatively , ulxs may be more ordinary bh ( or even neutron star ) binaries with mild x - ray beaming ( king et al .
2001 ) , although realistic x - ray beaming mechanisms are yet to be worked out . through two _
asca _ observations of the spiral galaxy ic 342 , kubota et al .
( 2001 ; hereafter paper i ) discovered that the two ulxs in it , called source 1 and source 2 ( fabbiano & trinchieri 1987 ) , make clear transitions between soft and hard spectral states .
since such a transition is characteristic of galactic / magellanic bhbs ( e.g. , maejima et al .
1984 ; zhang et al . 1997 ; wilms et al .
2000 ) , the bhb interpretation of ulxs has been much reinforced .
a similar spectral transition was observed from m81 x-9 ( la parola et al .
2001 ) , thought to be a ulx as well .
to conclusively establish the binary nature of ulxs , we in fact need one more step , i.e. , to detect their binary periods .
although the standard optical technique ( gies & bolton 1982 ; cowley et al . 1983 ; for review cowley et al .
1992 ) is currently unapplicable to ulxs due to the general lack of their optical identification , we may search their x - ray light curves for possible periodicity . according to the roche - lobe formula ( eggleton 1983 ) for a semi - detached binary system and kepler s law , the density @xmath5 ( g @xmath6 ) of the mass - donating star filling its roche - lobe is determined by the binary period @xmath7 ( hour ) as @xmath8 , if the mass - accreting star is more massive than the mass - donating star ( frank et al .
for example , if a ulx indeed consists of a bh of @xmath9 together with a @xmath10 main - sequence companion , we expect @xmath11 hours , because the latter has typically a radius of @xmath12 10 @xmath13 ( i.e. , @xmath14 ) , although it may finally evolve to a much larger size ( de loore et al .
1978 ; weiss 1994 ; stothers & chin 1999 ) .
actually , in ngc 5204 , a possible optical counterpart to a ulx has been discovered , and may be considered to be a supergiant o star ( roberts et al .
if the companion mass is instead @xmath15 ( @xmath12 4 @xmath13 radius ) , we expect @xmath16 hours .
thus , the orbital periods of such binary systems are expected in the range of several tens of hours . when the first
observation of ic 342 was conducted in 1993 , a significant x - ray variation was detected from source 1 on the time scale of @xmath12 10 hours ( okada et al .
1998 ) . however , the total span of that observation ( @xmath12 22 hours ) was too short to examine whether the variation is periodic or not .
in this paper , we report timing results on the two ulxs in ic 342 , source 1 and source 2 , based on the one - week _ asca _ observation conducted in 2000 .
we have indeed obtained evidence for @xmath12 31 or @xmath12 41 hour periodicity from source 2 .
the present _ asca _ observation of ic 342 , the second one described in paper i , was carried out from 2000 february 24 00:29 through march 1 21:09 ( ut ) , for a total time span of 557 ks .
after screening the data with the standard criteria , we obtained the net exposure of 262 ks with the gis and 244 ks with the sis .
as already described in paper i , this observation found source 1 in the hard spectral state while source 2 in the soft spectral state . referring to the x - ray images ( paper i ) , we accumulated photons over circular regions of @xmath17 radius around the two sources .
figure [ fig : lightcurves ] shows the gis and sis two - band light curves obtained from the two sources .
the photon arrival times have been converted to solar barycentric times .
the background counts contained in the light curves were evaluated using an off - source region in the image for the gis and other blank - sky observations for the sis , respectively , because no off - source region is available in the sis image .
thus , the background contributes @xmath18 to the light curves .
furthermore , the gis background is stable within @xmath18 ( rms , relative ) , implying that the background variation is below @xmath19 of the gis signal counts .
the variation of sis background is known to be similar to that of the gis .
thus , the background and its variation can be ignored in our subsequent timing study . in figure
[ fig : lightcurves ] , we can clearly see significant variations on @xmath12 one day time scale in the high - energy light curves of source 2 , consistently with the gis and sis data .
in contrast , the other light curves appear to be consistent with being constant .
actually , the value of reduced @xmath20 obtained by fitting a linear trend , indicated in figure [ fig : lightcurves ] , is 2.9 and 2.3 for the gis and sis high - energy light curves of source 2 , respectively , while it is below 1.5 for the other light curves . to examine the 210 kev gis and sis light curves of source 2 for the periodic behavior , we re - calculated them with a finer bin width of 256 s , and embedded them into a time span of 524 ks = 146 h ( 2048 bins ) .
we next subtracted the linear trend from the light curves and filled the data gaps with zeros . using an fft algorithm , we calculated power density spectra , separately for the two instruments , over a frequency range from @xmath21 hz ( @xmath22 s @xmath23 ks @xmath24 hours ) to @xmath25 hz ( @xmath26 s ) .
figure [ fig : power_spectra ] shows the power density spectra of the 210 kev gis and sis light curves of source 2 , together with that of source 1 in the same energy band .
we thus detect a strong power peak at @xmath27 hz ( period @xmath28 ks @xmath29 hours ) from source 2 , both instruments showing consistent results . the chance probability to detect such a strong power below @xmath30 hz with the two instruments , assuming that the source is varying randomly with the same rms amplitude as the actual data ,
is estimated by a monte carlo simulation to be @xmath31 .
the @xmath27 hz peak is absent in the 210 kev data of source 1 ( figure [ fig : power_spectra ] ) , indicating that the peak is unrelated to background . in the 0.72 kev band ,
neither of the two sources exhibits the @xmath27 hz peak .
another peak at @xmath32 hz ( @xmath33 ks ) corresponds to the orbital period of _
asca_. in order to further examine the possible periodicity found in the power - density spectra , we performed standard epoch - folding analysis .
that is , we folded the overall light curve ( of either gis or sis ) of source 2 into 7 bins at various trial periods , and examined the obtained folded light curves against the hypothesis of `` being constant '' through @xmath20-evaluation .
figure [ fig : epoch_folding ] shows the gis and sis periodograms of source 2 , thus calculated in the 210 kev band , together with that of sources 1 obtained in the same way . in the periodograms of both instruments for source 2 , we observe clear peaks at periods of @xmath34 ks ( @xmath0 hours ) and @xmath35 ks ( @xmath1 hours ) .
the reduced @xmath36 values are @xmath37 ( gis ) and @xmath38 ( sis ) for the 31 hour period , and @xmath39 ( gis ) and @xmath40 ( sis ) for the 41 hour period ; their significances are similarly high . these periods are not present in the 210 kev periodogram of source 1 ( figure [ fig : epoch_folding ] ) , or 0.72 kev periodogram of either source .
the peak periods do not depend significantly on the number of bins used in the folding .
the double period of 31 hours is also seen at @xmath41 hours .
we do not find other significant peaks in the periodograms .
the 31 hour period of source 2 means about 5 cycles across the total data span , and corresponds to the peak found at @xmath42 hz in the power density spectra ( figure [ fig : power_spectra ] ) .
the 41 hour period , in contrast , corresponds to about 3.8 overall cycles , and its counterpart ( to be expected at @xmath43 hz ) is weakly visible in the power density spectra , especially in the sis data .
figure [ fig : folded_lightcurves ] shows the gis and sis two - band folded light curves of source 2 , corresponding to the above two candidate periods ( 31 and 41 hours ) .
thus , a roughly sinusoidal modulation of @xmath445% amplitude is seen in the high - energy band , whereas the modulation is insignificant in the low - energy band with an upper limit of 4.0% and 3.9% ( rms ) for the 31 hour and 41 hour periods , respectively .
these properties do not differ by a large amount between the two candidate periods . to investigate spectral changes associated with these periods , we sorted the gis spectra into two phases ( bright and faint ) of the modulation , as indicated in figure [ fig : folded_lightcurves ] , separately for the two possible periods .
we did not analyze the sis spectra , since the long - term ccd degradation has made its response uncertain .
the obtained phase - resolved gis spectra for the 31 hour period are presented in figure [ fig : spectra ] , together with ratios to the mcd model that best fits the phase - averaged gis spectrum ( paper i ) .
thus , the spectrum changes by @xmath45% , with the change being larger in the harder band . similarly to paper
i , we fitted the two phase - resolved gis spectra successfully with the mcd model . because the insignificant modulation in lower energies means a constant absorption across the modulation phase
, we fixed the absorbing column to @xmath46 @xmath47 which was found through the phase - averaged spectral analysis ( paper i ) .
the inset in figure [ fig : spectra ] shows the obtained confidence contours on the plane of the two mcd model parameters , disk temperature and normalization .
thus , the two confidence regions do not overlap with each other , implying that the periodic modulation involves statistically significant changes in the accretion disk condition ( the temperature and/or the apparent radius ) .
although the linear trend shown in figure [ fig : lightcurves ] has a similar effect , it contributes little ( @xmath48 1% ) to this spectral modulation because the linear trend is averaged out by the folding process .
although figure [ fig : spectra ] refers to the 31 hour period , the 41 hour period yields very similar results .
during the _ asca _ long - look observation of ic 342 conducted in 2000 , source 2 , in its soft spectral state , varied clearly .
the variation has a typical time scale of one day , is more prominent in higher energies , and can be understood as changes in the mcd temperature and/or normalization .
these properties are reminiscent of the variation detected in the shorter _ asca _ observation in 1993 from source 1 , when it was in the high state ( okada et al .
mizuno ( 2000 ) and mizuno et al .
( 2001 ) observed similar variation from several other ulxs in the spectral high state .
thus , a mild intensity variation may be a relatively common phenomenon among high - state ulxs .
our timing analysis indicates that the source 2 variation is possibly periodic , with a suggested period of either 31 hours or 41 hours . of course
, we can not conclude for sure on the cyclic nature of the variation , nor can we distinguish between the two candidate periods , because the time span of the observation covers only 45 cycles of the suggested periods .
nevertheless , these values agree with what is expected for the orbital period of a semi - detached binary system consisting of a bh and a massive main - sequence star .
specifically , the @xmath7 vs @xmath5 relation described in 1 yields a main - sequence star mass of @xmath12 25 @xmath49 and @xmath12 50 @xmath49 for @xmath50 31 and 41 hours , respectively ; a lower mass is indicated , if the mass - donating star is evolved and hence has a larger size .
however , the bh mass remains rather unconstrained .
if the observed variability of source 2 really reflects an orbital effect , the phase - dependent spectral changes may be caused , e.g. , by an orbital modulation in the mass accreting rate , or eclipses of the inner disk .
actually , based on a 6.4-day observation with _
, boyd et al . ( 2001 ) report on a clear high - energy x - ray modulation in lmc x-3 synchronized with its 1.7-day orbital period .
recently , bauer et al .
( 2001 ) detected a very clear 7.5-hour periodic x - ray variation from a ulx in the circinus galaxy .
if the period is orbital in origin , the implied system has a few @xmath49 main - sequence star and a @xmath51 @xmath49 bh , assuming that its x - ray luminosity ( @xmath52 ) is close to the eddington limit without beaming .
this source may be a scale - down version of source 2 in ic 342 .
boyd , p. t. , smale , a. p. , & dolan , j. f. 2001 , , 555 , 822 bauer , f. e. , brandt , w. n. , sambruna , r. m. , chartas , g. , garmire , g. p. , kaspi , s. , & netzer , h. 2001 , , 122 , 182 colbert , e. j. m. , & mushotzky , r. f. 1999 , , 519 , 89 cowley , a. p. 1992
, , 30 , 287 cowley , a. p. , crampton , d. , hutchings , j. b. , remillard , r. , & penfold , j. e. 1983 , , 272 , 118 de loore , c. , de greve , j. p. , & vanbeveren , d. 1978 , , 34 , 363 ebisawa , k. , makino , f. , mitsuda , k. , belloni , t. , cowley , a. , schmidke , p. , & treves , a. 1993 , , 403 , 684 ebisuzaki , t. , et al . 2001 ( astro - ph/0106252 ) eggleton , p.
, , 268 , 368 fabbiano , g. 1989 , , 27 , 87 fabbiano , g. , & trinchieri , g. 1987 , , 315 , 46 fabbiano , g. , zezas , a. , & murray , s. s. 2001 , , 554 , 1035 frank , j. , king , a. , & raine , d. 1992 , cambridge university press , accretion power in astrophysics gies , d. r. , & bolton , c. t. 1982 , , 260 , 240 king , a. r. , davies , m. b. , ward , m. j. , fabbiano , g. , & elvis , m. 2001 , , 552 , 109 kotoku , j. , mizuno , t. , kubota , a. , & makishima , k. 2000 , , 52 , 1081 kubota , a. , mizuno , t. , makishima , k. , fukazawa , y. , kotoku , j. , ohnishi , t. , & tashiro , m. 2001 , , 547l , 119 ( paper i ) lira , p. , lawrence , a. , & johnson , r. a. 2000 , , 319 , 17 maejima , y. , makishima , k. , matsuoka , m. , ogawara , y. , oda , m. , tawara , y. , & doi , k. 1984 , , 285 , 712 makishima , k. , et al .
1986 , , 308 , 635 makishima , k. , et al .
2000 , , 535 , 632 mitsuda , k. , et al .
1984 , , 36 , 741 mizuno , t. 2000 , phd thesis , university of tokyo mizuno , t. , kubota , a. , & makishima , k. 2001 , , 554 , 1282 okada , k. , dotani , t. , makishima , k. , mitsuda , k. , & mihara , t. 1998 , , 50 , 25 la parola , v. , peres , g. , fabbiano , g. , kim , d. w. & bocchino , f. 2001 , , 556 , 47 read , a. m. , ponman , t. j. , & strickland , d. k. 1997 , , 286 , 626 roberts , t. p. , & warwick , r. s. 2000 , , 315 , 98 roberts , t. p. , goad , m. r. , ward , m. j. , warwick , p. t. o. , lira , p. , & hands a.
, , 325 , 7 stothers , r. b. , & chen , c. 1999 , , 522 , 960 strickland , d. k. , colbert , e. j. m. , & heckman , t. m. , 2001 ( astro - ph/0107115 ) wilms , j. , nowak , m. a. , pottschmidt , k. , heindl , w. a. , dove , j. b. , & begelman , m. c. 2000 , , 320 , 327 weiss , a. 1994 , , 284 , 138 zhang , s. n. , cui , w. , harmon , b. a. , paciesas , w. , remillard , r. e. , & van paradijs , j. 1997 , , 477 , l95 | a long ( 155 hours ) _ asca _ observation was performed of two ultra luminous compact x - ray sources , source 1 and source 2 , in the spiral galaxy ic 342 . while source 1 that was in a hard spectral state varied little , source 2 that was in its soft spectral state varied significantly on a time scale of about one day .
the rms variation amplitude amounts to 5% in the 210 kev band , but is less than 4% in 0.72 kev .
the variation involves statistically significant changes in the parameters describing multi - color disk blackbody emission from this source .
the variation is possibly periodic , with a period of either @xmath0 hours or @xmath1 hours
. both are consistent with the orbital period of a semi - detached binary formed by a black hole and a main - sequence star of several tens solar masses .
these results reinforce the interpretation of these x - ray objects in terms of accreting massive stellar black holes . |
a. details of the geometric argument for the current increase or decrease b. evaluation of the energy conservation @xmath71-function c. identification of distinct scattering processes in the asymptotic behavior of the rate of change of the current d. asymptotic behavior of the rate of change of the current for large excitation energies e. definition of the mean change of the current per electron scattering event
the dashed circles indicate constant energy contours and are a guide to the eye .
processes where the final states @xmath25 and @xmath26 are both either in the lower or upper band are similar to the case of @xmath90 and do not affect the current .
b ) for @xmath51 the allowed scattering processes can also be represented by an ellipse .
@xmath91 is the length of the semi - major axis and @xmath92 is the distance between the focal points .
the green dashed circle again indicates the fermi momentum . because of pauli s principle we have @xmath93 . note that here we do not have a restriction on @xmath94 other than @xmath95 .
c ) when averaging over all allowed @xmath24 , each @xmath24 parametrizes a different ellipse . for each @xmath24 and the resulting ellipse ( red )
there is a mirror image with respect to @xmath22 , @xmath96 ( blue ) , such that an increase in the current along @xmath39 averages to an increase along @xmath38 . in general there
can also be a change of the current in direction @xmath42 .
the change in the component parallel to @xmath22 , however , averages to zero when averaging over @xmath24 . by symmetry
there can be no change in current perpendicular to @xmath22 .
changes of the current in the @xmath42 direction are therefore not important for the average change in current .
the green dashes circle indicates the fermi momentum
. the allowed states @xmath24 must lie within this circle.,width=566 ] for the case of @xmath30 we have shown that for an arbitrary @xmath24 the component of the current along the direction given by the major axis of that specific ellipse , i.e. , along @xmath39 , increases due to e - e scattering . in general , the component of the current along @xmath40 might also change during a scattering event . however , we still need to sum over all @xmath24 . summing over
@xmath24 means summing over all possible ellipses and thus over all possible @xmath41 and @xmath42 . for a given @xmath24 and resulting @xmath97 , by symmetry
there is also a @xmath98 , i.e. , the mirror image of @xmath24 with respect to an axis parallel to @xmath22 , that leads to @xmath99 as shown in fig .
[ fig : suppmatellipse](c ) .
thus , when summing over all possible @xmath24 , the increase of the component of the current along @xmath39 averages to a current increase along @xmath38 . a change in the component along @xmath40 has components parallel and perpendicular to @xmath22 .
the component parallel to @xmath22 changes sign under the described reflection and thus averages to zero .
the component perpendicular to @xmath100 also has to average to zero because the rotational symmetry of our systems requires that the average change in current can only be in the @xmath38-direction . by symmetry
there can be no change in current perpendicular to @xmath38 .
an analogous argument to the case of @xmath30 can be made for the case of @xmath51 . here
the highly excited electron in the upper band scatters off an electron in the fermi sea in the lower band .
the scattering process that affects the current has final electronic states which are in different bands as illustrated in fig .
[ fig : suppmatellipse](a ) , i.e. , @xmath101 .
processes where both final states are in the same band are equivalent to the process for @xmath19 and thus do not change the current .
the condition for energy conservation becomes @xmath102 .
written in the following way , energy and momentum conservation can again be represented by an ellipse as illustrated in fig .
[ fig : suppmatellipse](b ) , @xmath103 note that in fig .
[ fig : suppmatellipse](b ) the orientations of @xmath24 and @xmath26 are reversed with respect to the analogous case of @xmath30 , and we have @xmath104 ( see fig . [
fig : suppmatellipse](a ) ) .
the initial and finals currents along @xmath105 ( see fig . [
fig : suppmatellipse](b ) ) are thus given by @xmath106 and @xmath107 and we have to compare @xmath108 to @xmath109 to analyze the change in current . because @xmath104 due to pauli s principle , the point of connection of @xmath22 and @xmath26 , as drawn in fig .
[ fig : suppmatellipse](b ) , will be inside the green dashed circle while the point of connection of @xmath25 and @xmath24 will lie outside this circle .
again it can be shown by elementary geometry that this implies @xmath110 for any scattering event , i.e. , the current decreases along @xmath111 .
analogously to above , averaging over all possibles scattering processes leads to a current decrease due to e - e scattering for @xmath51 .
after introducing the momentum transfer @xmath69 the energy conservation @xmath71-function can be written as @xmath112 where @xmath113 is the difference in energy of the initial and final scattering states . using the relation @xmath114 with @xmath115 , we can write @xmath116 where @xmath117 is the angle between @xmath22 and @xmath118 , such that @xmath119 where the factor of @xmath120 comes from the fact that @xmath121 has two zeroes in the interval @xmath122 $ ] with @xmath123
. analogously , the @xmath124-integration can be performed evaluating @xmath125 .
a ) scattering processes where @xmath126 is in the lower band are collinear and will not change the current .
b ) processes where @xmath127 , @xmath128 are in the same band are also collinear.,width=264 ] for @xmath30 we only need to consider processes like the one illustrated in fig . [
fig : holeprocess](a ) , in which the excited electron scatters off an electron in the upper band .
other allowed processes will be collinear and thus do not change the current . the rate of change of the electron current is then given by @xmath129 where the sum is over states with positive energy only and we decide to call the first term in the interaction matrix element direct and the second exchange. performing the sum
, we find that the contributions to the rate of change from @xmath130 and @xmath131 are equal , as can be easily seen by switching the labels @xmath132 in one of the terms .
we will call these contributions @xmath133 and @xmath134 respectively , with @xmath135 .
the remaining contribution from the interference term proportional to @xmath136 $ ] we will call @xmath137 .
analogously for the hole current , we only need to consider processes where the hole recombines with an electron in the lower(upper ) band thereby exciting an electron from the upper(lower ) band above the fermi energy , i.e. , the states @xmath127 , @xmath138 of the scattering event are in different bands as illustrated in fig .
[ fig : holeprocess](b ) . processes where @xmath127 , @xmath138 are in the same band are collinear and thus do not change the current .
the rate of change of the hole current can be written as @xmath139 where we restricted the sum to @xmath140 and @xmath141 and added the factor of @xmath120 in front for the other half of the sum .
we again call the first term of the interaction matrix element direct and the second exchange. analogously to above we call the corresponding contributions to the rate of change of the hole current @xmath142 and @xmath143 , and @xmath144 . here , the contributions from direct and exchange term are not equal because the states @xmath127 , @xmath138 are in different bands . switching the labels as for
the electron current does not transform one term into the other . for large excitation energies ,
i.e. , for @xmath75 , we find to lowest order that @xmath145 as shown in the next section this cancellation results in the fact that the rate of change of the total current to leading order is simply given by @xmath146 @xmath143 is governed by the interaction matrix element @xmath147 which describes processes where the photoexcited hole recombines with an electron from the upper band thereby exciting an electron from the lower band above the fermi energy .
these scattering processes involve large energy transfers of the order of the initial excitation energy @xmath11 .
we now want to calculate the asymptotic behavior of the rates of change of the individual electron and hole currents and of the total current .
we will show the calculation for @xmath133 in detail .
the calculations of the remaining contributions follow analogously . after introducing the momentum transfer @xmath69
, we use that the coulomb interaction in the direct term of eq .
is proportional to @xmath148 and the integral will be dominated by scattering events with small momentum transfer @xmath149 .
we now fix the initial momentum of the excited electron - hole pair @xmath150 such that the initial current is given by @xmath151 .
the difference of the velocities of states @xmath22 and @xmath25 can be approximated by zero , i.e. , @xmath152 we can also approximate the spin overlap of states @xmath22 and @xmath25 by @xmath153 , i.e. , @xmath154 the initial current flows in the negative @xmath155-direction so the rate of change of the current will only have an @xmath155-component . by the rotational symmetry of our problem there can be no change of the current along @xmath156 .
the sum of the quadratic direct and exchange contributions to the rate of change of the electron current given by eq .
can then be written as @xmath157 where @xmath158 is defined by @xmath159 .
as above , we use the identity @xmath160 and evaluate the @xmath124 and @xmath117 integrations with the two @xmath71-functions as shown in eqs . and .
in eq . , however , we not only have terms with @xmath161 and @xmath162 but also terms that contain @xmath163 . depending on the values of @xmath124 and @xmath117
, we can write @xmath164 since we have to integrate both @xmath124 and @xmath117 from @xmath165 to @xmath166 , integration of the terms proportional to @xmath167 gives zero . thus , in the integrand of eq . we can neglect the terms proportional to @xmath167 , leaving us with a function that only depends on @xmath161 and @xmath162 . performing the @xmath124 and @xmath117 integrations using eq . and simplifying the result , we are left with @xmath168 for the integral over small @xmath169 we can approximate the integrand further . introducing dimensionless parameters @xmath170 , @xmath171 , and @xmath172 and shifting the integration variable @xmath173 , we get @xmath174 with @xmath175 from numerical integration and @xmath176 and @xmath177 as in the main text . for the remaining integral we also use the dimensionless parameters @xmath178 , @xmath179 , and @xmath180 .
if the integral converges , then the integrand has to go to zero faster than @xmath181 , i.e. , the weight is negligible for large @xmath182 and we are still allowed to approximate @xmath169 . the integral becomes @xmath183 we are interested in the limit @xmath184 . to avoid numerical integration up to infinity we use that for the region @xmath185 we can approximate @xmath186 and get @xmath187 where @xmath188 is a cutoff that ensures that the approximation @xmath185 is valid .
the remaining part of the integral we can not approximate further and we have to integrate @xmath189 numerically . for @xmath190 we get @xmath191 and @xmath192 and for @xmath193 we get @xmath194 and @xmath195 . both cutoffs give us the same final result of @xmath196 an analogous calculation for the interference term shows that @xmath197 , which is of higher order . it can be easily shown that to lowest order in @xmath12 , @xmath198 and @xmath199 just differ by a sign . in fig .
[ fig : holeprocess](b ) , labeling the initial states by @xmath22 , @xmath26 and the final states by @xmath25 , @xmath24 , i.e. , switching @xmath200 , and making use of the approximations and , the direct term of the hole current can be written as @xmath201 where @xmath202 .
the transformation @xmath203 , leads to @xmath204 and @xmath205 . with @xmath206 and @xmath207
, we find to lowest order that @xmath208 analogous calculations to the one above for @xmath209 lead to @xmath210 calculating the rate of change of the total current , we then get for the asymptotic behavior in the limit @xmath211 , @xmath212 \nonumber\\ & \approx -\frac{d}{9 } \left(\frac{\epsilon_f}{\epsilon_1}\right)^{3/2 } \hat{j_0 } + \mathcal{o}\left[\left(\frac{\epsilon_f}{\epsilon_1}\right)^{2 } \right].\end{aligned}\ ] ]
the mean change in current per scattering event is defined by @xmath213 while @xmath214 is always well defined , @xmath4 diverges for a perfectly linear dispersion because the phase space for collinear scattering becomes infinite .
we regularize this by introducing a physical and commonly used particle - hole asymmetry , such that @xmath215 . when calculating the now well defined @xmath4 we have to take into account all allowed scattering processes .
processes that are collinear and can be neglected in the calculation of the rate of change of the current have to be included in @xmath4 . | we study how electron - electron ( e - e ) interactions affect the current carried by excited electrons and holes in dirac systems such as graphene or topological insulators .
we find that the current shows distinctly different behavior for e - e collisions involving the electron or the hole .
most surprisingly , for positive fermi energy , collisions of the electrons can substantially increase the current .
this remarkable amplification of the current can be of the order of @xmath0 per scattering event .
it also causes a strong suppression of the overall current relaxation rate of a photoexcited electron - hole pair , with the amplification of the electron current making up for a fast decay of the hole current . _
introduction._dirac systems like graphene and topological insulators ( tis ) are a central research topic in condensed matter physics .
the easy fabrication process of graphene @xcite and the discovery of a variety of materials that are 2d @xcite or 3d @xcite tis are but two reasons why dirac systems are studied intensively . because of their linear dispersion and unique helical ( pseudo)spin structure , graphene and tis might be valuable materials for spintronic devices @xcite .
in addition , graphene and tis are promising systems for applications in the rising field of optoelectronics , for instance as transparent conductors or photodetectors @xcite .
photocurrents provide an interesting probe of the optoelectronic properties of dirac materials , and have been measured in both systems @xcite .
their magnitude is governed by a competition between carrier excitation which is asymmetric in momentum space and current relaxation @xcite .
while there has been extensive research on energy relaxation of excited electrons both theoretically @xcite and experimentally @xcite , current relaxation of photoexcited carriers in dirac systems remains much less unexplored @xcite .
in general , current relaxation occurs through impurity , electron - phonon , or electron - electron ( e - e ) scattering . here
we study highly excited electrons , holes , or electron - hole pairs in the dirac cone and assume that e - e interactions provide the dominant relaxation mechanism [ see fig .
[ fig : relax1](a ) ]
. interestingly , e - e scattering does not contribute to current relaxation when the carrier dispersion is quadratic .
indeed , for quadratic dispersions velocity is proportional to momentum and thus momentum conservation implies current conservation .
in contrast , in dirac systems with their linear dispersion velocity is no longer proportional to momentum and the current can change and relax by e - e scattering .
relative to the initial current @xmath1 _ vs _ @xmath2 for fixed initial energy @xmath3 ev .
the blue and red shaded areas indicate current amplification and relaxation , respectively .
the rate of change of the current is normalized by the total scattering rate @xmath4 .
the results are obtained for realistic parameters for bi@xmath5se@xmath6 , including particle - hole asymmetry @xmath7 ev @xmath8 , @xmath9 m / s @xcite , and @xmath10 @xcite ( see text for definitions ) .
inset : schematic of the excitation of an electron - hole pair by the initial photoexcited electron .
, width=321 ] relative to the initial current @xmath1 _ vs _ @xmath2 for fixed initial energy @xmath3 ev .
the blue and red shaded areas indicate current amplification and relaxation , respectively .
the rate of change of the current is normalized by the total scattering rate @xmath4 .
the results are obtained for realistic parameters for bi@xmath5se@xmath6 , including particle - hole asymmetry @xmath7 ev @xmath8 , @xmath9 m / s @xcite , and @xmath10 @xcite ( see text for definitions ) .
inset : schematic of the excitation of an electron - hole pair by the initial photoexcited electron .
, width=124 ] we find that this current relaxation process has rather surprising properties .
consider first the relaxation of the current associated with a single excited electron or hole above the fermi sea .
we find that the change in current strongly depends on the position of the fermi level and , most importantly , changes sign as the fermi energy crosses the dirac point .
this leads to the surprising conclusions that for an excited electron , the current actually increases rather than decreases by e - e collisions when the fermi energy is above the dirac point .
similarly , for an excited hole , the current increases when the fermi energy lies below the dirac point . the mean change of the current per scattering event is illustrated in fig .
[ fig : meancurrkf ] and is of the order of @xmath0 for realistic parameters .
ultimately , these remarkable results can be traced back to the fact that at zero temperature , there is no current relaxation when the fermi energy is right at the dirac point . for a photoexcited electron - hole pair ,
the relaxation of the total current involves a subtle interplay between the electron and the hole contribution .
we find that overall , e - e scattering decreases the total current of the electron - hole pair , but the relaxation is strongly suppressed due to cancellations between electron and hole processes . in the limit of large excitation energies @xmath11 of the initial carriers ,
the rate of change of both electron and hole currents varies linearly with @xmath12 , but with opposite signs .
these linear terms cancel in the relaxation of the electron - hole pair and relaxation of the total current is dominated by subleading contributions which we find to scale as @xmath13 . and ( b ) @xmath14 .
( c ) a possible relaxation process of the excited hole.,width=302 ] _ analysis of kinematic constraints._the surprising possibility of an increase in current due to e - e scattering can be seen most directly by analyzing the kinematic constraints of the scattering process .
energy and momentum conservation demand @xmath15 here , @xmath16 ( @xmath17 ) is the momentum of the initial ( final ) electrons and @xmath18 for the upper ( conduction ) and lower ( valence ) band , respectively .
expressed in terms of momentum , eq . depends on whether the specific scattering process is intraband or interband , with the allowed scattering processes depending on the fermi energy .
if the fermi energy lies at the dirac point , @xmath19 , a typical relaxation process of a highly excited electron is illustrated in fig .
[ fig : relax1](a ) .
the excited electron relaxes by scattering off an electron in the fermi sea , creating a hole in the valence band and an additional electron in the conduction band .
energy conservation , eq .
, demands @xmath20 , which takes into account that the electron in the fermi sea has a negative energy , @xmath21 .
thus , the length of vector @xmath22 must be equal to the sum of the lengths of the remaining three vectors .
this is only satisfied for collinear scattering so that initial and final states have the same velocities , i.e. , @xmath23 , and the current remains unchanged in the e - e collision . for dirac systems with the fermi energy at the dirac point , e - e interactions therefore do not relax current . .
@xmath22 and @xmath24 are drawn head - to - tail starting and ending at the left and right focal points ( red dots ) . @xmath25 and @xmath26 are drawn in a similar fashion and touch at points that lie on the ellipse .
@xmath27 and @xmath28 are the lengths of the semi - major axis and the distance between focal points , respectively .
the green dashed circle indicates the fermi momentum . because of
pauli s principle we have @xmath29 and the point where @xmath25 is connected to @xmath26 must lie on the ellipse outside the green dashed circles , while the point of connection of @xmath22 and @xmath24 has to lie inside the green dashed circle.,width=207 ] when the fermi energy lies above the dirac point , i.e. , @xmath30 , the excited electron can interact with fermi - sea electrons which are either in the conduction ( @xmath31 ) or the valence ( @xmath32 ) band [ see fig .
[ fig : relax1](b ) ] , corresponding to the processes @xmath33 or @xmath34 .
the previous argument for @xmath19 implies that a collision with electrons in the valence band , @xmath34 , is collinear and does not relax current .
thus , we only need to consider intraband processes [ see fig . [
fig : relax1](b ) ] .
energy conservation , i.e. , @xmath35 from eq . , and
momentum conservation can now be graphically interpreted in terms of an ellipse as illustrated in fig .
[ fig : ellipse ] .
the energy @xmath36 defines the semi - major axis , while @xmath37 is the distance between the focal points .
this construction implies that the current actually increases along the direction of the initial current @xmath38 . in a first step , we assume that not only @xmath22 but also @xmath24 is fixed .
the resulting ellipse is defined by the axes @xmath39 and @xmath40 ( see fig .
[ fig : ellipse ] ) . we will show that a summation over all @xmath25 and @xmath26 restricted to the ellipse leads to a current increase along @xmath41 as well as a change in current along @xmath42 . in a second step ,
we sum over all @xmath24 , i.e. , over all ellipses .
it turns out that a possible change of the current along @xmath42 averages to zero due to the rotational symmetry of the problem .
remarkably , the increase in current along @xmath41 averages to an increase along the direction of the initial current @xmath38 ( see @xcite for more details ) .
the initial and final currents along @xmath41 are given by ( @xmath43 , @xmath44 for brevity ) @xmath45 where the angles are defined as in fig .
[ fig : ellipse ] . to analyze the change in current
, we have to compare @xmath46 to @xmath47 .
it can be shown by elementary geometry that the sum of the cosines , restricted to the ellipse , has a maximum for the symmetric case where the of connection of the corresponding vectors lies on @xmath42 , and falls off monotonically away from the maximum .
we know from pauli s principle that @xmath48 .
this implies that the point of connection of @xmath22 and @xmath24 must lie inside the green dashed circle of radius @xmath49 while the point of connection of @xmath25 and @xmath26 must lie outside this circle ( see fig .
[ fig : ellipse ] ) and thus closer to the @xmath42-axis .
hence , @xmath50 for any scattering event , i.e. , the current increases along @xmath41 .
averaging over @xmath24 , i.e. , averaging over all ellipses , leads to an average increase of the current along @xmath22 @xcite .
an analogous argument shows that e - e scattering decreases the current when @xmath51 @xcite . _
quantitative analysis._quantitatively , the current relaxation rate for the optically excited electron - hole pair can be obtained within a golden - rule approach . for definiteness
, we consider the surface states of the ti @xmath52 , described by the ( second - quantized ) dirac hamiltonian @xmath53 where @xmath54 is the coulomb interaction and @xmath55 describes the single - particle dirac dispersion with eigenenergies @xmath56 and eigenstates @xmath57 .
the initial photoexcitation creates an electron - hole pair with fixed momentum @xmath22 [ see fig . [
fig : relax1](a ) ] .
then , the rate of change of the electron and hole currents is @xmath58[1-f^{(e / h)}(\epsilon_{\bm{k}_2 ' } ) ] , \label{eq : currderiv}\end{aligned}\ ] ] where the velocity is @xmath59 and @xmath60 denote the fermi distribution function of electrons and holes , respectively .
the transition rate is given by @xmath61 with @xmath62 and interaction matrix element @xmath63.\ ] ] here , @xmath64 is the surface area of the system and @xmath65 the screened coulomb interaction where @xmath66 is the thomas - fermi wave vector with @xmath67 .
as photoexcitation creates highly excited electron - hole pairs , we can set @xmath68 .
. can be simplified by introducing the momentum transfer @xmath69 and the identity @xmath70 .
then , in the thermodynamic limit the two @xmath71-functions can be used to eliminate the angular integrals leaving us with a three - dimensional integral which can be solved numerically for general parameters and analytically in limiting cases @xcite .
we first evaluate the expressions numerically for a particular carrier type , namely the photoexcited electron , and compute the mean change of current per scattering event . to make our results realistic
, we include the particle - hole asymmetry of the dispersion through @xmath72 @xcite . as illustrated in fig .
[ fig : meancurrkf ] , the current relaxes for negative fermi energies but becomes amplified for positive fermi energies , with the current enhancement being of the order of @xmath0 . from now on , we assume a perfectly linear dispersion , i.e. , @xmath73 , and @xmath30 for definiteness .
the case of negative fermi energy follows by electron - hole symmetry .
the rates of change of the electron , hole , and total currents , obtained by numerically integrating eq . are illustrated in the inset of fig .
[ fig : currderiv ] .
as already seen , e - e scattering increases the electron current ( red squares ) and decreases the hole current ( green diamonds ) . for large fermi energies
@xmath74 the rate of change approaches zero for the electron current and remains finite for the hole current , reflecting the different behavior of the phase space for scattering in the two cases [ see figs .
[ fig : relax1](b ) and ( c ) ] .
the e - e scattering also relaxes the total current ( blue circles ) but there are substantial cancellations between the electron and hole contributions . to quantify these cancellations
, we analytically explore the asymptotic behavior of eq . for large excitation energies and small fermi energy , i.e. , for the limit @xmath75 .
here , we focus on the results . details of the calculations can be found in @xcite . for the electron current , we only need to consider the scattering process illustrated in fig .
[ fig : relax1](b ) because interband processes , i.e. , @xmath76 , are collinear and do not change the current , as shown above by the geometric argument . for the hole
, we have to consider scattering processes like the one illustrated in fig .
[ fig : relax1](c ) . the hole can recombine with an electron in the valence band , exciting an electron from the conduction band above the fermi energy , i.e. , @xmath77 , or the hole can recombine with an electron in the conduction band , exciting an electron from the valence band above the fermi energy , i.e. , @xmath78 .
other allowed processes will be collinear .
we find that the asymptotic behavior of the rate of change of the electron and hole currents is given by @xmath79 where @xmath80 @xcite , @xmath81 stands for the electron and hole current respectively , and @xmath82 is the initial current of magnitude @xmath83 of the photoexcited carrier .
this result has several interesting aspects .
first , the time scale on which the initial current changes is independent of the large initial excitation energy @xmath11 of the photoexcited carrier and instead depends on the fermi energy only .
this is a consequence of the fact that the typical energy transfer in the relevant e - e collisions is of the order of the fermi energy .
secondly , to this order the rates of change of electron and hole currents differ only in their sign and thus cancel exactly .
thus , the rate of change of the total current of the photoexcited electron - hole pair is indeed much smaller and must scale with a higher power of @xmath12 .
we find that @xcite @xmath84 the relaxation of the total current is suppressed for small @xmath12 and even vanishes in the limit @xmath85 .
fig .
[ fig : currderiv ] shows the rate of change of the total current for small @xmath12 determined by numerically integrating eq .
( blue circles ) , and the asymptotic behavior given by eq .
( red solid line ) .
these asymptotic behaviors of the total current and the individual electron and hole currents can be traced back to distinct scattering processes .
the amplification and relaxation of the individual currents are governed by scattering processes with small energy transfers of the order of @xmath2 .
in contrast , for the total current the contributions with small energy transfer cancel exactly to the order considered and the result in eq
. arises solely from scattering processes with large energy transfers of the order of @xmath11 . specifically , the relaxation of the total current is dominated by the interband hole process , @xmath78 , where the hole recombines with an electron from the conduction band while exciting an electron from the valence band to empty states in the conduction band .
the predominance of scattering events with large energy transfers of the order of @xmath11 also explains why the relaxation vanishes in the limit of @xmath86 . and the asymptotic behavior @xmath87 given by eq .
( red ) .
inset : rates of change of the electron ( red squares ) , hole ( green diamonds ) , and total ( blue circles ) currents .
parameters as in fig .
[ fig : meancurrkf ] .
relaxation of the total current is strongly suppressed due to cancellations between the electron and hole contributions .
results for @xmath51 follow by electron - hole symmetry.,width=321 ] _ conclusion._motivated by photocurrent measurements on various dirac systems , we investigated the interaction - induced relaxation of photocurrents in clean dirac systems and uncovered a surprising effect : for a single excited electron , the current actually increases upon scattering with the electrons in the fermi sea , as long as the chemical potential is above the dirac point .
even for a single collision , this increase is substantial for realistic parameters .
furthermore , since high - energy electrons decay via a cascade of e - e collisions , the current increase can be further amplified by an additional factor up to @xmath88 , as the typical energy loss per collision is of the order of the fermi energy @xcite . while the relaxation cascade underlying this argument has been predicted theoretically , the experimental situation remains inconclusive @xcite , possibly due to competing optical - phonon collisions or a large radiation intensity which produces a high density of photoelectrons and phonons @xcite .
thus , the amplification effect may be most pronounced for excited electrons with energies below optical - phonon frequencies ( @xmath89 mev for graphene ) and for low - intensity irradiation .
this current amplification has important implications for photocurrents where it results in a substantial suppression of the current - relaxation rate of photoexcited electron - hole pairs , but might also be observable more directly in other types of experiments .
most promising may be time - resolved measurements of photocurrents as recently performed on graphene @xcite .
one might also expect a strong non - linear signature in iv characteristics , since high energy electrons produce a jet of induced current .
for the same reason , the current amplification might enhance the photoconductivity ( electron conductivity in the presence of light ) , with the effect increasing with the frequency of the irradiating light .
we intend to pursue a quantitative analysis of such effects in future work . _
acknowledgements._we thank j. eisenstein , erik henriksen , justin song and feng wang for discussions and acknowledge financial support through spp 1666 of the deutsche forschungssemester and a helmholtz virtual institute `` new states of matter and their excitations '' ( berlin ) as well as darpa , the iqim , an nsf institute supported by the moore foundation , and the humboldt foundation ( pasadena ) .
36 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , , , , , * * , ( ) . , * * , ( ) . , , , , , , , , * * , ( ) . , , , , , * * , ( ) . , * * , ( ) . , , , , , , , * * , ( ) . , * * , ( ) . , * * , ( ) .
, * * , ( ) . , , , , * * , ( ) . ,
* * , ( ) .
, , , * * , ( ) . , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , , , , , , ,
* * , ( ) . , , , , , , * * ( ) . , , ,
* * , ( ) . ,
* * , ( ) . , , , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) .
, , , * * , ( ) . , , ,
* * , ( ) . , , , , * * , ( ) .
, , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , , , , , , , * * , ( ) . , , , , , , , , , ,
* * , ( ) . , , , , , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . .
, , , * * , ( ) . .
* current amplification and relaxation in dirac systems - supplemental material * |
j. bae is supported in part by the hanyang university fellowship and y. kwon is supported in part by the fund of hanyang university . | it is questionable that grover algorithm may be more valuable than a classical one , when a partial information is given in a unstructured database . in this letter , to consider quantum search when a partial information is given , we replace the fourier transform in the grover algorithm with the haar wavelet transform .
we then , given a partial information @xmath0 to a unstructured database of size @xmath1 , show that there is the improved speedup , @xmath2 .
suppose that we have a problem of finding a desired one of unstructured @xmath1 items .
it is known that a classical search algrithm may take @xmath3 times , but the fast quantum search algorithm provides the quadratic speedup , @xmath4.@xcite the quantum algorithm whose central idea is amplitude amplification was first provided by grover in 1996.@xcite@xcite it is well - known that grover algorithm is 1)optimal in the context of applying unitary operators repeatdly and 2)efficient in searching a target of a unstructured database .
however , it is not evident that , when a partial information is given , the grover algorithm is still more valuable than a classical one . in this letter , we provide the quantum search algorithm which is able to benefit from a partial information .
the key building block in our construction is the ( haar ) wavelet transform instead of the fourier transform in the grover algorithm .
+ let us first consider the grover algorithm .
a bijection between a database and quantum states is necessary before applying the grover algorithm .
if a superposition of @xmath1 states is initially prepared , the grover algorithm amplifies the amplitude of the target state up to around one , while those of other states dwindle down to nearly zeros .
the amplitude amplification is performed by two inversion operations : inversion about the target by the oracle and inversion about the initial state by the fourier transform .
noting the fact that two simultaneous reflections about two mirrors crossing by an angle @xmath5 induce @xmath6 rotation , one may imagine that the inversions in the grover algorithm rotate the initial state around the target state.@xcite if the target state and the initial state are denoted by @xmath7 and @xmath8 respectively,(here the initial state is prepared by the fourier transform of a state @xmath9 , i.e. @xmath10 ) the inversion operators is expressed as @xmath11 since @xmath12 , the grover operator is written as @xmath13 then , after applying the operator @xmath4 times , the final state comes to @xmath14 , which is @xmath15 , @xmath16 .
the query complexity of this algorithm , the number of callings of the oracle , is therefore @xmath4 .
we here note that the running time has nothing to do with the choice of @xmath9 . + to consider the partial information , let us think about the following situation .
+ _ suppose that hanyang university library has @xmath1(@xmath17 , for some @xmath18 ) books .
each book has a code number identifying itself .
recently , the library decides to label new code numbers to each book since a better way to classify books has provided .
a labelling machine is applied to the tedious job .
each code is composed of @xmath19 numbers .
the first numbers must be @xmath20 , since they stands for the region .
and hanyang university is in south korea , so the second number is @xmath20.the other numbers stand for the category each book is included .
an earlier number means a larger category .
the rules of labeling a code number is that the @xmath21-th number , @xmath22 , is in @xmath23 $ ] for @xmath24 .
the following codes are a good example , _
@xmath25 the first code means @xmath26 and the next one @xmath27 .
the first two numbers @xmath28 stands for the region , and hanyang university . to aid the searching a book in the library
, the library also has implemented the grover algorithm in a quantum computer .
+ after the labelling is completed , one discovers a fault in the machine .
therefore , the library decides to do labelling again , fixing the fault in the labelling machine .
they then recommend to apply the grover algorithm to whom may be concerned to search a book , during labelling .
+ dr .
lee , a postdoc in physics , has to submit his research paper by tomorrow @xmath29 a.m. it is now 6 p.m. he wish to fill the reference section in the paper , but the library in mess blocks his work .
the director of the library recommends him to try the grover algorithm , but it was calculated , based on the number of books @xmath1 , that @xmath30 hours is the running time of the algorithm .
therefore , he concludes that the grover algorithm can not help him . at that moment
, he gets a call from the library that the only @xmath21-th number of all code numbers is correct .
+ it is sure that the grover algorithm can not complete the search task in time , since it takes over @xmath30 hours .
the only thing that dr .
lee can use in order to overcome his hopeless situation , is the fact that only the @xmath21-th number was correctly labelled .
the partial information may save his problem .
however , the grover algorithm can not benefit from the partial information of this problem . +
we here introduce the fast wavelet quantum search algorithm(wqsa ) , which is a modification of the grover algorithm by replacing the fourier transform with the haar wavelet transform , to resolve the situation of dr .
lee .
let us note that to apply the following operator @xmath31 is one iteration of the wqsa .
the haar wavelet transform @xmath32 is represented @xmath33 where @xmath34.\end{aligned}\ ] ] and @xmath35 the haar 1-level decomposition operator as follows ; @xmath36_{2^k \times 2^k}\end{aligned}\ ] ] we have used @xmath37 as the @xmath38 unit matrix and @xmath39 as the @xmath40 zero matrix .
it is clear that the wavelet transform @xmath32 is unitary since the operator @xmath41 is unitary .
+ since the operator is composed of the wavelet transform , consistently , the initial state is prepared by applying the inverse wavelet transform @xmath42 to a state @xmath9 , i.e. , the initial state is now @xmath43 .
the power of our wqsa appears in the initialization procedure .
it is quite remarkable that the state @xmath44 is a superposition of @xmath45 states , where @xmath46(@xmath21 is given by @xmath47 ) , while the state @xmath48 is a superposition of @xmath1 states .
then it is expected that the running time is @xmath2 .
choosing the initial state as @xmath44 , @xmath49 when the target state exists in the restricted domain of the @xmath45 states , we look forward to an improved speedup with the partial information .
since @xmath50 , by setting @xmath51 , @xmath52 and @xmath24 , and @xmath53 , the state @xmath44 is explicitly , @xmath54 the following diagram shows the set of states composing @xmath44 . .
the @xmath55axis runs from @xmath56 to @xmath57 .
when the initial state is @xmath58 or @xmath59 , the running time is the same to that of the grover algorithm since the lowest two retangulars include all states.,scaledwidth=40.0% ] we thus arrive at the following proposition .
suppose that we solve the problem of finding a desired one in the set @xmath60 .
given a partial information that the target state is in the subset @xmath61 , we complete the search task in @xmath62 times by choosing the initial state as @xmath63 .
proof ) let the target state @xmath64 . let us take the initial state as @xmath65 .
it suffices to show that it takes @xmath62 times for the wqsa to find the target state with the following setting .
+ let @xmath66 .
the wavelet quantum search operator is @xmath31 where @xmath32 is the haar wavelet transform .
applying the operator @xmath42 to the @xmath9 , we have the initial state @xmath67 and the state @xmath68 iterations of the operator @xmath69 create the following state , @xmath70 iterations is @xmath71 , where @xmath72 and @xmath73 .
hence , we have shown that the total number of iterations is @xmath62 . if we denote @xmath74 and @xmath46 , then the running time is written as @xmath2 q.e.d .
+ let us revisit the dr .
lee s problem.the partial information that the @xmath21-th number @xmath22 is correctly labelled leads dr .
lee to apply the wqsa so that the reference section is filled in time .
however note that there is no improvement in running time when the intial state is @xmath58 or @xmath59 since , in those cases , the initial state is still a superposition of @xmath1 states .
therfore , from the proposition , we know that he can complete the submission in time if the @xmath21 is larger than @xmath75 .
+ we have exerted how to utilize a partial information , in order to enhance quantum search .
our construction provides a way for quantum search to benefit from a partial information .
since the running time of the grover algorithm has nothing to do with the choice of unitary operator , the complexity of the wqsa is the same to the grover algorithm .
however , we have obtained the speedup @xmath2 by preparing the initial state as @xmath44 .
the running time of the wqsa depends on the choice of @xmath47 , while that of the grover algorithm does not .
this is because the state @xmath44 is a superpositin of states in the restricted domain of @xmath45 states .
the speedup is indeed originated in the initialization .
+ finally , let us discuss the haar wavelet basis .
although other wavelet transforms may be applied to the wqsa , we chose the haar wavelet transform .
it is observed that the first half of the haar wavelet basis differs with the second half of the wavelet basis on the phase @xmath76 .
this implies that destructive and constructive interference between states accepts a set of states containing the target and rejects the other states . in this sense , other known wavelet basis , e.g. daubechies s ,
are not appropriate to play the role of seclecting a subset of the @xmath1 states . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.